Unsupervised Named Entity Classification Models and their Ensembles

Unsupervised Named Entity Classification Models and their Ensembles Jae-Ho Kim*, In-Ho Kang, Key-Sun Choi* Korea Advanced Institute of Science and Technology (KAIST) / Korea Terminology Research Center for Language and Knowledge Engineering* (KORTERM) 373-1, Guseong-dong, Yuseong-gu Daejeon, KOREA, 305-701 {jjaeh@world, ihkang@csone, kschoi@world}.kaist.ac.kr Abstract This paper proposes an unsupervised learning model for classifying named entities. This model uses a training set, built automatically by means of a small-scale named entity dictionary and an unlabeled corpus. This enables us to classify named entities without the cost for building a large hand-tagged training corpus or a lot of rules. Our model uses the ensemble of three different learning methods and repeats the learning with new training examples generated through the ensemble learning. The ensemble of various learning methods brings a better result than each individual learning method. The experimental result shows 73.16% in precision and 72.98% in recall for Korean news articles. 1 Introduction Named entity extraction is an important step for various applications in natural language processing. Named entity extraction involves identifying named entities in the text and classifying their types such as person, organization, location, time expressions, numeric expressions, and so on (Sekine and Eriguchi, 2000). One might think the named entities can be classified easily using dictionaries because most of named entities are proper nouns, but this is wrong opinion. As time passes, new proper nouns are created continuously. Therefore it is impossible to add all those proper nouns to a dictionary. Even though named entities are registered in the dictionary it is not easy to decide their senses. They have a semantic (sense) ambiguity that a proper noun has different senses according to the context (Nina Wacholder, et al., 1997). For example, ‘United States’ refers either to a geographical area or to the political body which governs this area. The semantic ambiguity is occured frequently in Korean (Seon, et al. 2001). Let us illustrate this. Example 1 : Location Let’s meet at KAIST. KAIST e-seo man-na-ja . (PN:KAIST) (PP:at) (V:meet) Example 2 : Organization KAIST announced the list of successful candidates. KAIST e-seo hab-gyeok-ja (PN:KAIST) (PP) (N:successful candidates) myeong-dan eul bal-pyo-haet-da . (N:list) (PP) (V:announced) PN : proper noun, N : noun, PP : postposition, V : verb In the above examples, ‘KAIST’ has different categories although same postposition, ‘e-seo’, followed. The classification of named entities in Korean is a little more difficult than in English. There are two main approaches to classify named entities. The first approach employs hand-crafted rules. It costs too much to maintain rules because rules and dictionaries have to be changed according to the application. The second belongs to a supervised learning approach, which employs a statistical method. As it is more robust and requires less human intervention, several statistical methods based on a hidden Markov model (Bikel et al., 1997), a Maximum Entropy model (Borthwich et al., 1998) and a Decision Tree model (Béchet et al. 2000) have been studied. The supervised learning approach requires a hand-tagged training corpus, but it can not …

Content truncated to 500 words