Integrating Punctuation Rules and Naïve Bayesian Model for Chinese Creation Title Recognition

R. Dale et al. (Eds.): IJCNLP 2005, LNAI 3651, pp. 838 – 848, 2005. © Springer-Verlag Berlin Heidelberg 2005 Integrating Punctuation Rules and Naïve Bayesian Model for Chinese Creation Title Recognition Conrad Chen and Hsin-Hsi Chen Department of Computer Science and Information Engineering, National Taiwan University, Taipei, Taiwan drchen@nlg.csie.ntu.edu.tw; hhchen@csie.ntu.edu.tw http://nlg.csie.ntu.edu.tw/ Abstract. Creation titles, i.e. titles of literary and/or artistic works, comprise over 7% of named entities in Chinese documents. They are the fourth large sort of named entities in Chinese other than personal names, location names, and organization names. However, they are rarely mentioned and studied before. Chinese title recognition is challenging for the following reasons. There are few internal features and nearly no restrictions in the naming style of titles. Their lengths and structures are varied. The worst of all, they are generally composed of common words, so that they look like common fragments of sentences. In this paper, we integrate punctuation rules, lexicon, and naïve Bayesian models to recognize creation titles in Chinese documents. This pioneer study shows a precision of 0.510 and a recall of 0.685 being achieved. The promising results can be integrated into Chinese segmentation, used to retrieve relevant informa- tion for specific titles, and so on. 1 Introduction Named entities are important constituents to identify roles, meanings, and relation- ships in natural language sentences. However, named entities are productive, so that it is difficult to collect them in a lexicon exhaustively. They are usually “unknown” when we process natural language sentences. Recognizing named entities in docu- ments is indispensable for many natural language applications such as information retrieval [2], summarization [3], question answering [7], and so on. Identifying named entities is even harder in Chinese than in many Indo-European languages like English. In Chinese, there are no delimiters to mark word boundaries and no special features such as capitalizations to indicate proper nouns, which constitute huge part of named entities. In the past, various approaches [1, 4, 10] have been pro- posed to recognize Chinese named entities. Most of them just focused on MUC-style named entities [8], i.e., personal names, location names, and organization names. The extensive studies cover nearly 80% of named entities in real documents [1]. Although the performance of such kinds of named entity recognizers is satisfiable, the rest 20% of named entities are so far rarely mentioned and often ignored in previous studies. These rarely mentioned ones belong to various sorts, such as terminologies, aliases and nicknames, brands, etc. These sorts may not occur as frequently as personal names or location names in a corpus, but the importance of the former in documents Integrating Punctuation Rules and Naïve Bayesian Model 839 of specific domains is no less than that of the latter. For example, knowing names of dishes would be very important to understand articles about cooking. Among these rarely addressed named entities, titles of creations, such as book names, song titles, sculpture titles, etc., are one of the most important sorts. According to Chen & Lee (2004)’s study …

Content truncated to 500 words