The advancement of text generation models has granted us the capability to produce coherent and convincing text on demand. Yet, in real-life circumstances, individuals do not continuously generate text or voice their opinions. For instance, consumers pen product reviews after weighing the merits and demerits of a product, and professional analysts issue reports following significant news releases. In essence, opinion expression is typically prompted by particular reasons or signals. Despite long-standing developments in opinion mining, the appropriate timing for expressing an opinion remains largely unexplored. To address this deficit, our study introduces an innovative task - the identification of news-triggered opinion expressing timing. We ground this task in the actions of professional stock analysts and develop a novel dataset for investigation. Our Generator-Assisted Decision-Focused Approach (GADFA) is decision-focused, leveraging text generation models to steer the classification model, thus enhancing overall performance. Our experimental findings demonstrate that the text generated by our model contributes fresh insights from various angles, effectively aiding in identifying the optimal timing for opinion expression.

We introduce MESAQA, a novel dataset focusing on multi-span contextual understanding question answering (QA).Unlike traditional single-span QA systems, questions in our dataset consider information from multiple spans within the context document. MESAQA supports evidence-grounded QA, demanding the model’s capability of answer generation and multi-evidence identification. Our automated dataset creation method leverages the MASH-QA dataset and large language models (LLMs) to ensure that each Q/A pair requires considering all selected spans. Experimental results show that current models struggle with multi-span contextual QA, underscoring the need for new approaches. Our dataset sets a benchmark for this emerging QA paradigm, promoting research in complex information retrieval and synthesis.

Sycophancy causes models to produce answers that cater to user expectations rather than providing truthful responses.Sycophantic behavior in models can erode user trust by creating a perception of dishonesty or bias.This lack of authenticity may lead users to question the reliability and objectivity of the system's responses.Although Reinforcement Learning from Human Feedback (RLHF) is effective in aligning models with human preferences, previous studies have observed that it can simultaneously amplify sycophantic behavior.However, these studies primarily focused on proprietary models and employed indirect analysis to demonstrate the influence of human feedback.Our study focuses on sycophancy in open-source models, which are more reproducible and transparent for research.We investigated the impact of human feedback on sycophancy by directly comparing models aligned with human feedback to those not aligned.To address sycophancy, we proposed assessing the user's expected answer rather than ignoring it.Consequently, we developed the Sycophancy Answer Assessment (SAA) dataset 1 and introduced Self-Augmented Preference Alignment, demonstrating that these methods effectively enhance the model's assessment ability and significantly reduce sycophancy across tasks.

Timely understanding of a patient’s clinical course is crucial for effective treatment. Extracting course-related information, such as temporal and medical events, from unstructured medical records is both challenging and time-consuming, especially when relying on manual identification by physicians. We introduce TimelyMed, a system powered by a locally deployed large language model (LLM) that ensures data security while efficiently organizing key psychiatric events and their corresponding temporal information. Additionally, our system is attributed, allowing clinicians to not only categorize events but also trace them back to their original textual descriptions, ensuring transparency and interpretability in clinical decision-making. By organizing temporal and medical event information into timelines, our system enables physicians to quickly grasp a patient’s medical history while effectively reducing their cognitive burden.

Legal question answering requires accurate retrieval of relevant laws, yet the significant writing style gap between user queries and legal provisions poses a major challenge. Existing datasets and retrieval methods often struggle to capture the complexity of legal language, limiting retrieval effectiveness. In this study, we introduce the Legal Query-to-Provision Retrieval (LQPR) task and construct Query2Provision (Q2P), a dataset designed to enhance law retrieval by incorporating diverse case scenarios and linguistic structures representative of real-world legal inquiries. To address the style disparity, we propose a style translation approach that transforms informal user queries into a more formal legal tone and simplifies complex legal provisions for better alignment. Our experiments demonstrate that integrating writing style transformation significantly improves retrieval performance. The dataset is available at https://github.com/ntunlplab/Query2Provision

People often struggle to remember specific details of past experiences, which can lead to the need to revisit these memories. Consequently, lifelog retrieval has emerged as a crucial application. Various studies have explored methods to facilitate rapid access to personal lifelogs for memory recall assistance. In this paper, we propose a Captioning-Integrated Visual Lifelog (CIVIL) Retrieval System for extracting specific images from a user’s visual lifelog based on textual queries. Unlike traditional embedding-based methods, our system first generates captions for visual lifelogs and then utilizes a text embedding model to project both the captions and user queries into a shared vector space. Visual lifelogs, captured through wearable cameras, provide a first-person viewpoint, necessitating the interpretation of the activities of the individual behind the camera rather than merely describing the scene. To address this, we introduce three distinct approaches: the single caption method, the collective caption method, and the merged caption method, each designed to interpret the life experiences of lifeloggers. Experimental results show that our method effectively describes first-person visual images, enhancing the outcomes of lifelog retrieval. Furthermore, we construct a textual dataset that converts visual lifelogs into captions, thereby reconstructing personal life experiences.

Headline generation, a key task in abstractive summarization, strives to condense a full-length article into a succinct, single line of text. Notably, while contemporary encoder-decoder models excel based on the ROUGE metric, they often falter when it comes to the precise generation of numerals in headlines. We identify the lack of datasets providing fine-grained annotations for accurate numeral generation as a major roadblock. To address this, we introduce a new dataset, the NumHG, and provide over 27,000 annotated numeral-rich news articles for detailed investigation. Further, we evaluate five well-performing models from previous headline-generation tasks using human evaluation in terms of numerical accuracy, reasonableness, and readability. Our study reveals a need for improvement in numerical accuracy, demonstrating the potential of the NumHG dataset to drive progress in number-focused headline generation and stimulate further discussions in numeral-focused text generation.

Argument mining has typically been researched for specific corpora belonging to concrete languages and domains independently in each research work. Human argumentation, however, has domain- and language-dependent linguistic features that determine the content and structure of arguments. Also, when deploying argument mining systems in the wild, we might not be able to control some of these features. Therefore, an important aspect that has not been thoroughly investigated in the argument mining literature is the robustness of such systems to variations in language and domain. In this paper, we present a complete analysis across three different languages and three different domains that allow us to have a better understanding on how to leverage the scarce available corpora to design argument mining systems that are more robust to natural language variations.

Large Language Models (LLMs) have demonstrated capability in “instruction induction,” generating instructions from demonstrations (input-output pairs). However, existing methods often rely on large datasets or numerous examples, which is impractical and costly in real-world scenarios. In this work, we propose a low-cost, task-level framework called Induct-Learn. It induces pseudo instructions from a few demonstrations and a short phrase, adding a CoT process into existing demonstrations. When encountering new problems, the learned pseudo instructions and demonstrations with the pseudo CoT process can be combined into a prompt to guide the LLM’s problem-solving process. We validate our approach on the BBH-Induct and Evals-Induct datasets, and the results show that the Induct-Learn framework outperforms state-of-the-art methods. We also exhibit cross-model adaptability and achieve superior performance at a lower cost compared to existing methods.

The popularization of networks fosters the convenience of communication. People can easily share their life experiences and thoughts with relatives and friends via instant messaging software. As time passes, individuals may forget certain details of life events, leading to difficulties in effectively communicating with others. The propensity of individuals to forget or mix up life events highlights the importance of services aimed at retrieving information about past experiences. This paper presents a conversational information recall system, ConvLogRecaller, which proactively supports real-time memory recall assistance during online conversations. Given a conversation of the user with others, ConvLogRecaller suggests a message if the user forgets the details of the life experiences. The services provided by our system can avoid hesitations or memory lapses that might hinder the efficiency of a conversation.

Contributions are essentially the core of every scientific research, highlighting their key values to the academic community. Systems that are capable of identifying the contributions from scientific papers precisely and organizing them into well-structured summaries can facilitate both text processing and human comprehension. In this paper, we present ContributionSum, a dataset consisting of 24K computer science papers with contributions explicitly listed by the authors, which are further classified into different contribution types based on a newly-proposed annotation scheme. In addition, we study the task of generating disentangled contributions that summarize the values of scientific papers into key points. We propose a fine-grained post-training strategy tailored to our task and leverage salient information of different contribution types in the papers. To assess the coherency and coverage of each contribution aspect, we perform summary-level and contribution-level evaluations for our task. Experimental results show that our method improves upon mainstream baselines.

As the significant growth of scientific publications in recent years, an efficient way to extract scholarly knowledge and organize the relationship among literature is necessitated. Previous works constructed scientific knowledge graph with authors, papers, citations, and scientific entities. To assist researchers to grasp the research context comprehensively, this paper constructs a fine-grained citation graph in which citation intents and their supporting evidence are labeled between citing and cited papers instead. We propose a model with a Transformer encoder to encode the long-lengthy paper. To capture the coreference relations of words and sentences in a paper, a coreference graph is created by utilizing Gated Graph Convolution Network (GGCN). We further propose a graph modification mechanism to dynamically update the coreference links. Experimental results show that our model achieves promising results on identifying multiple citation intents in sentences.

This paper introduces the DynamicESG dataset, a unique resource for dynamically extracting ESG ratings from news articles. The ESG rating, a novel metric employed annually to gauge a company's sustainability, relies heavily on corporate disclosure and other external information, especially news narratives. Our dataset, comprising a wide spectrum of news over a twelve-year span, annotates articles in accordance with MSCI ESG ratings methodology and SASB standards, with relevance to ESG issues. DynamicESG provides a comprehensive means of investigating the relationship between public discourse, ESG-related events, and subsequent ESG rating adjustments. We detail our data collection, curation, annotation procedure, and inter-rater agreement, ensuring high data quality and usability. Importantly, our dataset includes a temporal dimension, enabling the analysis of longitudinal trends in ESG ratings and their correlation with news coverage. Moreover, the dataset incorporates an opportunity/risk tendency, thus permitting analysis from diverse perspectives to discern if the news is beneficial or detrimental to the company. We believe this dataset will serve as a valuable resource for researchers in fields such as corporate social responsibility, sustainable investing, machine learning, and natural language processing. Initial analysis using the dataset underscores its potential to facilitate new insights into the dynamics of ESG ratings and the influence of news media on these ratings.

In this paper, we address the hallucination problem commonly found in natural language generation tasks.Language models often generate fluent and convincing content but lack consistency with the provided source, resulting in potential inaccuracies.We propose a new decoding method called Fidelity-Enriched Contrastive Search (FECS), which augments the Contrastive Search framework with contextaware regularization terms.FECS promotes tokens that are semantically similar to the provided source while penalizing repetitiveness in the generated text.We demonstrate its effectiveness across two tasks prone to hallucination: abstractive summarization and dialogue generation.Results show that FECS consistently enhances faithfulness across various language model sizes while maintaining output diversity comparable to well-performing decoding algorithms. 1

Large language models (LLMs) have exhibited striking in-context learning (ICL) ability to adapt to target tasks with a few inputoutput demonstrations.For better ICL, different methods are proposed to select representative demonstrations from existing training corpora.However, such settings are not aligned with real-world practices, as end-users usually query LMs without access to demonstration pools.In this work, we introduce SELF-ICL-a simple framework which bootstraps LMs' intrinsic capabilities to perform zero-shot ICL.Given a test input, SELF-ICL first prompts the model to generate pseudoinputs.Next, the model predicts pseudo-labels for the pseudo-inputs via zero-shot prompting.Finally, we perform ICL for the test input with the pseudo-input-label pairs as demonstrations.Evaluation on 23 BIG-Bench Hard tasks shows SELF-ICL outperforms zero-shot baselines on both average accuracy and head-to-head comparison.Moreover, with zero-shot chain-ofthought, SELF-ICL achieves results comparable to using real demonstrations.Additionally, we conduct a range of analyses to validate SELF-ICL's effectiveness and provide insights for its behaviors under different settings.1 * Equal contribution. 1 https://github.com/ntunlplab/Self-ICLFollowing is an example instance for the task: Evaluate the result of a random Boolean expression.Please come up with 3 new, diverse, and creative instances for the task.Example instance: Q: not ( True ) and ( True ) is New instance 1: Q: ( False ) or ( False ) and ( True ) is New instance 2: Q: ( True ) and ( False ) or ( False ) is New instance 3: Q: not ( False ) and ( True ) or ( False ) Task description: Evaluate the result of a random Boolean expression.Q: ( False ) or ( False ) and ( True ) is A: False Q: ( True ) and ( False ) or ( False ) is A: False Q: not ( False ) and ( True ) or ( False ) A: True Q: not ( True ) and ( True ) is

In various scenarios, such as conference oral presentations, company managers' talks, and politicians' speeches, individuals often contemplate the potential questions that may arise from their presentations.This common practice prompts the research question addressed in this study: to what extent can models generate multiple questions based on a given presentation transcript?To investigate this, we conduct pilot explorations using earnings conference call transcripts, which serve as regular meetings between professional investors and company managers.We experiment with different task settings and methods and evaluate the results from various perspectives.Our findings highlight that incorporating key points retrieval techniques enhances the accuracy and diversity of the generated questions.

Abstract With the aid of recently proposed word embedding algorithms, the study of semantic relatedness has progressed rapidly. However, word-level representations are still lacking for many natural language processing tasks. Various sense-level embedding learning algorithms have been proposed to address this issue. In this paper, we present a generalized model derived from existing sense retrofitting models. In this generalization, we take into account semantic relations between the senses, relation strength, and semantic strength. Experimental results show that the generalized model outperforms previous approaches on four tasks: semantic relatedness, contextual word similarity, semantic difference, and synonym selection. Based on the generalized sense retrofitting model, we also propose a standardization process on the dimensions with four settings, a neighbor expansion process from the nearest neighbors, and combinations of these two approaches. Finally, we propose a Procrustes analysis approach that inspired from bilingual mapping models for learning representations that outside of the ontology. The experimental results show the advantages of these approaches on semantic relatedness tasks.

Peer review is an essential part of the scientific process in which the research papers are assessed by several reviewers. The author rebuttal phase, which is held at most top conferences, provides an opportunity for the authors to defend their work against the arguments made by the reviewers. The strengths and the weaknesses pointed out by the reviewers, as well as the authors' responses, will be evaluated by the area chair. The final decisions generally accompany meta-reviews regarding the reason for acceptance/rejection. Previous research has studied the generation of meta-review using transformer-based summarization models. However, few of them consider the rebuttals' content and the interaction between reviews and rebuttals' arguments, where the argumentation persuasiveness plays an important role in affecting the final decision. To generate a comprehensive meta-review that well organizes reviewers' opinions and authors' responses, we present a novel generation model that is capable of explicitly modeling the complicated argumentation structure from not only arguments between the reviewers and the authors but also the inter-reviewer discussions. Experimental results show that our model outperforms baselines in terms of both automatic evaluation and human evaluation, demonstrating the effectiveness of our approach.

Handling unanswerable questions in knowledge base question answering (KBQA) has been a focus in recent years. However, how to explain why a given question is unanswerable is rarely discussed. In this work, we seek not only to correct unanswerable questions based on a personal knowledge base, but also to explain the reason of the correction. We argue that different types of questions need heterogeneous subgraphs with different types of connections. We thus propose a heterogeneous subgraph aggregation network with a two-level attention mechanism to detect important entities and relations in subgraphs and attend to informative subgraphs for different questions. We conduct comprehensive experiments on five subgraphs and their combinations, with results that attest the effectiveness of incorporating heterogeneous subgraphs.

Explaining the reasoning of neural models has attracted attention in recent years. Providing highly-accessible and comprehensible explanations in natural language is useful for humans to understand model’s prediction results. In this work, we present a pilot study to investigate explanation generation with a narrative and causal structure for the scenario of health consulting. Our model generates a medical suggestion regarding the patient’s concern and provides an explanation as the outline of the reasoning. To align the generated explanation with the suggestion, we propose a novel discourse-aware mechanism with multi-task learning. Experimental results show that our model achieves promising performances in both quantitative and human evaluation.

When recalling life experiences, people often forget or confuse life events, which necessitates information recall services.Previous work on information recall focuses on providing such assistance reactively, i.e., by retrieving the life event of a given query.Proactively detecting the need for information recall services is rarely discussed.In this paper, we use a humanannotated life experience retelling dataset to detect the right time to trigger the information recall service.We propose a pilot modelstructured event enhancement network (SEEN) that detects life event inconsistency, additional information in life events, and forgotten events.A fusing mechanism is also proposed to incorporate event graphs of stories and enhance the textual representations.To explain the need detection results, SEEN simultaneously provides support evidence by selecting the related nodes from the event graph.Experimental results show that SEEN achieves promising performance in detecting information needs.In addition, the extracted evidence can be served as complementary information to remind users what events they may want to recall.➊ My brother decided to propose to his fiancée, Ellie.Afterward, ➋ She hosted the party in their backyard.It was so fun!They had a beautifully stocked bar with lots of alcohol.➌ The shots and cocktails were fantastic!PRE-RETOLD STORY POST-RETOLD STORY Match Match Conflict Forgotten My brother's engagement party was hosted in my hometown.➍ It was hosted by his fiancée, Ellen.➎ She had it outside in her backyard.Even though it was 90 degrees, we all had so much fun !➏ In addition, it was also the first time that

In a debate, two debaters with opposite stances put forward arguments to fight for their viewpoints. Debaters organize their arguments to support their proposition and attack opponents’ points. The common purpose of debating is to persuade the opponents and the audiences to agree with the mentioned propositions. Previous works have investigated the issue of identifying which debater is more persuasive. However, modeling the interaction of arguments between rounds is rarely discussed. In this paper, we focus on assessing the overall performance of debaters in a multi-round debate on online forums. To predict the winner in a multi-round debate, we propose a novel neural model that is aimed at capturing the interaction of arguments by exploiting raw text, structure information, argumentative discourse units (ADUs), and the relations among ADUs. Experimental results show that our model achieves competitive performance compared with the existing models, and is capable of extracting essential argument relations during a multi-round debate by leveraging argumentative structure and attention mechanism.

Argument structure elaborates the relation among claims and premises. Previous works in persuasiveness prediction do not consider this relation in their architectures. To take argument structure information into account, this paper proposes an approach to persuasiveness prediction with a novel graph-based neural network model, called heterogeneous argument attention network (HARGAN). By jointly training on the persuasiveness and stance of the replies, our model achieves the state-of-the-art performance on the ChangeMyView (CMV) dataset for the persuasiveness prediction task. Experimental results show that the graph setting enables our model to aggregate information across multiple paragraphs effectively. In the meanwhile, our stance prediction auxiliary task enables our model to identify the viewpoint of each party, and helps our model perform better on the persuasiveness prediction.

People often encounter situations where they need to recall past experiences from their daily life. In this paper, we aim to construct a question answering system that enables human to query their past experiences over personal knowledge base. Previous works on knowledge base question answering focus on finding answers for answerable questions. In the real world applications, however, people often muddle up facts and ask those questions that cannot be answered with knowledge base. This work presents a novel system consisting of question answering model and question generation model. It not only answers answerable questions, but also corrects unanswerable questions if necessary. Our question answering model recognizes the question that is inconsistent with the state of the personal knowledge base and suggests facts that can form a feasible question. Then, the facts are converted to an answerable question by the question generation model. For refining question, we propose a question generation model based on the reinforcement learning (RL) with question editing mechanism. Experimental results show that our proposed system is effective for correcting unanswerable questions in personal knowledge base question answering.

Freedom of speech is one of the principles in the constitution of most countries. However, in the 2020 United States presidential election, Donald Trump's Twitter account is suspended due to the risk of further incitement of violence. That leads to the question: Which kind of rumors may undermine society? In this paper, we discuss this question based on the case studies of real-world court orders, which are the judges' official proclamations. We point out the possible research directions that NLP researchers may need to consider before applying our systems to society.

Numeral information plays an important role in narratives of several domains such as medicine, engineering, and finance. Previous works focus on the foundation exploration toward numeracy and show that fine-grained numeracy is a challenging task. In machine reading comprehension, our statistics show that only a few numeral-related questions appear in previous datasets. It indicates that few benchmark datasets are designed for numeracy learning. In this paper, we present a Numeral-related Question Answering Dataset, NQuAD, for fine-grained numeracy, and propose several baselines for future works. We compare NQuAD with three machine reading comprehension datasets and show that NQuAD is more challenging than the numeral-related questions in other datasets. NQuAD is published under the CC BY-NC-SA 4.0 license for academic purposes.

The economic policy uncertainty (EPU) index is one of the important text-based indexes in finance and economics fields. The EPU indexes of more than 26 countries have been constructed to reflect the policy uncertainty on country-level economic environments and serve as an important economic leading indicator. The EPU indexes are calculated based on the number of news articles with some manually-selected keywords related to economic, uncertainty, and policy. We find that the keyword-based EPU indexes contain noise, which will influence their explainability and predictability. In our experimental dataset, over 40% of news articles with the selected keywords are not related to the EPU. Instead of using keywords only, our proposed models take contextual information into account and get good performance on identifying the articles unrelated to EPU. The noise free EPU index performs better than the keyword-based EPU index in both explainability and predictability.

The volatility of stock price reflects the risk of stock and influences the risk of investor's portfolio. It is also a crucial part of pricing derivative securities. Researchers have paid their attention to predict the stock volatility with different kinds of textual data. However, most of them focus on using word information only. Few touch on capturing the numeral information in textual data, providing fine-grained clues for financial document understanding. In this paper, we present a novel dataset, ECNum, for understanding the numerals in the transcript of earnings conference calls. We propose a simple but efficient method, Numeral-Aware Model (NAM), for enhancing the capacity of numeral understanding of neural network models. We employ the distilled information in the stock volatility forecasting task and achieve the best performance compared to the previous works in short-term scenarios.

Filming video blogs, which is shortened to vlog, becomes a popular way for people to record their life experiences in recent years. In this work, we present a novel task that is aimed at extracting life events from videos and constructing personal knowledge bases of individuals. In contrast to most existing researches in the field of computer vision that focus on identifying low-level script-like activities such as moving boxes, our goal is to extract life events where high-level activities like moving into a new house are recorded. The challenges to be tackled include: (1) identifying which objects in a given scene related to the life events of the protagonist we concern, and (2) determining the association between an extracted visual concept and a more high-level description of a video clip. To address the research issues, we construct a video life event extraction dataset VidLife by exploiting videos from the TV series The Big Bang Theory, in which the plot is around the daily lives of several characters. A pilot multitask learning model is proposed to extract life events given video clips and subtitles for storing in the personal knowledge base.

Textual information extraction is a typical research topic in the NLP community.Several NLP tasks such as named entity recognition and relation extraction between entities have been well-studied in previous work.However, few works pay their attention to the implicit information.For example, a financial news article mentioned "Apple Inc." may be also related to Samsung, even though Samsung is not explicitly mentioned in this article.This work presents a novel dynamic graph transformer that distills the textual information and the entity relations on the fly.Experimental results confirm the effectiveness of our approach to implicit tag recognition.

Both the issues of data deficiencies and semantic consistency are important for data augmentation.Most of previous methods address the first issue, but ignore the second one.In the cases of aspect-based sentiment analysis, violation of the above issues may change the aspect and sentiment polarity.In this paper, we propose a semantics-preservation data augmentation approach by considering the importance of each word in a textual sequence according to the related aspects and sentiments.We then substitute the unimportant tokens with two replacement strategies without altering the aspect-level polarity.Our approach is evaluated on several publicly available sentiment analysis datasets and the real-world stock price/risk movement prediction scenarios.Experimental results show that our methodology achieves better performances in all datasets.

This paper presents a conversational lifelog mining system, ConvLogMiner, which detects personal life events from the human online conversation in real-time. Given a daily conversation of two speakers, ConvLogMiner identifies the new life events specific to each speaker that occur in the latest utterances. The lifelogs mined by our system are useful to provide complementary information to support lifestyle analysis and memory assistance service.

With the passage of time, people often have misty memories of their past experiences. Information recall support for people by collecting personal lifelogs is emerging. Recently, people tend to record their daily life via filming Video Weblog (VLog), which contains visual and audio data. These large scale multimodal data can be used to support information recall service that enables users to query their past experiences. The challenging issue is the semantic gap between the visual concept and the textual query. In this paper, we aim to extract personal life events from vlogs shared on YouTube and construct a personal knowledge base (PKB) for individuals. A multitask learning model is proposed to extract the components of personal life events, such as subjects, predicates and objects. The evaluation is performed on a video collection from three YouTubers who are English native speakers. Experimental results show our model achieves promising performance.

With the advance of science and technology, people are used to recording their daily life events via writing blogs, uploading social media posts, taking photos, or filming videos. Such rich repository personal information is useful for supporting human living assistance, such as information recall service. The main challenges are how to store and manage personal knowledge from various sources, and how to provide support for people who may have difficulty recalling past experiences. In this position paper, we propose a research agenda on personal knowledge mining from various sources of lifelogs, personal knowledge base construction, and information recall for assisting people to recall their experiences. Ten research questions are formulated.

Previous studies on the financial news focus mainly on the news articles explicitly mentioning the target financial instruments, and may suffer from data sparsity. As taking into consideration other related news, e.g., sector-related news, is a crucial part of real-world decision-making, we explore the use of news without explicit target mentions to enrich the information for the prediction model. We develop a neural network framework that jointly learns with a news selection mechanism to extract implicit information from the chaotic daily news pool. Our proposed model, called the news distilling network (NDN), takes advantage of neural representation learning and collaborative filtering to capture the relationship between stocks and news. With NDN, we learn latent stock and news representations to facilitate similarity measurements, and apply a gating mechanism to prevent noisy news representations from flowing to a higher level encoding stage, which encodes the selected news representation of each day. Extensive experiments on real-world stock market data demonstrate the effectiveness of our framework and show improvements over previous techniques.

This paper demonstrates FinSense, a system that improves the working efficiency of financial information processing. Given the draft of a financial news story, FinSense extracts the explicit-mentioned stocks and further infers the implicit stocks, providing insightful information for decision making. We propose a novel graph convolutional network model that performs implicit financial instrument inference toward the in-domain data. In addition, FinSense generates candidate headlines for the draft, reducing a significant amount of time in journalism production. The proposed system also provides assistance to investors to sort out the information in the financial news articles.

Social media’s rise in popularity has demonstrated the usefulness of the wisdom of the crowd. Most previous works take into account the law of large numbers and simply average the results extracted from tasks such as opinion mining and sentiment analysis. Few attempt to identify high-quality opinions from the mined results. In this paper, we propose an approach for capturing expert-like rationales from social media platforms without the requirement of the annotated data. By leveraging stylistic and semantic features, our approach achieves an F1-score of 90.81%. The comparison between the rationales of experts and those of the crowd is done from stylistic and semantic perspectives, revealing that stylistic and semantic information provides complementary cues for professional rationales. We further show the advantage of using these superlative analysis results in the financial market, and find that top-ranked opinions identified by our approach increase potential returns by up to 90.31% and reduce downside risk by up to 71.69%, compared with opinions ranked by feedback from social media users. Moreover, the performance of our method on downside risk control is comparable with that of professional analysts.

This work proposes a standalone, complete Chinese discourse parser for practical applications.We approach Chinese discourse parsing from a variety of aspects and improve the shift-reduce parser not only by integrating the pre-trained text encoder, but also by employing novel training strategies.We revise the dynamic-oracle procedure for training the shift-reduce parser, and apply unsupervised data augmentation to enhance rhetorical relation recognition.Experimental results show that our Chinese discourse parser achieves the state-of-the-art performance.

The goal of claim detection in argument mining is to sort out the key points from a long narrative. In this paper, we design a novel task for argument mining in the financial domain, and provide an expert-annotated dataset, NumClaim, for the proposed task. Based on the statistics, we discuss the differences between the claims in other datasets and the claims of the investors in NumClaim. With the ablation analysis, we show that encoding numeral and co-training with the auxiliary task of the numeral understanding, i.e., the category classification task, can improve the performance of the proposed task under different neural network architectures. The annotations in the NumClaim is published for academic usage under the CC BY-NC-SA 4.0 license.

Most recent works in the field of grammatical error correction (GEC) rely on neural machine translation-based models. Although these models boast impressive performance, they require a massive amount of data to properly train. Furthermore, NMT-based systems treat GEC purely as a translation task and overlook the editing aspect of it. In this work we propose a heterogeneous approach to Chinese GEC, composed of a NMT-based model, a sequence editing model, and a spell checker. Our methodology not only achieves a new state-of-the-art performance for Chinese GEC, but also does so without relying on data augmentation or GEC-specific architecture changes. We further experiment with all possible configurations of our system with respect to model composition order and number of rounds of correction. A detailed analysis of each model and their contributions to the correction process is performed by adapting the ERRANT scorer to be able to score Chinese sentences.

A dialogue dataset is an indispensable resource for building a dialogue system. Additional information like emotions and interpersonal relationships labeled on conversations enables the system to capture the emotion flow of the participants in the dialogue. However, there is no publicly available Chinese dialogue dataset with emotion and relation labels. In this paper, we collect the conversions from TV series scripts, and annotate emotion and interpersonal relationship labels on each utterance. This dataset contains 25,548 utterances from 4,142 dialogues. We also set up some experiments to observe the effects of the responded utterance on the current utterance, and the correlation between emotion and relation types in emotion and relation classification tasks.

In this paper, we investigate the annotation of financial social media data from several angles. We present Fin-SoMe, a dataset with 10,000 labeled financial tweets annotated by experts from both the front desk and the middle desk in a bank’s treasury. These annotated results reveal that (1) writer-labeled market sentiment may be a misleading label; (2) writer’s sentiment and market sentiment of an investor may be different; (3) most financial tweets provide unfounded analysis results; and (4) almost no investors write down the gain/loss results for their positions, which would otherwise greatly facilitate detailed evaluation of their performance. Based on these results, we address various open problems and suggest possible directions for future work on financial social media data. We also provide an experiment on the key snippet extraction task to compare the performance of using a general sentiment dictionary and using the domain-specific dictionary. The results echo our findings from the experts’ annotations.

Chinese discourse parsing, which aims to identify the hierarchical relationships of Chinese elementary discourse units, has not yet a consistent evaluation metric. Although Parseval is commonly used, variations of evaluation differ from three aspects: micro vs. macro F1 scores, binary vs. multiway ground truth, and left-heavy vs. right-heavy binarization. In this paper, we first propose a neural network model that unifies a pre-trained transformer and CKY-like algorithm, and then compare it with the previous models with different evaluation scenarios. The experimental results show that our model outperforms the previous systems. We conclude that (1) the pre-trained context embedding provides effective solutions to deal with implicit semantics in Chinese texts, and (2) using multiway ground truth is helpful since different binarization approaches lead to significant differences in performance.

Sense embedding models handle polysemy by giving each distinct meaning of a word form a separate representation. They are considered improvements over word models, and their effectiveness is usually judged with benchmarks such as semantic similarity datasets. However, most of these datasets are not designed for evaluating sense embeddings. In this research, we show that there are at least six concerns about evaluating sense embeddings with existing benchmark datasets, including the large proportions of single-sense words and the unexpected inferior performance of several multi-sense models to their single-sense counterparts. These observations call into serious question whether evaluations based on these datasets can reflect the sense model’s ability to capture different meanings. To address the issues, we propose the Multi-Sense Dataset (MSD-1030), which contains a high ratio of multi-sense word pairs. A series of analyses and experiments show that MSD-1030 serves as a more reliable benchmark for sense embeddings. The dataset is available at http://nlg.csie.ntu.edu.tw/nlpresource/MSD-1030/.

The advance in wearable technology has made lifelogging more feasible and more popular. Visual lifelogs collected by wearable cameras capture every single detail of individual's life experience, offering a promising data source for deeper lifestyle analysis and better memory recall assistance. However, building a system for organizing and accessing visual lifelogs is a challenging task due to the semantic gap between visual data and semantic descriptions of life events. In this paper, we introduce semantic knowledge to reduce such a semantic gap for daily activity recognition and lifestyle understanding. We incorporate the semantic knowledge derived from external resources to enrich the training data for the proposed supervised learning model. Experimental results show that incorporating external semantic knowledge is beneficial for improving the performance of recognizing life events.

This paper introduces visual story ordering, a challenging task in which images and text are ordered in a visual story jointly. We propose a neural network model based on the reader-processor-writer architecture with a self-attention mechanism. A novel bidirectional decoder is further proposed with bidirectional beam search. Experimental results show the effectiveness of the approach. The information gained from multimodal learning is presented and discussed. We also find that the proposed embedding narrows the distance between images and their corresponding story sentences, even though we do not align the two modalities explicitly. As it addresses a general issue in generative models, the proposed bidirectional inference mechanism applies to a variety of applications.

In this research, we propose 3 different approaches to measure the semantic relatedness between 2 words: (i) boost the performance of GloVe word embedding model via removing or transforming abnormal dimensions; (ii) linearly combine the information extracted from WordNet and word embeddings; and (iii) utilize word embedding and 12 linguistic information extracted from WordNet as features for Support Vector Regression. We conducted our experiments on 8 benchmark data sets, and computed Spearman correlations between the outputs of our methods and the ground truth. We report our results together with 3 state‐of‐the‐art approaches. The experimental results show that our method can outperform state‐of‐the‐art approaches in all the selected English benchmark data sets.

In this paper, we attempt to answer the question of whether neural network models can learn numeracy, which is the ability to predict the magnitude of a numeral at some specific position in a text description.A large benchmark dataset, called Numeracy-600K, is provided for the novel task.We explore several neural network models including CNN, GRU, BiGRU, CRNN, CNN-capsule, GRU-capsule, and BiGRU-capsule in the experiments.The results show that the BiGRU model gets the best micro-averaged F1 score of 80.16%, and the GRU-capsule model gets the best macroaveraged F1 score of 64.71%.Besides discussing the challenges through comprehensive experiments, we also present an important application scenario, i.e., detecting exaggerated information, for the task.

Social trading platforms provide a forum for investors to share their analysis and opinions. Posts on these platforms are characterized by narrative styles which are much different from posts on general social platforms, for instance tweets. As a result, recommendation systems for social trading platforms should leverage tailor-made latent features. This paper presents a representation for these latent features in both textual data and market information. A real-world dataset is adopted to conduct experiments involving a novel task called next cashtag prediction. We propose a joint learning model with an attentive capsule network. Experimental results show positive results with the proposed methods and the corresponding auxiliary tasks.

This paper demonstrates an opinion indicator (OI) generation system, named Crowd View, with which traders can refer to the fine-grained opinions, beyond the market sentiment (bullish/bearish), from crowd investors when trading financial instruments. We collect the real-time textual information from Twitter, and convert it into five kinds of OIs, including the support price level, resistance price level, price target, buy-side cost, and sell-side cost. The OIs for all component stocks in Dow Jones Industrial Average Index (DJI) are provided, and shown with the real-time stock price for comparison and analysis. The information embedding in the OIs and the application scenarios are introduced.

In this paper we propose the task of numeral attachment to detect the attached target of a numeral. Compared with other kinds of named entities, numerals provide richer and more crucial information in some domains. Fine-grained understanding of the information embedded in numerals is a fundamental challenge. We develop NumAttach, a pilot dataset for the proposed task based on tweets. Two main challenges of this task include the informal writing style in tweets and the representation of numerals. To address these challenges, we present an embedding technique that considers word and numeral information simultaneously. Furthermore, we design a joint learning model with the capsule network to accomplish the proposed task. We also release NumAttach to the research community as a resource.

Previous work on lifelogging focuses on life event extraction from image, audio, and video data via wearable sensors. In contrast to wearing an extra camera to record daily life, people are used to log their life on social media platforms. In this paper, we aim to extract life events from textual data shared on Twitter and construct personal knowledge bases of individuals. The issues to be tackled include (1) not all text descriptions are related to life events, (2) life events in a text description can be expressed explicitly or implicitly, (3) the predicates in the implicit events are often absent, and (4) the mapping from natural language predicates to knowledge base relations may be ambiguous. A joint learning approach is proposed to detect life events in tweets and extract event components including subjects, predicates, objects, and time expressions. Finally, the extracted information is transformed to knowledge base facts. The evaluation is performed on a collection of lifelogs from 18 Twitter users. Experimental results show our proposed system is effective in life event extraction, and the constructed personal knowledge bases are expected to be useful to memory recall applications.

People engage in lifelogging by taking photos with cameras and cellphones anytime anywhere and share the photos, intertwined with captions or descriptions, on social media platforms. The image-text intertwined data provides richer information for image recall. When images cannot keep the complete information, the textual information is a complement to describe the life experiences under the photos. This work proposes a multimodal retrieval model for image recall in image-text intertwined lifelogs. Our Attentive Image-Story model combines an Image model, which transfers visual information and textual information to a single representation space, and a Story model, which captures text-based contextual information, with an attention mechanism to reduce the semantic gap between visual and textual information. Experimental results show our model outperforms a state-of-the-art image-based retrieval model and the image/text hybrid system.

Understanding how information propagates among social media users can allow researchers to provide interesting insights into online social networks and lead to applications such as precise advertising and misinformation management. In this work, we focus on information diffusion through post sharing. Given an information propagation tree, our goal is to predict a list of potential users of the tree. A framework based on graph convolutional network (GCN) is proposed to learn the latent representation of a propagation tree and match it with the latent representation of a user. A novel strategy for tree pruning is further investigated to improve the GCN. Experimental results show that our framework outperforms the existing methods for modeling information diffusion.

This paper demonstrates a novel analytics service, CrowdPT, for capturing the key information, price target (PT), of individual investors on social media. PT, which is mentioned as a conclusion in most of analysts' reports, indicates not only the market sentiment (bullish/bearish) of investors, but also the analysis results. In order to provide the latest opinions of individual investors, we monitor Twitter in real time and update the information in price chart daily. For all component stocks in Dow Jones Industrial Average, textual information from numerous tweets is summarized into a single number, PT, in CrowdPT. Case studies confirm the effectiveness of our analytics service in the financial domain, and show that capturing the PT of individual investors is promising for stock price prediction. The Web API of CrowdPT is also provided for academic purpose.

With the aid of recently proposed word embedding algorithms, the study of semantic similarity has progressed and advanced rapidly. However, many natural language processing tasks need sense level representation. To address this issue, some researches propose sense embedding learning algorithms. In this paper, we present a generalized model from existing sense retrofitting model. The generalization takes three major components: semantic relations between the senses, the relation strength and the semantic strength. In the experiment, we show that the generalized model can outperform previous approaches in three types of experiment: semantic relatedness, contextual word similarity and semantic difference.

With more and more people around the world learning Chinese as a second language, the need of Chinese error correction tools is increasing. In the HSK dynamic composition corpus, word usage error (WUE) is the most common error type. In this paper, we build a neural network model that considers both target erroneous token and context to generate a correction vector and compare it against a candidate vocabulary to propose suitable corrections. To deal with potential alternative corrections, the top five proposed candidates are judged by native Chinese speakers. For more than 91% of the cases, our system can propose at least one acceptable correction within a list of five candidates. To the best of our knowledge, this is the first research addressing general-type Chinese WUE correction. Our system can help non-native Chinese learners revise their sentences by themselves.

This work demonstrates a writing assistant system that provides high level advice for Chinese scientific writing. Cross-lingual approaches are investigated to analyze the information structure of a given Chinese abstract and retrieve useful knowledge in the related work written in both English and Chinese. To the best of our knowledge, this is the first study on Chinese information structure identification. Without the need of labeled Chinese data, our novel model is capable of dealing with Chinese instances by acquiring language-invariant knowledge from the labeled English data. Adversarial learning is employed to enhance the cross-lingual sentence representation.

Numerals that contain much information in financial documents are crucial for financial decision making. They play different roles in financial analysis processes. This paper is aimed at understanding the meanings of numerals in financial tweets for fine-grained crowd-based forecasting. We propose a taxonomy that classifies the numerals in financial tweets into 7 categories, and further extend some of these categories into several subcategories. Neural network-based models with word and character-level encoders are proposed for 7-way classification and 17-way classification. We perform backtest to confirm the effectiveness of the numeric opinions made by the crowd. This work is the first attempt to understand numerals in financial social media data, and we provide the first comparison of fine-grained opinion of individual investors and analysts based on their forecast price. The numeral corpus used in our experiments, called FinNum 1.0, is available for research purposes.

Verbalization of knowledge base (KB) facts about an entity allows users to absorb information from KB more easily. The drawback of most previous work is that they cannot generalize to unseen frames. This work introduces the task of precise KB verbalization that is aimed at generating an exact description for the given factual triples. We propose a novel sequence-to-sequence (seq2seq) model with the local pointer network to deal with this task. The approach to training data construction is also explored. Experimental results show our method improves the performances in terms of Meteor and slot error rates. Human evaluation is also performed to confirm the effectiveness of our model.

With the aid of recently proposed word embedding algorithms, the study of semantic relatedness has progressed and advanced rapidly. In this research, we propose a novel structural-fitting method that utilizes the linguistic ontology into vector space representations. The ontological information is applied in two ways. The fine2coarse approach refines the word vectors from fine-grained to coarse-grained terms (word types), while the coarse2fine approach refines the word vectors from coarse-grained to fine-grained terms. In the experiments, we show that our proposed methods outperform previous approaches in seven publicly available benchmark datasets.

This paper demonstrates DISA, a higher-level writing assistant system, which analyzes the information structure of abstracts, and retrieves the knowledge according to the research goals from the related work. By incorporating the latest neural-network technologies including linguistically-informed neural-network and autoencoder, we construct an intelligent system which extends the scope of computer-aided writing.

This paper presents an approach to identify subject, type and property from knowledge base (KB) for answering simple questions. We propose new features to rank entity candidates in KB. Besides, we split a relation in KB into type and property. Each of them is modeled by a bi-directional LSTM. Experimental results show that our model achieves the state-of-the-art performance on the SimpleQuestions dataset. The hard questions in the experiments are also analyzed in detail.

There has been a surge in opinion-sharing in the public domain. Some opinions greatly influence our decisions, e.g., the choice of purchase. Malicious parties or individuals exploit social media by generating fake reviews for opinion manipulation. This paper aims to investigate the phenomenon of online paid restaurant reviews by bloggers. Our research provides an insight into some characteristics of paid reviews and their authors. We then explore a set of features based on our observations and detect paid reviews and paid bloggers using supervised machine learning techniques. Experimental results show the effectiveness of our approach.

This paper analyzes the lexical semantics of domain-specific terms based on various pre-trained specific domain and general domain word vectors, and addresses the semantic drift between domains. To capture lexical semantics in the specific domain, we propose a bridge mechanism to introduce domain-specific data into general data, and re-train word vectors. We find that even a small-scale fusion can result in the similar lexical semantics learned by using the large-scale domain-specific dataset. Experiments on sentiment analysis and outlier detection show that application of word embedding by the fusion dataset has the better performance than applications of word embeddings by pure large domain-specific and pure large general datasets. The simple, but effective methodology facilitates the domain adaptation of distributed word representations.

Opinion words are crucial information for sentiment analysis. In some text, however, opinion words are absent or highly ambiguous. The resulting implicit opinions are more difficult to extract and label than explicit ones. In this paper, cutting‐edge machine‐learning approaches – deep neural network and word‐embedding – are adopted for implicit opinion mining at the snippet and clause levels. Hotel reviews written in Chinese are collected and annotated as the experimental data set. Results show the convolutional neural network models not only outperform traditional support vector machine models, but also capture hidden knowledge within the raw text. The strength of word‐embedding is also analyzed.

Misuse of Chinese prepositions is one of common word usage errors in grammatical error diagnosis. In this paper, we adopt the Chinese Gigaword corpus and HSK corpus as L1 and L2 corpora, respectively. We explore gated recurrent neural network model (GRU), and an ensemble of GRU model and maximum entropy language model (GRU-ME) to select the best preposition from 43 candidates for each test sentence. The experimental results show the advantage of the GRU models over simple RNN and n-gram models. We further analyze the effectiveness of linguistic information such as word boundary and part-of-speech tag in this task.

This paper explores the role of tense information in Chinese causal analysis. Both tasks of causal type classification and causal directionality identification are experimented to show the significant improvement gained from tense features. To automatically extract the tense features, a Chinese tense predictor is proposed. Based on large amount of parallel data, our semi-supervised approach improves the dependency-based convolutional neural network (DCNN) models for Chinese tense labelling and thus the causal analysis.

In this paper, we investigate four important issues together for explicit discourse relation labelling in Chinese texts: (1) discourse connective extraction, (2) linking ambiguity resolution, (3) relation type disambiguation, and (4) argument boundary identification. In a pipelined Chinese discourse parser, we identify potential connective candidates by string matching, eliminate non-discourse usages from them with a binary classifier, resolve linking ambiguities among connective components by ranking, disambiguate relation types by a multiway classifier, and determine the argument boundaries by conditional random fields. The experiments on Chinese Discourse Treebank show that the F1 scores of 0.7506, 0.7693, 0.7458, and 0.3134 are achieved for discourse usage disambiguation, linking disambiguation, relation type disambiguation, and argument boundary identification, respectively, in a pipelined Chinese discourse parser.

Word embedding methods revolve around learning continuous distributed vector representations of words with neural networks, which can capture semantic and/or syntactic cues, and in turn be used to induce similarity measures among words, sentences and documents in context. Celebrated methods can be categorized as prediction-based and count-based methods according to the training objectives and model architectures. Their pros and cons have been extensively analyzed and evaluated in recent studies, but there is relatively less work continuing the line of research to develop an enhanced learning method that brings together the advantages of the two model families. In addition, the interpretation of the learned word representations still remains somewhat opaque. Motivated by the observations and considering the pressing need, this paper presents a novel method for learning the word representations, which not only inherits the advantages of classic word embedding methods but also offers a clearer and more rigorous interpretation of the learned word representations. Built upon the proposed word embedding method, we further formulate a translation-based language modeling framework for the extractive speech summarization task. A series of empirical evaluations demonstrate the effectiveness of the proposed word representation learning and language modeling techniques in extractive speech summarization.

Since more and more multimedia data associated with spoken documents have been made available to the public, spoken document retrieval (SDR) has become an important research subject in the past two decades. Following the research tendency, many efforts have been devoted towards developing indexing and modeling techniques for representing spoken documents, but only few have been made on improving query formulation for better representing users' information needs. The i-vector based language modeling (IVLM) framework, stemming from the state-of-the-art i-vector framework for language identification and speaker recognition, has been proposed and formulated to represent documents in SDR with good promise recently. However, a major challenge of using IVLM for query modeling is that a query usually consists of only a few words; thus, it is hard to learn a reliable representation accordingly. In this paper, we focus our attention on query reformulation and propose three novel methods on top of IVLM to more accurately represent users' information needs. In addition, we also explore the use of multi-levels of index features, including word- and subword-level units, to work in concert with the proposed methods. A series of empirical SDR experiments conducted on the TDT-2 (Topic Detection and Tracking) collection demonstrate the good effectiveness of our proposed methods as compared to existing state-of-the-art methods.

Owing to the rapidly growing multimedia content available on the Internet, extractive spoken document summarization, with the purpose of automatically selecting a set of representative sentences from a spoken document to concisely express the most important theme of the document, has been an active area of research and experimentation. On the other hand, word embedding has emerged as a newly favorite research subject because of its excellent performance in many natural language processing (NLP)-related tasks. However, as far as we are aware, there are relatively few studies investigating its use in extractive text or speech summarization. A common thread of leveraging word embeddings in the summarization process is to represent the document (or sentence) by averaging the word embeddings of the words occurring in the document (or sentence). Then, intuitively, the cosine similarity measure can be employed to determine the relevance degree between a pair of representations. Beyond the continued efforts made to improve the representation of words, this paper focuses on building novel and efficient ranking models based on the general word embedding methods for extractive speech summarization. Experimental results demonstrate the effectiveness of our proposed methods, compared to existing state-of-the-art methods.

In this paper, a real case study on opinion spammer detection in web forum is presented. We explore user profiles, maximum spamicity of first posts of users, burstiness of registration of user accounts, and frequent poster set to build a model with SVM with RBF kernel and frequent itemset mining. The proposed model achieves 0.6753 precision, 0.6190 recall, and 0.6460 F1 score. The result is promising because the ratio of opinion spammers in the test set is only 0.98%.

Opinion spamming refers to the illegal marketing practice which involves delivering commercially advantageous opinions as regular users. In this paper, we conduct a real case study based on a set of internal records of opinion spams leaked from a shady marketing campaign. We explore the characteristics of opinion spams and spammers in a web forum to obtain some insights, including subtlety property of opinion spams, spam post ratio, spammer accounts, first post and replies, submission time of posts, activeness of threads, and collusion among spammers. Then we present features that could be potentially helpful in detecting spam opinions in threads. The results of spam detection on first posts show: (1) spam first posts put more focus on certain topics such as the user experiences' on the promoted items, (2) spam first posts generally use more words and pictures to showcase the promoted items in an attempt to impress people, (3) spam first posts tend to be submitted during work time, and (4) the threads that spam first posts initiate are more active to be placed at striking positions. The spam detection on replies is more challenging. Besides lower spam ratio and less content, replies even do not mention the promoted items. Their major intention is to keep the discussion in a thread alive to attract more attention on it. Submission time of replies, thread activeness, position of replies, and spamicity of first post are more useful than content-based features in spam detection on replies.

This article explores users' browsing intents to predict the category of a user's next access during web surfing and applies the results to filter objectionable content, such as pornography, gambling, violence, and drugs. Users' access trails in terms of category sequences in click‐through data are employed to mine users' web browsing behaviors. Contextual relationships of URL categories are learned by the hidden Markov model. The top‐level domains (TLDs) extracted from URLs themselves and the corresponding categories are caught by the TLD model. Given a URL to be predicted, its TLD and current context are empirically combined in an aggregation model. In addition to the uses of the current context, the predictions of the URL accessed previously in different contexts by various users are also considered by majority rule to improve the aggregation model. Large‐scale experiments show that the advanced aggregation approach achieves promising performance while maintaining an acceptably low false positive rate. Different strategies are introduced to integrate the model with the blacklist it generates for filtering objectionable web pages without analyzing their content. In practice, this is complementary to the existing content analysis from users' behavioral perspectives.

Chinese spelling check (CSC) is still an unsolved problem today since there are many homonymous or homomorphous characters. Recently, more and more CSC systems have been proposed. To the best of our knowledge, language modeling is one of the major components among these systems because of its simplicity and moderately good predictive power. After deeply analyzing the school of research, we are aware that most of the systems only employ the conventional n -gram language models. The contributions of this article are threefold. First, we propose a novel probabilistic framework for CSC, which naturally combines several important components, such as the substitution model and the language model, to inherit their individual merits as well as to overcome their limitations. Second, we incorporate the topic language models into the CSC system in an unsupervised fashion. The topic language models can capture the long-span semantic information from a word (character) string while the conventional n -gram language models can only preserve the local regularity information. Third, we further integrate Web resources with the proposed framework to enhance the overall performance. Our rigorously empirical experiments demonstrate the consistent and utility performance of the proposed framework in the CSC task.

Extractive text or speech summarization manages to select a set of salient sentences from an original document and concatenate them to form a summary, enabling users to better browse through and understand the content of the document. A recent stream of research on extractive summarization is to employ the language modeling (LM) approach for important sentence selection, which has proven to be effective for performing speech summarization in an unsupervised fashion. However, one of the major challenges facing the LM approach is how to formulate the sentence models and accurately estimate their parameters for each sentence in the document to be summarized. In view of this, our work in this paper explores a novel use of recurrent neural network language modeling (RNNLM) framework for extractive broadcast news summarization. On top of such a framework, the deduced sentence models are able to render not only word usage cues but also long-span structural information of word co-occurrence relationships within broadcast news documents, getting around the need for the strict bag-of-words assumption. Furthermore, different model complexities and combinations are extensively analyzed and compared. Experimental results demonstrate the performance merits of our summarization methods when compared to several well-studied state-of-the-art unsupervised methods.

This study explores the existing blacklists to discover suspected URLs that refer to on-the-fly phishing threats in real time. We propose a PhishTrack framework that includes redirection tracking and form tracking components to update the phishing blacklists. It actively finds phishing URLs as early as possible. Experimental results show that our proactive phishing update method is an effective and efficient approach for improving the coverage of the blacklists. In practice, our solution is complementary to the existing anti-phishing techniques for providing secured web surfing.

Most cross-domain sentiment classification techniques consider a domain as a whole set of opinionated instances for training. However, many online shopping websites organize their data in terms of taxonomy. With multiple domains (or, nodes) organized in a tree-structured representation, we propose a general ensemble algorithm which takes into account: 1) the model application, 2) the model weight and 3) the strategies for selecting the most related models with respect to a target node. The traditional sentiment classification technique SVM and the transfer learning algorithm Spectral Features Alignment (SFA) were applied as our model applications. In addition, the model weight takes the tree information and the similarity between domains into account. Finally, two strategies, cosine function and taxonomy-based regression model (TBRM) are proposed to select the most related models with respect to a target node. Experimental results showed both (cosine function and TBRM) proposed strategies outperform two baselines on an Amazon dataset. Three tasks of the proposed methods surpass the gold standard generated by the in-domain classifiers trained on the labeled data from the target nodes. Good results from the three tasks enable this algorithm to shed some new light on eliminating the major difficulties in transfer learning research: the distribution gap.

Yuen-Hsien Tseng, Lung-Hao Lee, Shu-Yen Lin, Bo-Shun Liao, Mei-Jun Liu, Hsin-Hsi Chen, Oren Etzioni, Anthony Fader. Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics, volume 2: Short Papers. 2014.

Statistical language modeling (LM) that purports to quantify the acceptability of a given piece of text has long been an interesting yet challenging research area.In particular, language modeling for information retrieval (IR) has enjoyed remarkable empirical success; one emerging stream of the LM approach for IR is to employ the pseudo-relevance feedback process to enhance the representation of an input query so as to improve retrieval effectiveness.This paper presents a continuation of such a general line of research and the main contribution is threefold.First, we propose a principled framework which can unify the relationships among several widely-used query modeling formulations.Second, on top of the successfully developed framework, we propose an extended query modeling formulation by incorporating critical query-specific information cues to guide the model estimation.Third, we further adopt and formalize such a framework to the speech recognition and summarization tasks.A series of empirical experiments reveal the feasibility of such an LM framework and the performance merits of the deduced models on these two tasks.

Since more and more multimedia data associated with spoken documents have been made available to the public, spoken document retrieval (SDR) has become an important research subject in the past two decades. The i-vector based framework has been proposed and introduced to language identification (LID) and speaker recognition (SR) tasks recently. The major contribution of the i-vector framework is to reduce a series of acoustic feature vectors of a speech utterance to a low-dimensional vector representation, and then numbers of well-developed postprocessing techniques (such as probabilistic linear discriminative analysis, PLDA) can be readily and effectively used. However, to our best knowledge, there is no research up to date on applying the i-vector framework for SDR or information retrieval (IR). In this paper, we make a step forward to formulate an i-vector based language modeling (IVLM) framework for SDR. Furthermore, we evaluate the proposed IVLM framework with both inductive and transductive learning strategies. We also exploit multi-levels of index features, including word- and subword-level units, in concert with the proposed framework. The results of SDR experiments conducted on the TDT-2 (Topic Detection and Tracking) collection demonstrate the performance merits of our proposed framework when compared to several existing approaches.

In this paper, we describe the development of the tagging editor for learner corpora annotation and computer-aided error analysis. We collect essays written by learners of Chinese as a foreign language for grammatical error annotation and correction. Our tagging editor is effective and enables the annotated corpus to be used in a shared task in ICCE 2014.

In this paper, we present a question-answering (QA) system as a virtual tutor for students in the 5th and 6th grades. Students ask questions and the QA system gives answers to their questions based on a knowledge base. Teaching materials for history and geography are considered as a knowledge source. Because question log is not available in developing QA systems, multiple choice questions (MCQs) in the learning and testing materials are regarded as a training corpus to learn question types, answer types and keywords for retrieval, where an MCQ consists of a stem and a set of options. Options from the same MCQ are grouped into a cluster. Clusters with common elements are merged into a larger cluster. A cluster is labelled with a nominal element selected from the corresponding stems. We also mine question patterns from the stems for question type analysis in the QA system. Because the questions created by instructors in MCQs and the questions asked by students may be different, we develop a procedure to collect possible questions from students in the 6th grade. In the experiments, we first evaluate the question type classification systems using the MCQ corpus and the student corpus with 5-fold cross validation, respectively. Then we train question type classifiers with the complete MCQ corpus, and test them on the student corpus. The student’s and the instructor’s questions are compared and analyzed.

Extractive speech summarization, with the purpose of automatically selecting a set of representative sentences from a spoken document so as to concisely express the most important theme of the document, has been an active area of research and development. A recent school of thought is to employ the language modeling (LM) approach for important sentence selection, which has proven to be effective for performing speech summarization in an unsupervised fashion. However, one of the major challenges facing the LM approach is how to formulate the sentence models and accurately estimate their parameters for each spoken document to be summarized. This paper presents a continuation of this general line of research and its contribution is two-fold. First, we propose a novel and effective recurrent neural network language modeling (RNNLM) framework for speech summarization, on top of which the deduced sentence models are able to render not only word usage cues but also long-span structural information of word co-occurrence relationships within spoken documents, getting around the need for the strict bag-of-words assumption. Second, the utilities of the method originated from our proposed framework and several widely-used unsupervised methods are analyzed and compared extensively. A series of experiments conducted on a broadcast news summarization task seem to demonstrate the performance merits of our summarization method when compared to several state-of-the-art existing unsupervised methods.

This paper explores users' browsing intents to predict the category of a user's next access during web surfing, and applies the results to objectionable content filtering. A user's access trail represented as a sequence of URLs reveals the contextual information of web browsing behaviors. We extract behavioral features of each clicked URL, i.e., hostname, bag-of-words, gTLD, IP, and port, to develop a linear chain CRF model for context-aware category prediction. Large-scale experiments show that our method achieves a promising accuracy of 0.9396 for objectionable access identification without requesting their corresponding page content. Error analysis indicates that our proposed model results in a low false positive rate of 0.0571. In real-life filtering simulations, our proposed model accomplishes macro-averaging blocking rate 0.9271, while maintaining a favorably low macro-averaging over-blocking rate 0.0575 for collaboratively filtering objectionable content with time change on the dynamic web.

Most cross-domain sentiment classification techniques consider a domain as a whole set of instances for training. However, many online shopping websites organize their data in terms of taxonomy. This paper takes Amazon shopping website as an example, and proposes a tree-structured domain representation scheme in which each node in the tree is encoded as a bit sequence to preserve its relationship with all the other nodes in the tree. To select an appropriate source node for training in the domain taxonomy, we propose a Taxonomy-Based Regression Model (TBRM) which predicts the accuracy loss from multiple source nodes to a target node using the tree-structured domain representation combined with domain similarity and domain complexity. The source node with the smallest accuracy loss is used to train a classifier which makes a prediction on the target node. The results show that our TBRM achieves better performance than the regression models without considering the taxonomy information.

Since more and more multimedia data associated with spoken documents have been made available to the public, spoken document retrieval (SDR) has become an important research subject in the past two decades. Recently, topic models have been successfully used in SDR as well as general information retrieval (IR). These models fall into two categories: probabilistic topic models (PTM) and non-probabilistic topic models (NPTM). One major difference between PTM and NPTM is that the former only takes the words occurring in a document into account, whereas the latter, such as latent semantic analysis (LSA), explicitly models all the words in the vocabulary (including both occurring and non-occurring words). We believe that the non-occurring words can provide additional information that is also useful for SDR. However, to our best knowledge, there is a dearth of work investigating the effectiveness of the non-occurring words for SDR and IR. In order to make effective use of those non-occurring words of documents for semantic analysis, we propose a weighted matrix factorization (WMF) framework, in which the impact of the non-occurring words on the semantic analysis can be modulated properly. The results of SDR experiments conducted on the TDT-2 (Topic Detection and Tracking) collection highlight the performance merits of our proposed framework when compared to several existing topic models.

This paper studies the feasibility of an early warning system that prevents users from the dangerous situations they may fall into during web surfing. Our approach adopts behavioral Hidden Markov Models to explore collective intelligence embedded in users' browsing behaviors for context-aware category prediction, and applies the results to web security threat prevention. Large-scale experiments show that our proposed method performs accuracy 0.463 for predicting the fine-grained categories of users' next accesses. In real-life filtering simulations, our method can achieve macro-averaging blocking rate 0.4293 to find web security threats that cannot be detected by the existing security protection solutions at the early stage, while accomplishes a low macro-averaging over-blocking rate 0.0005 with the passage of time. In addition, behavioral HMM is able to alert users for avoiding security threats by 8.4 hours earlier than the current URL filtering engine does. Our simulations show that the shortening of this lag time is critical to avoid severe diffusions of security threats.

In Internet ad campaign, ranking of an ad on search result pages depends on a cost-per-click (CPC) of ad words offered by an advertiser and a quality score estimated by a search engine. Bidding for ad words with a higher CPC is more competitive than bidding for the same ad words with a lower CPC in the ad ranking competition. However, offering a higher CPC will increase a burden on advertisers. In contrast, offering a lower CPC may decrease the exposure rate of their ads. Thus, how to select an appropriate CPC for ad words is indispensable for advertisers. In this paper, we extract different semantic levels of features, such as named entities, topic terminologies, and individual words from a large-scale real-world ad words corpus, and explore various learning based prediction algorithms. The thorough experimental results show that the CPC prediction models considering more ad words semantics achieve better prediction performance, and the prediction model using the support vector regression (SVR) and features from all semantic levels performs the best.

A benchmark evaluation dataset which reflects users' search behaviors in the real world is indispensable for evaluating the performance of information retrieval applications. A typical evaluation dataset consists of a document set, a topic set and relevance judgments. Manual preparation of an evaluation dataset needs much human cost, and human-made topics may not fully capture users' real search needs. This paper aims at automatically constructing an evaluation dataset from wisdom of the crowds in search query logs for information retrieval applications. We begin with collecting documents of clicked documents in search query logs, selecting suitable queries in terms of topics, sampling documents from the document collection for each query and estimating the multi-level relevance of document samples based on click count, normalized count and average count functions. The machine-made evaluation dataset is trained and tested by three learning to rank algorithms, including linear regression, SVMRank and FRank. We compare their performance on a testing collection MQ2007 of LETOR which is a well-known human-made benchmark dataset for learning to rank. The experimental results show that the performance tendency is similar by using machine-made and human-made evaluation datasets. That demonstrates our proposed models can construct an evaluation dataset with similar quality of human-made.

Abstract This article presents a search‐intent‐based method to generate pornographic blacklists for collaborative cyberporn filtering. A novel porn‐detection framework that can find newly appearing pornographic web pages by mining search query logs is proposed. First, suspected queries are identified along with their clicked URLs by an automatically constructed lexicon. Then, a candidate URL is determined if the number of clicks satisfies majority voting rules. Finally, a candidate whose URL contains at least one categorical keyword will be included in a blacklist. Several experiments are conducted on an MSN search porn dataset to demonstrate the effectiveness of our method. The resulting blacklist generated by our search‐intent‐based method achieves high precision (0.701) while maintaining a favorably low false‐positive rate (0.086). The experiments of a real‐life filtering simulation reveal that our proposed method with its accumulative update strategy can achieve 44.15% of a macro‐averaging blocking rate, when the update frequency is set to 1 day. In addition, the overblocking rates are less than 9% with time change due to the strong advantages of our search‐intent‐based method. This user‐behavior‐oriented method can be easily applied to search engines for incorporating only implicit collective intelligence from query logs without other efforts. In practice, it is complementary to intelligent content analysis for keeping up with the changing trails of objectionable websites from users' perspectives.

This paper presents an intent conformity model to collaboratively generate blacklists for cyberporn filtering. A novel porn detection framework via searches-and-clicks is proposed to explore collective intelligence embedded in query logs. Firstly, the clicked pages are represented in terms of the weighted queries to reflect the degrees related to pornography. Consequently, these weighted queries are regarded as discriminative features to calculate the pornography indicator by an inverse chi-square method for candidate determination. Finally, a candidate whose URL contains at least one pornographic keyword is included in our collaborative blacklists. The experiments on a MSN porn data set indicate that the generated blacklist achieves a high precision, while maintaining a favorably low false positive rate. In addition, real-life filtering simulations reveal that our blacklist is more effective than some publicly released blacklists.

This paper presents a user intent method to generate blacklists for collaborative cyberporn filtering. A novel porn detection framework that finds new pornographic web pages by mining user search behaviors is proposed. It employs users' clicks in search query logs to select the suspected web pages without extra human efforts to label data for training, and determines their categories with the help of URL host name and path information, but without web page content. We adopt an MSN porn data set to explore the effectiveness of our method. This user intent approach achieves high precision, while maintaining favorably low false positive rate. In addition, real-life filtering simulation reveals that our user intent method with its accumulative update strategy achieves 43.36% of blocking rate, while maintaining a steadily less than 7% of over-blocking rate.

This paper mines sequences of actions called search scripts from query logs which keep large scale users' search experiences. Search scripts can be applied to predict users' search needs, improve the retrieval effectiveness, recommend advertisements, and so on. Information quality, topic diversity, query ambiguity, and URL relevancy are major challenging issues in search scripts mining. In this paper, we calculate the relevance of URLs, adopt the Open Directory Project (ODP) categories to disambiguate queries and URLs, explore various features and clustering algorithms for intent clustering, and identify critical actions from each intent cluster to form a search script. Experiments show that the model based on a complete link hierarchical clustering algorithm with the features of query terms, relevant URLs, and disambiguated ODP categories performs the best. Search scripts are generated from the best model. When only search scripts containing a single intent are considered to be correct, the accuracy of the action identification algorithm is 0.4650. If search scripts containing a major intent are also counted, the accuracy increases to 0.7315.

Capturing users' future search actions has many potential applications such as query recommendation, web page re-ranking, advertisement arrangement, and so on. This paper predicts users' future queries and URL clicks based on their current access behaviors and global users' query logs. We explore various features from queries and clicked URLs in the users' current search sessions, select similar intents from query logs, and use them for prediction. Because of an intent shift problem in search sessions, this paper discusses which actions have more effects on the prediction, what representations are more suitable to represent users' intents, how the intent similarity is measured, and how the retrieved similar intents affect the prediction. MSN Search Query Log excerpt (RFP 2006 dataset) is taken as an experimental corpus. Three methods and the back-off models are presented.

Abstract As the web has grown into an integral part of daily life, social annotation has become a popular manner for web users to manage resources. This method of management has many potential applications, but it is limited in applicability by the cold‐start problem, especially for new resources on the web. In this article, we study automatic tag prediction for web pages comprehensively and utilize the predicted tags to improve search performance. First, we explore the stabilizing phenomenon of tag usage in a social bookmarking system. Then, we propose a two‐stage tag prediction approach, which is efficient and is effective in making use of early annotations from users. In the first stage, content‐based ranking, candidate tags are selected and ranked to generate an initial tag list. In the second stage, random‐walk re‐ranking, we adopt a random‐walk model that utilizes tag co‐occurrence information to re‐rank the initial list. The experimental results show that our algorithm effectively proposes appropriate tags for target web pages. In addition, we present a framework to incorporate tag prediction in a general web search. The experimental results of the web search validate the hypothesis that the proposed framework significantly enhances the typical retrieval model.

Good and generous knowledge sources, reliable and efficient induction patterns, and automatic and controllable quality assertion approaches are three critical issues to commonsense knowledge (CSK) acquisition. This paper employs Open Mind Common Sense (OMCS), a volunteers-contributed CSK database, to study the first and the third issues. For those stylized CSK, our result shows that over 40% of CSK for four predicate types in OMCS can be found in the web, which contradicts to the assumption that CSK is not communicated in texts. Moreover, we propose a commonsense knowledge classifier trained from OMCS, and achieve high precision in some predicate types, e.g., 82.6% in HasProperty. The promising results suggest new ways of analyzing and utilizing volunteer-contributed knowledge to design systems automatically mining commonsense knowledge from the web.

Identifying intent boundary in search query logs is important for learning users' behaviors and applying their experiences. Time-based, query-based, and cluster-based approaches are proposed. Experiments show that the integration of intent clusters and dynamic time model performs the best.

This paper defines a novel relatedness measure by conditional query, explores snippets in various web domains as corpora, and evaluates the relatedness measure on three famous benchmarks, including WordSimilarity-353, Miller-Charles and Rubenstein-Goodenough datasets. Conditional query QY|X on a web domain estimates frequency fY|X by querying Y to search engine results of X. Dependency score is in terms of frequencies fY|X and fX|Y, and content overlap of search results of X and Y by various operations. A transfer function projects dependency score to mutual dependency of X and Y. Two transfer functions based on Poisson and Gompertz models are considered. Gompertz model reports the correlation score 0.706 in the WordSimilarity-353 dataset. Gompertz model also shows the best performance among all the web-based approaches in Rubenstein-Goodenough and Miller-Charles datasets.

This paper introduces the novel research of emotion analysis from both the writer's and reader's perspectives. A challenge that comes up is the lack of a corpus annotated with both writer and reader emotions. We tackle this problem by combining an online writer-emotion corpus and an online reader-emotion corpus. Statistical analyses are then performed on this newly-generated corpus. It is discovered that there is indeed a relationship between writer and reader emotions.

Abstract We present CopeOpi, an opinion‐analysis system, which extracts from the Web opinions about specific targets, summarizes the polarity and strength of these opinions, and tracks opinion variations over time. Objects that yield similar opinion tendencies over a certain time period may be correlated due to the latent causal events. CopeOpi discovers relationships among objects based on their opinion‐tracking plots and collocations. Event bursts are detected from the tracking plots, and the strength of opinion relationships is determined by the coverage of these plots. To evaluate opinion mining, we use the NTCIR corpus annotated with opinion information at sentence and document levels. CopeOpi achieves sentence‐ and document‐level f‐measures of 62% and 74%. For relationship discovery, we collected 1.3M economics‐related documents from 93 Web sources over 22 months, and analyzed collocation‐based, opinion‐based, and hybrid models. We consider as correlated company pairs that demonstrate similar stock‐price variations, and selected these as the gold standard for evaluation. Results show that opinion‐based and collocation‐based models complement each other, and that integrated models perform the best. The top 25, 50, and 100 pairs discovered achieve precision rates of 1, 0.92, and 0.79, respectively.

Client-side data buffering is a common technique to deal with media playout interruptions of streaming video caused by network jitters and packet losses of best-effort networks. However, stronger playout interruption protection inevitably amounts to larger data buffering and results in more memory requirements and longer playout delay. Adaptive media playout (AMP), also a client-side technique, can reduce the buffer requirement and avoid buffer outage but at the expense of visual quality degradation because of the fluctuation of playout speed. In this paper, we propose a novel AMP scheme to keep the video playout as smooth as possible while adapting to the channel condition. The triggering of the playout control is based on buffer variation rather than buffer fullness. Experimental results show that our AMP scheme surpasses conventional schemes in unfriendly network conditions. Unlike previous schemes that are tuned for a specific range of packet loss and network instability, the proposed AMP scheme maintains consistent performance across a wide range of network conditions.

This paper predicts the stabilized tag set of a resource, with feedback of a small amount of user annotations, aiming to reduce the requirement of sufficient user annotations and to resolve the cold-start problem in a social annotation system.

In this technical demonstration, we showcase an interactive content presentation (ICP) system that integrates media-expressed-emotion-based composition, user-perceived preference feedback, and interactive digital art creation. ICP harmonizes the browsing of multimedia contents by presenting them in the form of music videos (photos, blog articles with accompanied music) based on their expressed emotion similarity. ICP facilitates content browsing by automatically and dynamically selecting the media to be played next in real time, responding to user's preference feedback measured from physiological signals. In addition, ICP enhances the enjoyments of content browsing by incorporating interactive digital art creation. ICP achieves these goals by properly integrating recent researches on media-expressed emotion classification,cross-media composition, and physiological signal processing.

This paper proposes a learning approach for the merging process in multilingual information retrieval (MLIR). To conduct the learning approach, we also present a large number of features that may influence the MLIR merging process; these features are mainly extracted from three levels: query, document, and translation. After the feature extraction, we then use the FRank ranking algorithm to construct a merge model; to our knowledge, this practice is the first attempt to use a learning-based ranking algorithm to construct a merge model for MLIR merging. In our experiments, three test collections for the task of crosslingual information retrieval (CLIR) in NTCIR3, 4, and 5 are employed to assess the performance of our proposed method; moreover, several merging methods are also carried out for a comparison, including traditional merging methods, the 2-step merging strategy, and the merging method based on logistic regression. The experimental results show that our method can significantly improve merging quality on two different types of datasets. In addition to the effectiveness, through the merge model generated by FRank, our method can further identify key factors that influence the merging process; this information might provide us more insight and understanding into MLIR merging.

In this paper, we propose a tag normalization algorithm to unify the userspsila annotations. Meanwhile, we explore some general phenomena in a social annotation system and propose a supervised tag prediction model to predict the stabilized tag set of a resource, with feedback of a small amount of user annotation records. The experiments show that a large potion of the stabilized tag set is predicted, and it is feasible to reduce the requirement of sufficient user annotations in the applications of social annotations.

Past studies on emotion classification focus on the writerpsilas emotional state. This research addresses the reader aspect instead. The classification of documents into reader-emotion categories has several applications. One of them is to integrate reader-emotion classification into a Web search engine to allow users to retrieve documents that contain relevant contents and at the same time instill proper emotions. In this paper, we automatically classify documents into reader-emotion categories, and examine classification performance under different feature settings. Experiments show that certain feature combinations achieve good accuracy. We also compare the best classifierpsilas classification results with the emotional distributions of documents to determine how closely the classifier models the underlying reader behavior. Finally, we investigate the feasibility of emotion ranking.

Summary generation for multiple documents poses a number of issues including sentence selection, sentence ordering, and sentence reduction over single-document summarization. In addition, the temporal resolution among extracted sentences is also important. This article considers informative words and event words to deal with multidocument summarization. These words indicate the important concepts and relationships in a document or among a set of documents, and can be used to select salient sentences. We present a temporal resolution algorithm, using focusing time and coreference chains, to convert Chinese temporal expressions in a document into calendrical forms. Moreover, we consider the last calendrical form of a sentence as a sentence time stamp to address sentence ordering. Informative words, event words, and temporal words are introduced to a sentence reduction algorithm, which deals with both length constraints and information coverage. Experiments on Chinese-news data sets show significant improvements of both information coverage and readability.

We study the classification of news articles into emotions they invoke in their readers. Our work differs from previous studies, which focused on the classification of documents into their authors' emotions instead of the readers'. We use various combinations of feature sets to find the best combination for identifying the emotional influences of news articles on readers.

Ranking problem is becoming important in many fields, especially in information retrieval (IR). Many machine learning techniques have been proposed for ranking problem, such as RankSVM, RankBoost, and RankNet. Among them, RankNet, which is based on a probabilistic ranking framework, is leading to promising results and has been applied to a commercial Web search engine. In this paper we conduct further study on the probabilistic ranking framework and provide a novel loss function named fidelity loss for measuring loss of ranking. The fidelity loss notonly inherits effective properties of the probabilistic ranking framework in RankNet, but possesses new properties that are helpful for ranking. This includes the fidelity loss obtaining zero for each document pair, and having a finite upper bound that is necessary for conducting query-level normalization. We also propose an algorithm named FRank based on a generalized additive model for the sake of minimizing the fedelity loss and learning an effective ranking function. We evaluated the proposed algorithm for two datasets: TREC dataset and real Web search dataset. The experimental results show that the proposed FRank algorithm outperforms other learning-based ranking methods on both conventional IR problem and Web search.

In this paper, we investigate the emotion classification of web blog corpora using support vector machine (SVM) and conditional random field (CRF) machine learning techniques. The emotion classifiers are trained at the sentence level and applied to the document level. Our methods also determine an emotion category by taking the context of a sentence into account. Experiments show that CRF classifiers outperform SVM classifiers. When applying emotion classification to a blog at the document level, the emotion of the last sentence in a document plays an important role in determining the overall emotion.

Abstract Documents discussing public affairs, common themes, interesting products, and so on, are reported and distributed on the Web. Positive and negative opinions embedded in documents are useful references and feedbacks for governments to improve their services, for companies to market their products, and for customers to purchase their objects. Web opinion mining aims to extract, summarize, and track various aspects of subjective information on the Web. Mining subjective information enables traditional information retrieval (IR) systems to retrieve more data from human viewpoints and provide information with finer granularity. Opinion extraction identifies opinion holders, extracts the relevant opinion sentences, and decides their polarities. Opinion summarization recognizes the major events embedded in documents and summarizes the supportive and the nonsupportive evidence. Opinion tracking captures subjective information from various genres and monitors the developments of opinions from spatial and temporal dimensions. To demonstrate and evaluate the proposed opinion mining algorithms, news and bloggers' articles are adopted. Documents in the evaluation corpora are tagged in different granularities from words, sentences to documents. In the experiments, positive and negative sentiment words and their weights are mined on the basis of Chinese word structures. The f‐measure is 73.18% and 63.75% for verbs and nouns, respectively. Utilizing the sentiment words mined together with topical words, we achieve f‐measure 62.16% at the sentence level and 74.37% at the document level.

A web search with double checking model is proposed to explore the web as a live corpus. Five association measures including variants of Dice, Overlap Ratio, Jaccard, and Cosine, as well as Co-Occurrence Double Check (CODC), are presented. In the experiments on Rubenstein-Goodenough's benchmark data set, the CODC measure achieves correlation coefficient 0.8492, which competes with the performance (0.8914) of the model using WordNet. The experiments on link detection of named entities using the strategies of direct association, association matrix and scalar association matrix verify that the double-check frequencies are reliable. Further study on named entity clustering shows that the five measures are quite useful. In particular, CODC measure is very stable on word-word and name-name experiments. The application of CODC measure to expand community chains for personal name disambiguation achieves 9.65% and 14.22% increase compared to the system without community expansion. All the experiments illustrate that the novel model of web search with double checking is feasible for mining associations from the web.

This paper employs ConceptNet, which covers a rich set of commonsense concepts, to retrieve images with text descriptions by focusing on spatial relationships. Evaluation on test data of the 2005 ImageCLEF shows that integrating commonsense knowledge in information retrieval is feasible.

As the blogs have become pervasive media in these years, the number of bloggers has increased at an exponential rate. Mining the bloggers' characteristics, such as age, gender, emotions, etc., has attracted much attention recently. In this paper, we describe our work on the detection of bloggers' interest from three kinds of important features containing in blogs. Textual features include the interest-related words. Temporal features are used to analyze bloggers' posting frequency in order to model the strength of bloggers' interest. Interactive features are represented by the comments posted by bloggers to identify the bloggers' interactivity in blogsphere. By incorporating textual, temporal, and interactive features, our system can be able to automatically identify the interest of bloggers. Experimental results in several topics show that these features work well in the detection of bloggers' interest

This paper considers five features, personal titles, community chains, terms, temporal expressions, and hostnames for personal name disambiguation. In 9 test data sets covering 3 ambiguous personal names, we address the issues of awareness degree of an entity, the source of materials and Web pages in different areas. Two approaches, single-clusterer and cascaded multiple-clusterer, are proposed. In the experiments, the proposed features are quite useful; the multiple-clusterer approach is better than the single-clusterer approach; and expanding community chains using the Web has positive effects on personal name disambiguation

Abstract Named entities are major constituents of a document but are usually unknown words. This work proposes a systematic way of dealing with formulation, transformation, translation, and transliteration of multilingual‐named entities. The rules and similarity matrices for translation and transliteration are learned automatically from parallel‐named‐entity corpora. The results are applied in cross‐language access to collections of images with captions. Experimental results demonstrate that the similarity‐based transliteration of named entities is effective, and runs in which transliteration is considered outperform the runs in which it is neglected.

No abstract available.

This paper presents a framework of using Chinese speech to access images via English captions. The formulation and the structure mapping rules of Chinese and English named entities are extracted from an NICT foreign location name corpus. For

Abstract This article proposes a summarization system for multiple documents. It employs not only named entities and other signatures to cluster news from different sources, but also employs punctuation marks, linking elements, and topic chains to identify the meaningful units (MUs). Using nouns and verbs to identify the similar MUs, focusing and browsing models are applied to represent the summarization results. To reduce information loss during summarization, informative words in a document are introduced. For the evaluation, a question answering system (QA system) is proposed to substitute the human assessors. In large‐scale experiments containing 140 questions to 17,877 documents, the results show that those models using informative words outperform pure heuristic voting‐only strategy by news reporters. This model can be easily further applied to summarize multilingual news from multiple sources.

A WordNet-like linguistic resource is useful, but difficult to construct. This article proposes a method to integrate five linguistic resources, including English/Chinese sense-tagged corpora, English/Chinese thesauruses, and a bilingual dictionary. Chinese words are mapped into WordNet. A Chinese WordNet and a Chinese-English WordNet are derived by following the structures of WordNet. Experiments with Chinese-English information retrieval are developed to evaluate the applicability of the Chinese-English WordNet. The best model achieves 0.1010 average precision, 69.23% of monolingual information retrieval. It also gains a 10.02% increase relative to a model that resolves translation ambiguity and target polysemy problems together.

This paper presents a spoken language processing system for parsing spoken dialogues. The differences between spoken data and written data are clarified. At first, we employ acoustic and prosodic cues to remove noises and identify the linguistic boundaries. Then a fast multi-level chunking-and-raising parser is used to analyze the more "clean " spoken data. The experimental results in parsing a Chinese spoken corpus show that the labeled precision and recall rates are 94.14 % and 90.97%, respectively. 1.

Language barrier is the major problem that people face in searching for, retrieving, and understanding multilingual collections on the Internet. This paper deals with query translation and document translation in a Chinese-English information retrieval system called MTIR. Bilingual dictionary and monolingual corpus-based approaches are adopted to select suitable translated query terms. A machine transliteration algorithm is introduced to resolve proper name searching. We consider several design issues for document translation, including which material is translated, what roles the HTML tags play in translation, what the tradeoff is between the speed performance and the translation performance, and what form the translated result is presented in. About 100,000 Web pages translated in the last four months of 1997 are used for quantitative study of online and real-time Web page translation.

This paper deals with translation ambiguity and target polysemy problems together.Two monolingual balanced corpora are employed to learn word co-occurrence for translation ambiguity resolution, and augmented translation restrictions for target polysemy resolution.Experiments show that the model achieves 62.92% of monolingual information retrieval, and is 40.80% addition to the select-all model.Combining the target polysemy resolution, the retrieval performance is about 10.11% increase to the model resolving translation ambiguity only.

A lazy decision tree approach is described in this paper for the selection of concatenative units for Mandarin speech synthesis. The concept is not to induce a concise hypothesis from a given training data; the selection is delayed until a test instance is given. Thus we can construct the “best” decision tree for each selection. The selection of waveform units is guided with other simultaneously selected prosodic parameters. The selected waveform units can be directly concatenated into output speech or modified with the selected prosodic parameters. This method can produce synthetic speech that sounds very natural and resembles the acoustic and prosodic characteristics of the original speaker. A Mandarin speech synthesizer is described in this paper. However, most of the technologies are language independent and can be extended to a multilingual system.

Spoken languages are more flexible in usage than written languages. Thus, tagging spoken corpus differs from tagging traditional written corpus. This paper proposes a new framework for tagging spoken corpus. The framework adopts the written tagger to process spoken data with the special consideration of the characteristics of spoken language. Besides, the problems of different tagging sets between the written and spoken corpora are also considered in the framework. The presented approach makes an attempt at reducing the differences between these two kinds of language systems and the preliminary tests give very encouraging results. 1.

Speech repairs introduce much noise in spoken language processing. Properly correcting speech repairs can help the speech recognizer to avoid the textual errors, and prevent the interpretation errors during the subsequent processing. Because the task of repair processing cannot defer to the latter (word segmentation, part-of-speech tagging and sentence parsing) stages, this paper employs acoustic and prosodic cues to correct Chinese repetition and addition repairs. The experimental results show that the precision rate of 93.87 % (76.09%) and the recall rate of 90.65 % (70%) can be achieved for correcting Chinese repetition (addition) repairs. 1

To acquire noun phrases from running texts is useful for many applications, such as word grouping, terminology indexing, etc. The reported literatures adopt pure probabilistic approach, or pure rule-based noun phrases grammar to tackle this problem. In this paper, we apply a probabilistic chunker to deciding the implicit boundaries of constituents and utilize the linguistic knowledge to extract the noun phrases by a finite state mechanism. The test texts are SUSANNE Corpus and the results are evaluated by comparing the parse field of SUSANNE Corpus automatically. The results of this preliminary experiment are encouraging.

By corpus analyses, about 75% of Chinese sentences are composed of more than two sentence segments separated by commas or semicolons. A segment may be a sentence, a noun phrase, a verb phrase, an adjective phrase, an adverbial phrase, or a prepositional phrase. An NP segment may serve as a subject of the next segment or an object of the previous segment. The empty category pro may also appear in the VP segment. The maximal freedom of the uses of pros, the large number of segments, the various segment types, and the associativity problem make sentence parsing difficult. Few parsing systems deal with these problems. The authors regard a segment as a basic parsing unit. It also uses characteristic words, subcategories of verbs, topic chains and some heuristic rules to link the segments into meaningful units. The pro resolution and segment linking are useful for practical applications.

The movement of constituents in natural language is a very common linguistic phenomenon— topicalization and relativization in Mandarin Chinese, for example. Any successful natural language processing system must be able to deal with these movements and, at the same time, justify them. However, it is difficult to achieve these goals due to the gaps in between the constituents. In this paper, a logic programming approach with Chomsky’s Government-Binding Theory (GB) attempts to solve this type of problems in Chinese language processing. According to GB, the rule of “move-α” moves anything anywhere, and the universal principles operate interactively to rule the illegal movements out. With this point of view, no specific movement constraints are specified in our logic programming approach. The universal principles are embedded in the logic grammars implicitly, and generated by a translator. The specific features of this approach enable grammar-writing to capture extrapositions in Chinese much easier than the other approaches.