Sujian Li

A Dependency-Based Neural Network for Relation Classification

Previous research on relation classification has verified the effectiveness of using dependency shortest paths or subtrees. In this paper, we further explore how to make full use of the combination of these dependency information. We first propose a new structure, termed augmented dependency path (ADP), which is composed of the shortest dependency path between two entities and the subtrees attached to the shortest path. To exploit the semantic representation behind the ADP structure, we develop dependency-based neural networks (DepNN): a recursive neural network designed to model the subtrees, and a convolutional neural network to capture the most important features on the shortest path. Experiments on the SemEval-2010 dataset show that our proposed method achieves state-of-art results.

Constructing Information Networks Using One Single Model

In this paper, we propose a new framework that unifies the output of three information extraction (IE) tasks - entity mentions, relations and events as an information network representation, and extracts all of them using one single joint model based on structured prediction. This novel formulation allows different parts of the information network fully interact with each other. For example, many relations can now be considered as the resultant states of events. Our approach achieves substantial improvements over traditional pipelined approaches, and significantly advances state-of-the-art end-toend event argument extraction.

Learning Summary Prior Representation for Extractive Summarization

In this paper, we propose the concept of summary prior to define how much a sentence is appropriate to be selected into summary without consideration of its context. Different from previous work using manually compiled documentindependent features, we develop a novel summary system called PriorSum, which applies the enhanced convolutional neural networks to capture the summary prior features derived from length-variable phrases. Under a regression framework, the learned prior features are concatenated with document-dependent features for sentence ranking. Experiments on the DUC generic summarization benchmarks show that PriorSum can discover different aspects supporting the summary prior and outperform state-of-the-art baselines.

Text-level Discourse Dependency Parsing

Previous researches on Text-level discourse parsing mainly made use of constituency structure to parse the whole document into one discourse tree. In this paper, we present the limitations of constituency based discourse parsing and first propose to use dependency structure to directly represent the relations between elementary discourse units (EDUs). The state-of-the-art dependency parsing techniques, the Eisner algorithm and maximum spanning tree (MST) algorithm, are adopted to parse an optimal discourse dependency tree based on the arcfactored model and the large-margin learning techniques. Experiments show that our discourse dependency parsers achieve a competitive performance on text-level discourse parsing.

Exploring hypergraph-based semi-supervised ranking for query-oriented summarization

Traditional graph based sentence ranking algorithms such as LexRank and HITS model the documents to be summarized as a text graph where nodes represent sentences and edges represent pairwise relations. Such modeling cannot capture complex group relationship shared among multiple sentences which can be useful for sentence ranking. In this paper, we propose to take advantage of hypergraph to remedy this defect. In a text hypergraph, nodes still represent sentences, yet hyperedges are allowed to connect more than two sentences. With a text hypergraph, we are thus able to integrate both group relationship and pairwise relationship into a unified framework. Then, a hypergraph based semi-supervised sentence ranking algorithm is developed for query-oriented extractive summarization, where the influence of query is propagated to sentences through the structure of the constructed text hypergraph. When evaluated on DUC datasets, performance of our proposed approach shows improvements compared to a number of baseline systems.

Component-Enhanced Chinese Character Embeddings

Distributed word representations are very useful for capturing semantic information and have been successfully applied in a variety of NLP tasks, especially on English. In this work, we innovatively develop two component-enhanced Chinese character embedding models and their bigram extensions. Distinguished from English word embeddings, our models explore the compositions of Chinese characters, which often serve as semantic indictors inherently. The evaluations on both word similarity and text classification demonstrate the effectiveness of our models.

Bring you to the past: Automatic Generation of Topically Relevant Event Chronicles

An event chronicle provides people with an easy and fast access to learn the past. In this paper, we propose the first novel approach to automatically generate a topically relevant event chronicle during a certain period given a reference chronicle during another period. Our approach consists of two core components – a timeaware hierarchical Bayesian model for event detection, and a learning-to-rank model to select the salient events to construct the final chronicle. Experimental results demonstrate our approach is promising to tackle this new problem.

Experiments of Ontology Construction with Formal Concept Analysis

Introduction Ontologies are constructs of domain-specific concepts, and their relationships are used to reason about or define that domain. While an ontology may be constructed either manually or semi-automatically, it is never a trivial task. Manual methods usually require that the concept architecture be constructed by experts who consult dictionaries and other text sources. For example, the Upper Cyc Ontology built by Cycorp was manually constructed with approximately 3,000 terms (Lenat, 1998). Automatic and semi-automatic methods require two separate steps in which the first step acquires domain-specific terms followed by the second step of identifying relations among them from available lexicons or corpora. As lexicons are a good resource and are helpful for ontology construction, Chapters 5 and 15 discuss the problems involving ontology construction and lexicons. To use the available corpus resource, a common approach for automatic acquisition employs heuristic rules (Hearst, 1992; Maedche and Staab, 2000). However, such a method can only acquire limited relations. One new approach in the automatic construction of ontologies (Cimiano et al ., 2004) is FCA (Formal Concept Analysis), a mathematical data analysis approach based on the lattice theory. Because formal concept lattices are a natural representation of hierarchies and classifications, FCA has evolved from a pure mathematical tool to an effective method in computer science (Stumme, 2002), such as in the automatic construction of an ontology (Cimiano et al ., 2004). The focus of this work is on how to use FCA to construct a domainspecific ontology based on different Chinese data sources.

Interactive Attention Networks for Aspect-Level Sentiment Classification

Aspect-level sentiment classification aims at identifying the sentiment polarity of specific target in its context. Previous approaches have realized the importance of targets in sentiment classification and developed various methods with the goal of precisely modeling their contexts via generating target-specific representations. However, these studies always ignore the separate modeling of targets. In this paper, we argue that both targets and contexts deserve special treatment and need to be learned their own representations via interactive learning. Then, we propose the interactive attention networks (IAN) to interactively learn attentions in the contexts and targets, and generate the representations for targets and contexts separately. With this design, the IAN model can well represent a target and its collocative context, which is helpful to sentiment classification. Experimental results on SemEval 2014 Datasets demonstrate the effectiveness of our model.

A novel topic model for automatic term extraction

Automatic term extraction (ATE) aims at extracting domain-specific terms from a corpus of a certain domain. Termhood is one essential measure for judging whether a phrase is a term. Previous researches on termhood mainly depend on the word frequency information. In this paper, we propose to compute termhood based on semantic representation of words. A novel topic model, namely i-SWB, is developed to map the domain corpus into a latent semantic space, which is composed of some general topics, a background topic and a documents-specific topic. Experiments on four domains demonstrate that our approach outperforms the state-of-the-art ATE approaches.