Pose estimation is an essential technology for industrial robots to perform precise gripping and assembly. The state-of-the-art deep learning-based approach uses an indirect strategy, i.e., first finding local correspondence between the 2-D image and 3-D model, and then using the perspective-n-point and RANSAC methods to calculate the poses of ordinary objects. However, the metal parts in industry are reflective and textureless, making it difficult to identify distinguishable point features to establish 2-D–3-D correspondences. To address this problem, in this article, we propose a novel deep learning based two-stage method for pose estimation of reflective textureless metal parts, which accurately estimates the target pose using monocular red green blue (RGB) images. Since contours play an important role in both keypoints prediction and pose estimation stages, our method is named ContourPose. First, an additional contour decoder is adopted to implicitly constrain the keypoints prediction in the former stage, which improves the accuracy of the keypoints prediction. Then, the predicted contour of the previous stage is taken as geometric prior that is used to iteratively solve for the optimal pose. Experiments indicate that the proposed approach for reflective textureless metal parts has a significant improvement over the state-of-the-art approaches.

Social media streams contain large and diverse amount of information, ranging from daily-life stories to the latest global and local events and news. Twitter, especially, allows a fast spread of events happening real time, and enables individuals and organizations to stay informed of the events happening now. Event detection from social media data poses different challenges from traditional text and is a research area that has attracted much attention in recent years. In this paper, we survey a wide range of event detection methods for Twitter data stream, helping readers understand the recent development in this area. We present the datasets available to the public. Furthermore, a few research opportunities are discussed as potential future research directions.

Millions of patent applications are submitted every year. Patent examiners spend tremendous amount of time to evaluate the quality of them for approval or denial. A system that can automatically evaluate patents and expedite the evaluation process is much needed. This kind of tool can also help the small enterprises, patent attorneys and agents in preparing their patent applications. This study proposes a model that can rate the quality of a patent by utilizing both the semantic and novelty information of the patent. We also built a dataset of more than 32,000 Chinese patents with manual ratings from professional patent examiners. Our experiments show that the proposed model outperforms other approaches.

In this paper, we present a large Chinese news article dataset with 4.4 million articles. These articles are obtained from different news channels and sources. They are labeled with multi-level topic categories, and some of them also have summaries. This is the first Chinese news dataset that has both hierarchical topic labels and article full texts. And it is also the largest Chinese news topic dataset. We describe the data collection, annotation and quality evaluation process. The basic statistics of the dataset, comparison with other datasets and benchmark experiments are also presented.

In this study, we proposed an approach to automatically generating court view from the fact description of a legal case. This is a text-to-text natural language generation problem, and it can help the automatic legal document generation. Due to the specialty of the legal domain, our model exploits the charge and law article information in the generation process, instead of utilizing just the fact description text. The BERT model is used as the encoder and a Transformer architecture is used as decoder. To smoothly integrate these two parts together, we employ two separate optimizers for the two components during the training process. The experiments on two data sets of Chinese legal cases show that our approach outperforms other methods.

Twitter is a fast communication channel for gathering and spreading breaking news, and it generates a large volume of tweets for most events. Automatically creating a summary for an event is necessary and important. In this study, we explored two extractive approaches for summarizing events on Twitter. The first one exploits the semantic types of event related terms, and ranks the tweets based on the score computed from these semantic terms. The second one utilizes a graph convolutional network built from a tweet relation graph to generate tweet hidden features for tweet salience estimation. And the most salient tweets are selected as the summary of the event. Our experiments show that these two approaches outperform the compared methods.

Millions of trademarks were registered last year in China, and thousands of applications are submitted daily. A trademark must be unique in the category it belongs to. Therefore, each new trademark application needs to be checked against all the existing ones in its category. A trademark can be a text string (characters, words or phrases), a figure (symbol or design), or both. In this study, we focus on the textual trademark in Chinese, and propose a model for finding similar trademarks for a given one. This neural network model exploits the semantic, phonetic and visual similarities between two textual trademarks. We evaluated our model based on a dataset that were built from the real trademark application data. Our evaluation shows that the proposed model outperforms other approaches.

There is massive amount of news on financial events every day. In this paper, we present a unified model for detecting, classifying and summarizing financial events. This model exploits a multi-task learning approach, in which a pre-trained BERT model is used to encode the news articles, and the encoded information are shared by event type classification, detection and summarization tasks. For event summarization, we use a Transformer structure as the decoder. In addition to the input document encoded by BERT, the decoder also utilizes the predicted event type and cluster information, so that it can focus on the specific aspects of the event when generating summary. Our experiments show that our approach outperforms other methods.

In this study, we propose a new multi-task learning approach for rumor detection and stance classification tasks.This neural network model has a shared layer and two task specific layers.We incorporate the user credibility information into the rumor detection layer, and we also apply attention mechanism in the rumor detection process.The attended information include not only the hidden states in the rumor detection layer, but also the hidden states from the stance detection layer.The experiments on two datasets show that our proposed model outperforms the state-of-the-art rumor detection approaches.

Recently more and more personal stories about sexual harassment are shared online, mainly inspired by the \#MeToo movement. Safecity is an online forum for victims of sexual harassment to share their personal experience. Previous study applied neural network models to classify the harassment forms of the stories. To uncover patterns of sexual harassment, the extraction of the key elements and the categorization of these stories in different dimensions can be useful as well. In this study, we proposed neural network models to extract key elements including harasser, time, location and trigger words. In addition, we categorized these stories from different dimensions, such as location, time, and harassers' characteristics, including their age range, single/multiple harassers, profession, and relationship with the victims. We further demonstrated that encoding the key element information in the story categorization model can improve its performance. The proposed approaches and analysis would be helpful in automatically filing reports, raising public awareness, making preventing strategies and etc.

Yingchi Liu, Quanzhi Li, Marika Cifor, Xiaozhong Liu, Qiong Zhang, Luo Si. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

TweetSenti is a system for analyzing the sentiment of an entity in tweets. A sentence or tweet may contain multiple entities, and they do not always have the same sentiment polarity. Therefore, it is necessary to detect the sentiment for a specific target entity. This type of target-dependent (entity level) sentiment analysis has become attractive and has been used in many applications, but it is still a challenging task. TweetSenti employs a new approach for detecting the entity level sentiment. Our model splits a sentence into a left context and a right context according to the target entity, and it also exploits two different types of word embeddings to represent a word, the general word embedding and the sentiment specific word embedding. A hybrid neural network is used to capture both the sequence and structure information of the two sides of the target entity. The sequence information is learned by attention-based bi-directional LSTM models. The structure information is captured by multi-context CNN models. Based on this algorithm, we built a web-based application that users can interact with and analyze an entity's sentiment in Twitter at real-time.

Event trigger detection remains a challenging task. Most of previous studies focused on variations of model structures to extract features from the local context of the trigger words. However, few studies focused on the utilization of document level information. In this work, we studied the benefit of exploiting the document level information for event trigger detections in textual data. Two approaches of extracting document features are proposed, and the document features are integrated with the embeddings generated from the local context of the trigger word using a convolutional neural network (CNN) model. Our experiment shows that these two methods both outperform the CNN-based baseline model.

To deal with the sheer volume of information and gain competitive advantage, the news industry has started to explore and invest in news automation. In this paper, we present Reuters Tracer, a system that automates end-to-end news production using Twitter data. It is capable of detecting, classifying, annotating, and disseminating news in real time for Reuters journalists without manual intervention. In contrast to other similar systems, Tracer is topic and domain agnostic. It has a bottom-up approach to news detection, and does not rely on a predefined set of sources or subjects. Instead, it identifies emerging conversations from 12+ million tweets per day and selects those that are news-like. Then, it contextualizes each story by adding a summary and a topic to it, estimating its newsworthiness, veracity, novelty, and scope, and geotags it. Designing algorithms to generate news that meets the standards of Reuters journalists in accuracy and timeliness is quite challenging. But Tracer is able to achieve competitive precision, recall, timeliness, and veracity on news detection and delivery. In this paper, we reveal our key algorithm designs and evaluations that helped us achieve this goal, and lessons learned along the way.

Previous studies have shown that investor sentiment indicators can predict stock market change.A domain-specific sentiment lexicon and sentiment-oriented word embedding model would help the sentiment analysis in financial domain and stock market.In this paper, we present a new approach to learning stock market lexicon from StockTwits, a popular financial social network for investors to share ideas.It learns word polarity by predicting message sentiment, using a neural network.The sentiment-oriented word embeddings are learned from tens of millions of StockTwits posts, and this is the first study presenting sentiment-oriented word embeddings for stock market.The experiments of predicting investor sentiment show that our lexicon outperformed other lexicons built by the state-of-the-art methods, and the sentiment-oriented word vector was much better than the general word embeddings.

In this paper, we present a new approach for detecting novel events from social media, specially Twitter, at real-time. An event is usually defined by who, what, where and when, and an event tweet usually contains terms corresponding to these aspects. To exploit this information, we propose a method that incorporates simple semantics by splitting the tweet term space into groups of terms that have the meaning of the same type. These groups are called semantic categories (classes) and each reflects one or more event aspects. The semantic classes include named entity, mention, location, hashtag, verb, noun and embedded link. To group tweets talking about the same event into the same cluster, similarity measuring is conducted by calculating class-wise similarity and then aggregating them together. Users of a real-time event detection system are usually only interested in novel (new) events, which are happening now or just happened a short time ago. To fulfill this requirement, a temporal identification module is used to filter out event clusters that are about old stories. The clustering module also computes a novelty score for each event cluster, which reflects how novel the event is, compared to previous events. We evaluated our event detection method using multiple quality metrics and a large-scale event corpus having millions of tweets. The experiment results show that the proposed online event detection method achieves the state-of-the-art performance. Our experiment also shows that the temporal identification module can effectively detect old events.

Tables are significant document components. Table extraction and classification are critical for us to explore, retrieve and mine knowledge encoded in tables. This paper presents a learning based approach for classifying tables based on their content and structural information, with focus on financial document tables. To the best of our knowledge, this is the first study on classifying tables in financial domain, and also the first study of table classification based on its semantics, a more fine-grained level than previous studies. The experimental results show that it can effectively classify financial tables. We also analyzed what features are important and how to generate them. The feature identification and generation approach can potentially apply to other domains.

Two of the major problems in social media message classification are the data sparseness issue and the high degree of lexical variation. Paraphrases, or synonyms, are alternative ways of expressing the same meaning using different lexical variations. In this study, we try to use paraphrases to improve tweet topic classification performance. We explored two approaches to generating paraphrases, WordNet, which is a lexical database grouping English words into sets of synonyms, and word embeddings, which are learned from millions of tweets and billions of words. Our experiment shows that using paraphrases can improve the topic classification task, and the word embedding approach outperforms the WordNet method. To our knowledge, this is the first study exploiting paraphrases for tweet classification.

Classifying tweets into topic categories is necessary and important for many applications, since tweets are about a variety of topics and users are only interested in certain topical areas. Many tweet classification approaches fail to achieve high accuracy due to data sparseness issue. Tweet, as a special type of short text, in additional to its text, also has other metadata that can be used to enrich its context, such as user name, mention, hashtag and embedded link. In this demonstration, we present TweetSift, an efficient and effective real time tweet topic classifier. TweetSift exploits external tweet-specific entity knowledge to provide more topical context for a tweet, and integrates them with topic enhanced word embeddings for topic classification. The demonstration will show how TweetSift works and how it is incorporated with our social media event detection system.

In this paper, we present a new approach of recommending hashtags for tweets. It uses Learning to Rank algorithm to incorporate features built from topic enhanced word embeddings, tweet entity data, hashtag frequency, hashtag temporal data and tweet URL domain information. The experiments using millions of tweets and hashtags show that the proposed approach outperforms the three baseline methods -- the LDA topic, the tf.idf based and the general word embedding approaches.

News professionals are facing the challenge of discovering news from more diverse and unreliable information in the age of social media. More and more news events break on social media first and are picked up by news media subsequently. The recent Brussels attack is such an example. At Reuters, a global news agency, we have observed the necessity of providing a more effective tool that can help our journalists to quickly discover news on social media, verify them and then inform the public.

Previous studies have used many manually identified features and word embeddings for tweet sentiment classification. In this paper, we propose a new approach, which incorporates sentiment-specific word embeddings (SSWE) and a weighted text feature model (WTFM). WTFM produces features based on text negation, tf.idf weighting scheme, and a Rocchio text classification method. Compared to other tweet sentiment feature generation approaches, WTFM is easy to build, simple, yet effective. Experiments show that the proposed approach outperforms the two state-of-the-art tweet sentiment classification methods, SSWE and National Research Council Canada's (NRC) model.

Many classification tasks on short text, such as tweet, fail to achieve high accuracy due to data sparseness. One approach to solving this problem is to enrich the context of data by using external data sources, or distributed language representations trained on huge amount of data. In this paper, we present several tweet topic classification methods by exploiting different types of data: tweet text, tweet text plus entity knowledge base, word embeddings derived from tweet text, distributed representations of tweets, and topical word embeddings. The word embedding, topical word embedding and sentence representation models are generated from billions of words from tweets without supervision. To the best of our knowledge, this is the first study of applying distributed language representations to tweet topic classification task.

In this paper, we propose the first real time rumor debunking algorithm for Twitter. We use cues from 'wisdom of the crowds', that is, the aggregate 'common sense' and investigative journalism of Twitter users. We concentrate on identification of a rumor as an event that may comprise of one or more conflicting microblogs. We continue monitoring the rumor event and generate real time updates dynamically based on any additional information received. We show using real streaming data that it is possible, using our approach, to debunk rumors accurately and efficiently, often much faster than manual verification by professionals.

Abstract On the Web, there are limited ways of finding people sharing similar interests with a given person. The current methods are either ineffective or time consuming. In this paper, we present a new approach for searching people sharing similar interests from the Web. Given a person, to find similar people from the Web, there are two major research issues: person representation and matching persons. In this study, we propose a person representation method which uses a person's website to represent this person. Our design of matching process takes person representation into consideration to allow the same representation to be used when composing the query. Under this person representation method, the proposed algorithm integrates textual content and hyperlink information of all the pages belonging to a personal website to represent a person and match persons. Other algorithms are also explored and compared to the proposed algorithm. Experimental results are presented.

This paper presents an approach for an extensible multimodal interaction dialogue system, R-Flow, based on a recursive application of Model-View-C'ontroller (MVC) design patterns to derive system components and interfaces. This approach leads to a clear separation of three self-contained functional layers in a multimodal dialogue system: modality independent dialog control, synchronization of logical modalities, and physical presentation. These layers are codified and weaved together through standard based XML languages. In particular, it utilizes the standard State-Chart XML (SCXML) for dialog control, SMIL and EMMA based XM-Flow for modality synchronization and interpretation, and a generic XML based binding mechanism to map logical modalities to physical presentations. A prototype system has been implemented for multimodal (e.g. speech, text, and mouse) manipulation of Google map. Our experimental results indicated that such layered and component-based XML MMI system is feasible and the performance of such MMI system is studied and measured.

Abstract Document keyphrases provide a concise summary of a document's content, offering semantic metadata summarizing a document. They can be used in many applications related to knowledge management and text mining, such as automatic text summarization, development of search engines, document clustering, document classification, thesaurus construction, and browsing interfaces. Because only a small portion of documents have keyphrases assigned by authors, and it is time‐consuming and costly to manually assign keyphrases to documents, it is necessary to develop an algorithm to automatically generate keyphrases for documents. This paper describes a Keyphrase Identification Program (KIP), which extracts document keyphrases by using prior positive samples of human identified phrases to assign weights to the candidate keyphrases. The logic of our algorithm is: The more keywords a candidate keyphrase contains and the more significant these keywords are, the more likely this candidate phrase is a keyphrase. KIP's learning function can enrich the glossary database by automatically adding new identified keyphrases to the database. KIP's personalization feature will let the user build a glossary database specifically suitable for the area of his/her interest. The evaluation results show that KIP's performance is better than the systems we compared to and that the learning function is effective.

This paper presents a hybrid concept hierarchy development technique for web returned documents retrieved by a meta-search engine. The aim of the technique is to separate the initial retrieved documents into topical oriented categories, prior to the actual concept hierarchy generation. The topical categories correspond to different semantic aspects of the query. This is done using a 1-of-n automatic document classification, on the initial set of returned documents. Then, an individual topical concept hierarchy is automatically generated inside each of the resulted categories. Both steps are executed on the fly at retrieval time. Due to the efficiency constraints imposed by the web retrieval context, the algorithm only uses document snippets (rather than full web pages) for both document classification and concept hierarchy generation. Experimental results show that the algorithm is able to improve the quality of the concept hierarchy presented to the searcher; at the same time, the efficiency parameters are kept within reasonable intervals.

Document keyphrases provide semantic metadata characterizing documents and producing an overview of the content of a document. They can be used in many text-mining and knowledge management related applications. This paper describes a Keyphrase Identification Program (KIP), which extracts document keyphrases by using prior positive samples of human identified domain keyphrases to assign weights to the candidate keyphrases. The logic of our algorithm is: the more keywords a candidate keyphrase contains and the more significant these keywords are, the more likely this candidate phrase is a keyphrase. To obtain prior positive inputs, KIP first populates its glossary database using manually identified keyphrases and keywords. It then checks the composition of all noun phrases of a document, looks up the database and calculates scores for all these noun phrases. The ones having higher scores will be extracted as keyphrases.