Fang Kong

Exploiting Constituent Dependencies for Tree Kernel-Based Semantic Relation Extraction

This paper proposes a new approach to dynamically determine the tree span for tree kernel-based semantic relation extraction. It exploits constituent dependencies to keep the nodes and their head children along the path connecting the two entities, while removing the noisy information from the syntactic parse tree, eventually leading to a dynamic syntactic parse tree. This paper also explores entity features and their combined features in a unified parse and semantic tree, which integrates both structured syntactic parse information and entity-related semantic information. Evaluation on the ACE RDC 2004 corpus shows that our dynamic syntactic parse tree outperforms all previous tree spans, and the composite kernel combining this tree kernel with a linear state-of-the-art feature-based kernel, achieves the so far best performance.

Building Chinese Discourse Corpus with Connective-driven Dependency Tree Structure

In this paper, we propose a Connectivedriven Dependency Tree (CDT) scheme to represent the discourse rhetorical structure in Chinese language, with elementary discourse units as leaf nodes and connectives as non-leaf nodes, largely motivated by the Penn Discourse Treebank and the Rhetorical Structure Theory. In particular, connectives are employed to directly represent the hierarchy of the tree structure and the rhetorical relation of a discourse, while the nuclei of discourse units are globally determined with reference to the dependency theory. Guided by the CDT scheme, we manually annotate a Chinese Discourse Treebank (CDTB) of 500 documents. Preliminary evaluation justifies the appropriateness of the CDT scheme to Chinese discourse analysis and the usefulness of our manually annotated CDTB corpus.

A Constituent-Based Approach to Argument Labeling with Joint Inference in Discourse Parsing

Discourse parsing is a challenging task and plays a critical role in discourse analysis. In this paper, we focus on labeling full argument spans of discourse connectives in the Penn Discourse Treebank (PDTB). Previous studies cast this task as a linear tagging or subtree extraction problem. In this paper, we propose a novel constituent-based approach to argument labeling, which integrates the advantages of both linear tagging and subtree extraction. In particular, the proposed approach unifies intra- and intersentence cases by treating the immediately preceding sentence as a special constituent. Besides, a joint inference mechanism is introduced to incorporate global information across arguments into our constituent-based approach via integer linear programming. Evaluation on PDTB shows significant performance improvements of our constituent-based approach over the best state-of-the-art system. It also shows the effectiveness of our joint inference mechanism in modeling global information across arguments.

Global Learning of Noun Phrase Anaphoricity in Coreference Resolution via Label Propagation

Knowledge of noun phrase anaphoricity might be profitably exploited in coreference resolution to bypass the resolution of non-anaphoric noun phrases. However, it is surprising to notice that recent attempts to incorporate automatically acquired anaphoricity information into coreference resolution have been somewhat disappointing. This paper employs a global learning method in determining the anaphoricity of noun phrases via a label propagation algorithm to improve learning-based coreference resolution. In particular, two kinds of kernels, i.e. the feature-based RBF kernel and the convolution tree kernel, are employed to compute the anaphoricity similarity between two noun phrases. Experiments on the ACE 2003 corpus demonstrate the effectiveness of our method in anaphoricity determination of noun phrases and its application in learning-based coreference resolution.

Semi-Supervised Learning for Semantic Relation Classification using Stratified Sampling Strategy

This paper presents a new approach to selecting the initial seed set using stratified sampling strategy in bootstrapping-based semi-supervised learning for semantic relation classification. First, the training data is partitioned into several strata according to relation types/subtypes, then relation instances are randomly sampled from each stratum to form the initial seed set. We also investigate different augmentation strategies in iteratively adding reliable instances to the labeled set, and find that the bootstrapping procedure may stop at a reasonable point to significantly decrease the training time without degrading too much in performance. Experiments on the ACE RDC 2003 and 2004 corpora show the stratified sampling strategy contributes more than the bootstrapping procedure itself. This suggests that a proper sampling strategy is critical in semi-supervised learning.

Improve tree kernel-based event pronoun resolution with competitive information

Event anaphora resolution plays a critical role in discourse analysis. This paper proposes a tree kernel-based framework for event pronoun resolution. In particular, a new tree expansion scheme is introduced to automatically determine a proper parse tree structure for event pronoun resolution by considering various kinds of competitive information related with the anaphor and the antecedent candidate. Evaluation on the OntoNotes English corpus shows the appropriateness of the tree kernel-based framework and the effectiveness of competitive information for event pronoun resolution.

A chinese sentence segmentation approach based on comma

Chinese sentence segmentation is considered to be a very fundamental step in natural language processing. A successful solution for sentence boundary detection is a key step in the subsequent NLP tasks, such as parsing and machine translation, etc. In this paper, we consider comma as a sign-of-the-sentence boundary, and then divide it into two major types, i.e., the true (EOS) and the pseudo (Non-EOS). Finally, a system framework of Chinese sentence segmentation based on two-layer classifiers is presented and implemented. The experimental results on Chinese Treebank 6.0. Results show that our model achieve the F-measure of 90.7% overall, which improves by 1.5%.

Exploring Syntactic Features for Pronoun Resolution Using Context-Sensitive Convolution Tree Kernel

This paper proposes to use a convolution kernel over parse tree to model syntactic structure information for pronoun resolution. Our study reveals that the syntactic structure features embedded in a parse tree are very effective for pronoun resolution and these features can be well captured by the context-sensitive convolution tree kernel. Evaluation on the ACE 2003 corpus shows that among all structured syntactic feature space, Shortest Path Tree achieves the best performance. Then we incorporate more features into SPT, result shows that SPT can use successfully with normal features. Finally, we compare our system with other pronoun resolution systems, our results are outstanding in success rate than normal features and tree kernel-based method of Yang.

Research on question classification for Automatic Question Answering

Automatic Question Answering (QA) is a hot topic in both Natural Language Processing (NLP) and Information Retrieval (IR). And question classification is the key step of a successful automatic QA system. In this paper, an SVM-based approach is firstly proposed as our baseline system. Then two additional features, i.e., top-words and dependency relations, are introduced to improve the performance of our baseline system. Experiments on the UIUC corpus show that the introduced features can improve our baseline system significantly. In comparison with the state-of-the-art system, our proposed approach also achieves better performance.

Research of Noun Phrase Coreference Resolution

Coreference resolution is an important subtask in natural language processing systems. The process of it is to find whether two expressions in natural language refer to the same entity in the world. Machine learning approaches to this problem have been reasonably successful, operating primarily by recasting the problem as a classification task. A great deal of research has been done on this task in English, using approaches ranging from those based on linguistics to those based on machine learning. In Chinese, however, much less work has been done in this area. The lack of public resources is a big problem in the research of Chinese NLP. The other problem is that some features are more difficult to get than those features of English. In this paper, We present a noun phrase coreference system that refers to the work of Soon et al. (2001). We also explore the impact of various features on our systems performance. Experiments on the Chinese portion of OntoNotes 3.0 show that the platform achieves a good performance.