Question generation, the task of automatically creating questions that can be answered by a certain span of text within a given passage, is important for question-answering and conversational systems in digital assistants such as Alexa, Cortana, Google Assistant and Siri.Recent sequence to sequence neural models have outperformed previous rule-based systems.Existing models mainly focused on using one or two sentences as the input.Long text has posed challenges for sequence to sequence neural models in question generation -worse performances were reported if using the whole paragraph (with multiple sentences) as the input.In reality, however, it often requires the whole paragraph as context in order to generate high quality questions.In this paper, we propose a maxout pointer mechanism with gated self-attention encoder to address the challenges of processing long text inputs for question generation.With sentence-level inputs, our model outperforms previous approaches with either sentence-level or paragraph-level inputs.Furthermore, our model can effectively utilize paragraphs as inputs, pushing the state-of-the-art result from 13.9 to 16.3 (BLEU 4).

Understanding conversations is crucial to enabling conversational search in technologies such as chatbots, digital assistants, and smart home devices that are becoming increasingly popular. Conventional search engines are powerful at answering open domain queries but are mostly capable of stateless search. In this paper, we define a conversational query as a query that depends on the context of the current conversation, and we formulate the conversational query understanding problem as context-aware query reformulation, where the goal is to reformulate the conversational query into a search engine friendly query in order to satisfy users» information needs in conversational settings. Such context-aware query reformulation problem lends itself to sequence to sequence modeling. We present a large scale open domain dataset of conversational queries and various sequence to sequence models that are learned from this dataset. The best model correctly reformulates over half of all conversational queries, showing the potential of sequence to sequence modeling for this task.

Topical classification of user queries is critical for general-purpose web search systems. It is also a challenging task, due to the sparsity of query terms and the lack of labeled queries. On the other hand, search contexts embedded in query sessions and unlabeled queries free on the web have not been fully utilized in most query classification systems. In this work, we leverage these information to improve query classification accuracy.

Query type classification aims to classify search queries into categories like navigational, informational and transactional, etc., according to the type of information need behind the queries. Although this problem has drawn many research attentions, previous methods usually require editors to label queries as training data or need domain knowledge to edit rules for predicting query type. Also, the existing work has been mainly focusing on the classification of informational and navigational query types. Transactional query classification has not been well addressed. In this work, we propose an unsupervised approach for transactional query classification. This method is based on the observation that, after the transactional queries are issued to a search engine, many users will click the search result pages and then have interactions with Web forms on these pages. The interactions, e.g., typing in text box, making selections from dropdown list, clicking on a button to execute actions, are used to specify detailed information of the transaction. By mining toolbar search log data, which records the associations between queries and Web forms clicked by users, we can get a set of good quality transactional queries without using manual labeling efforts. By matching these automatically acquired transactional queries and their associated Web form contents, we can generalize these queries into patterns. These patterns can be used to classify queries which are not covered by search log. Our experiments indicate that transactional queries produced by this method have good quality. The pattern based classifier achieves 83% F1 classification result. This is very effective considering the fact that we do not adopt any labeling efforts to train the classifier.

In traditional clustering methods, a document is often represented as "bag of words" (in BOW model) or n-grams (in suffix tree document model) without considering the natural language relationships between the words. In this paper, we propose a novel approach DGDC (Dependency Graph-based Document Clustering algorithm) to address this issue. In our algorithm, each document is represented as a dependency graph where the nodes correspond to words which can be seen as meta-descriptions of the document; whereas the edges stand for the relations between pairs of words. A new similarity measure is proposed to compute the pairwise similarity of documents based on their corresponding dependency graphs. By applying the new similarity measure in the Group-average Agglomerative Hierarchial Clustering (GAHC) algorithm, the final clusters of documents can be obtained. The experiments were carried out on five public document datasets. The empirical results have indicated that the DGDC algorithm can achieve better performance in document clustering tasks compared with other approaches based on the BOW model and suffix tree document model.

Learning distance metrics is a fundamental problem in machine learning. Previous distance-metric learning research assumes that the training and test data are drawn from the same distribution, which may be violated in practical applications. When the distributions differ, a situation referred to as covariate shift, the metric learned from training data may not work well on the test data. In this case the metric is said to be inconsistent. In this paper, we address this problem by proposing a novel metric learning framework known as consistent distance metric learning (CDML), which solves the problem under covariate shift situations. We theoretically analyze the conditions when the metrics learned under covariate shift are consistent. Based on the analysis, a convex optimization problem is proposed to deal with the CDML problem. An importance sampling method is proposed for metric learning and two importance weighting strategies are proposed and compared in this work. Experiments are carried out on synthetic and real world datasets to show the effectiveness of the proposed method.

This paper investigates how to effectively do cross lingual text classification by leveraging a large scale and multilingual knowledge base, Wikipedia. Based on the observation that each Wikipedia concept is described by documents of different languages, we adapt existing topic modeling algorithms for mining multilingual topics from this knowledge base. The extracted topics have multiple types of representations, with each type corresponding to one language. In this work, we regard such topics extracted from Wikipedia documents as universal-topics, since each topic corresponds with same semantic information of different languages. Thus new documents of different languages can be represented in a space using a group of universal-topics. We use these universal-topics to do cross lingual text classification. Given the training data labeled for one language, we can train a text classifier to classify the documents of another language by mapping all documents of both languages into the universal-topic space. This approach does not require any additional linguistic resources, like bilingual dictionaries, machine translation tools, or labeling data for the target language. The evaluation results indicate that our topic modeling approach is effective for building cross lingual text classifier.

Sentiment classification aims to automatically predict sentiment polarity (e.g., positive or negative) of users publishing sentiment data (e.g., reviews, blogs). Although traditional classification algorithms can be used to train sentiment classifiers from manually labeled text data, the labeling work can be time-consuming and expensive. Meanwhile, users often use some different words when they express sentiment in different domains. If we directly apply a classifier trained in one domain to other domains, the performance will be very low due to the differences between these domains. In this work, we develop a general solution to sentiment classification when we do not have any labels in a target domain but have some labeled data in a different domain, regarded as source domain. In this cross-domain sentiment classification setting, to bridge the gap between the domains, we propose a spectral feature alignment (SFA) algorithm to align domain-specific words from different domains into unified clusters, with the help of domain-independent words as a bridge. In this way, the clusters can be used to reduce the gap between domain-specific words of the two domains, which can be used to train sentiment classifiers in the target domain accurately. Compared to previous approaches, SFA can discover a robust representation for cross-domain data by fully exploiting the relationship between the domain-specific and domain-independent words via simultaneously co-clustering them in a common latent space. We perform extensive experiments on two real world datasets, and demonstrate that SFA significantly outperforms previous approaches to cross-domain sentiment classification.

In this paper, we try to leverage a large-scale and multilingual knowledge base, Wikipedia, to help effectively analyze and organize Web information written in different languages. Based on the observation that one Wikipedia concept may be described by articles in different languages, we adapt existing topic modeling algorithm for mining multilingual topics from this knowledge base. The extracted 'universal' topics have multiple types of representations, with each type corresponding to one language. Accordingly, new documents of different languages can be represented in a space using a group of universal topics, which makes various multilingual Web applications feasible.

Weblogs have become a prevalent source of information for people to express themselves. In general, there are two genres of contents in weblogs. The first kind is about the webloggers' personal feelings, thoughts or emotions. We call this kind of weblogs affective articles. The second kind of weblogs is about technologies and different kinds of informative news. In this paper, we present a machine learning method for classifying informative and affective articles among weblogs. We consider this problem as a binary classification problem. By using machine learning approaches, we achieve about 92% on information retrieval performance measures including precision, recall and F1. We set up three studies on the applications of above classification approach in both research and industrial fields. The above classification approach is used to improve the performance of classification of emotions from weblog articles. We also develop an intent-driven weblog-search engine based on the classification techniques to improve the satisfaction of Web users. Finally, our approach is applied to search for weblogs with a great deal of informative articles.

Chinese Weblogs have been expanded in an incredible speed in recent years. There is plentiful personal information in Weblogs. In this paper, we propose a text classification based approach to automatically identify the interests of a Weblogger. To solve the problems arising out of class Weblog documents, the technique of heterogeneous classifiers combination is used. We also use hierarchical classification technique to identify much specific interests. Experiments show that our interest identification approach has a high accuracy and, for most Webloggers in our experiments, their interests implied in the contents of blogs could be well identified by using this approach