International Journal of Computer Processing of Languages

The calculating of similarity is an important process in NLP. As the basis of sentence and article similarity, the research on calculating of word similarity seems especially vital. There are two word similarity calculating methods: semantic based and statistics based. Former semantic based method mostly relies on Wordnet or Hownet. This article introduces a new method which is based on CSD and which also addes a statistics based method. It offers good support for further processing of corpus.

In this study, we propose to use two corpus-driven linguistic approaches for a sense prediction study. We will concentrate on the character similarity clustering approach and the concept similarity clustering approach to predict the senses of non-assigned words by using corpora and tools, such as the Chinese Gigaword Corpus and HowNet. In this study, we will evaluate sense predictions via the sense divisions of Chinese Wordnet (CWN) and Xiandai Hanyu Cidian (Xian Han). Using these corpora, we will determine the clusters of our four target words — chi1 "eat", wan2 "play", huan4 "change", and shao1 "burn" — in order to predict all possible senses and then evaluate them. This process will demonstrate the viability of the corpus-based approaches.

The representation of lexical semantic knowledge has been one of the most important research topics in the field of computational lexical semantics. Among relevant lexical resources, the design architecture of Princeton WordNet is the most popular one. In this paper, however, we argue that the current synset scheme requires more extensions when applied to the analysis of deeper sense structure in Chinese Wordnet. Issues involved include the underlying structure of sense, meaning facet and their relations. Based on a large amount of empirical analysis of sense data, this paper proposes a fine-grained framework in representing lexical semantic knowledge for Chinese Wordnet, which we believe will be an important consideration for the envisioned cross-lingual global wordnet grid construction. The systematic polysemy patterns found among meaning facets can also be used as a human gold standard of hand-annotated data for metonymy resolution task.

The location of a passage is a kind of semantic information that may prove useful for a variety of applications dealing with inference over passages described in natural language texts. In this paper, we propose a method for automatic discovery of pairs of an event and a place term related by the location, such as wash clothes ⇒ laundry room. In contrast to previous approaches that extracting associations between particular actions and locations when those actions occur on dunning likelihood ratio, the underlying assumption of our method is that correlation between an event and a place term is in the regular co-occurrence of a verb and a place noun within locally coherent text. Based on the analogy with the problem of inferring semantic information from text corpus statistical method of Dunning's likelihood ratio is used to score the extracted pairs for association in order to discover the correlation in each pair. In our experimental evaluation, we examine the effect that various statistical methods produce on the accuracy of this model of inferring locations. After that we carried out a direct evaluation of rare pairs against different statistical method.

Tags are widely used in Web 2.0 applications, such as blogs and online bookmarks. Many characteristics of tags have been studied in previous works, but the granularity level of tags still remains untouched. Granularity level describes how focused a tag is, the lower the level, the more concrete concept the tag covers. The paper has three main contributions: 1) We propose a method to discover the concept level of tags automatically. 2) We apply the method to a real-world tag data set, and study the distribution of levels. 3) We perform tag recommendation concerning their concept levels. Experiments are conducted on a large-scale real world Chinese blog data set. We find that most tags are in more concrete levels, and the users tend to choose tags within the same concept level for a document. When predicting tags for a new document, tags in higher concept levels provide little help, yet may hurt the performance.

MARVS is a theory for the representation of verbal semantics. This paper examines two types of modules and two sets of attributes, namely, role modules, event modules, role-internal attributes, and event-internal attributes, to account for the eventive and semantic factors of verbal polysemy. This study focuses on the polysemous verb chī. Based on MARVS, chī is divided into six senses. These senses are derived metonymically or metaphorically from the prototypical sense.