xcit'ed

Abstract. This paper explores different strategies for extracting similarity relations between words from partially parsed text corpora. The strategies we have analysed do not require supervised training nor semantic information available from general lexical resources. They differ in the amount and the quality of the syntactic contexts against which words are compared. The paper presents in details the notion of syntactic context and how syntactic information could be used to extract semantic regularities of word sequences. Finally, experimental tests with Portuguese corpus demonstrate that similarity measures based on fine-grained and elaborate syntactic contexts perform better than those based on poorly defined contexts.

The term word association is used in a very particular sense in the p!ycholinguistic literature. (Generally speaking, subjects respond quicker than normal to the word "nurse" if it follows a highly associated word such as "doctor.") We will extend the term to provide the basis for a statistical description of a variety of interesting linguistic phenomena, ranging from semantic rehtions of the doctor/nurse type (content word/content word) to lexico-syntactlc co-occurrence constraints between

Abstract: Context is used in many NLP systems as an indicator of a term's syntactic and semantic function. The accuracy of the system is dependent on the quality and quantity of contextual information available to describe each term. However, the quantity variable is no longer fixed by limited corpus resources. Given fixed training time and computational resources, it makes sense for systems to invest time in extracting high quality contextual information from a fixed corpus. However, with an effectively limitless quantity of text available, extraction rate and representation size need to be considered. We use thesaurus extraction with a range of context extracting tools to demonstrate the interaction between context quantity, time and size on a corpus of 300 million words.

This paper describes a method to construct related terms thesauri automatically based on Web information. We utilize Web search engine to obtain word co-occurrence information and propose a new efficient similarity metrics applying \chi^2 value to solve problems of the existing methods. We also introduce a new method to identify related terms using word-clustering. We do word-clustering on that associative network to identify related terms using latest clustering methods, "Newman method". We make evaluations and show the effectiveness of our approach using sets of related terms extracted from a corpus and a current thesaurus.

This paper presents Japanese morphological analysis based on Conditional Random Fields (CRF). Previous work in CRF assumed that observation sequence (word) boundaries were fixed. However, word boundaries are not clear in Japanese, and hence a straightforward application of CRF is not possible. We show how CRF can be applied to situations where word boundary ambiguity exists. CRF offer an elegant solution to the long-standing problems in Japanese morphological analysis using HMM or MEMM. First, flexible feature designs for hierarchical tagsets become possible. Second, influences of label and length bias are minimized. The former compensate weakness in HMM, while the latter overcomes noticed problems in MEMM. We experiment with CRF, HMM, and MEMM on Japanese annotated corpora, and CRF outperform the other approaches.

There have been many proposals to compute similarities between words based on their distributions in contexts. However, these approaches do not distinguish between synonyms and antonyms. We present two methods for identifying synonyms among distributionally similar words.

To improve the proficiency of text processing such as information retrieval or text mining, it is necessary to construct a synonym dictionary, but it is very tiresome to make it by hands. In some fields, such as aviation, synonym nouns are mixed with kanji/hiragana, katakana, alphabet and their abbreviations. As new words always come to be used, the dictionary update is a big issue. In this paper, we propose a tool for constructing a synonym dictionary. The system will return synonym candidates against a query. A synonym can be easily registered in dictionary by looking the synonym candidates. We experimented the system performance by aviation pilot report and evaluated it by average precision.

Abstract: Bootstrapping semantics from text is one of the greatest challenges in natural language learning. We first define a word similarity measure based on the distributional pattern of words. The similarity measure allows us to construct a thesaurus using a parsed corpus. We then present a new evaluation methodology for the automatically constructed thesaurus. The evaluation results show that the thesaurus is significantly closer to WordNet than Roget Thesaurus is.

Abstract: We describe and evaluate experimentally a method for clustering words according to their distribution in particular syntactic contexts. Words are represented by the relative frequency distributions of contexts in which they appear, and relative entropy between those distributions is used as the similarity measure for clustering. Clusters are represented by average context distributions derived from the given words according to their probabilities of cluster membership. In many cases, the clusters can be thought of as encoding coarse sense distinctions. Deterministic annealing is used to find lowest distortion sets of clusters: as the annealing parameter increases, existing clusters become unstable and subdivide, yielding a hierarchical "soft" clustering of the data. Clusters are used as the basis for class models of word coocurrence, and the models evaluated with respect to held-out test data.