NERSIL - the Named-Entity Recognition System for Iban Language

NERSIL: the Named-Entity Recognition System for Iban Language Soo-Fong Yong a, Bali Ranaivo-Malanҫon b, Alvin Yeo Wee b aFaculty of Computer Science and Information Technology, Universiti Malaysia Sarawak 94300 Kota Samarahan, Sarawak, Malaysia soofong0629@gmail.com bFaculty of Computer Science and Information Technology, Universiti Malaysia Sarawak 94300 Kota Samarahan, Sarawak, Malaysia {mbranaivo, alvin}@fit.unimas.my Abstract. This paper presents NERSIL, the first Iban named entity recognition using ANNIE, an information extraction tool available within GATE. We proposed a method for building rules and gazetteers for Iban language. Rules were determined based on named entities that were not recognized by ANNIE. Then, the investigation of the contexts of these non-recognized Iban named entities allowed us to write the rules. NERSIL achieves 76.4% F-measure. Keywords: Named Entity Recognition, GATE, Information Extraction, Rules-based, Gazetteers 1 Introduction Named entities (NEs) are recognized as an important source of information for many applications. In information retrieval, they improve the detection of relevant documents. In machine translation, post-editing is very expensive when the errors of a machine translation system are mainly due to NEs. Named entity recognition (NER) is the process of detecting NEs and classifying them into semantic categories. NERSIL is the first Iban NER. One of the policies of the Malaysian government is to develop more Malaysian digital contents by including indigenous languages like Iban. Presently, NER can accommodate only major languages such as English and Chinese. Indigenous languages are largely neglected. Hence, NER which can accommodate indigenous languages such as Iban is required. Besides the development of the tool itself, our project will provide access to indigenous knowledge. These may in turn encourage more research, which in long term will preserve the culture. In addition, the proposed method may help other under-resourced languages. The paper is structured as follows. Section 2 presents the background and related works. Section 3 explains the proposed method for building Iban rules and gazetteers. Section 4 describes the experiment performed to evaluate NERSIL, and section 5 depicts the analysis of the errors. 2 Section 6 concludes the presentation and highlights our plans for future work. 2.1 Background and Related Works Iban language According to Dewan Bahasa dan Pustaka (Malay for The Institute of Language and Literature) there are 63 indigenous languages in Sarawak, an east Malaysian state. Iban is the largest ethnic group making up about 44% of the population of Sarawak. Iban language is the vernacular for Iban 25th Pacific Asia Conference on Language, Information and Computation, pages 549–558 549 people. The language is written using Latin alphabet. In 2008, Iban language was introduced as a new elective subject in the national fifth form examination. The digitization and content processing of Iban texts are parts of our center research objective. Currently, we have already developed for Iban, a morphological analyzer and generator, a syntactic parser, a POS tagger, a spell checker, and an interactive alignment tool. The development of the Iban NER is in line with the research center objective. Even though a few processing tools have been developed for Iban, this …

Content truncated to 500 words