Pendekatan Teknik Pengecaman Entiti Nama Bagi Capaian Berita Jenayah Bahasa Melayu (Named Entity Recognition Approach for Malay Crime News Retrieval)

Abstract

© 2018, Universiti Kebangsaan Malaysia Press. All rights reserved. Information extraction is a process of obtaining an important concept in representing the textual content of unstructured documents. At present, there are a lot of unstructured documents such as news, articles, blogs, forums, tweets and micro-blogs of social networks. These documents are very difficult to be understood by the computer. Therefore, studies on the extraction of information is very important to overcome this problem. One extractiontechnique that is widely used is the entity name recognition. This research aims to implement the entity name recognition techniques of crime news source document in Malay language. The main objective of this study is to develop a prototype system model information extraction crime news in the Malay language using name entity recognition through a rule-based approach. This assessment is done by creating a corpus of crime news in the Malay language which is derived from the archival source; BERNAMA news. The corpus is then examined manually by linguists to identify individual entities such as name, organization, location, date, time, financial, percentage, crime and weapons. At the same time, a prototype system was developed and tested with the same corpus and the results of these tests were compared with the results of an expert. Overall, these tests showed good results with the findings for the recall at 78.67%, while precision is at 71.11% and for F-measure at 74.7%. The results of this study are expected to contribute knowledge regarding the effectiveness of the entity's name recognition techniques for crime news Malay language. This could further assist investigators, police, lawyers and authorities involved in the field of crime in solving crimes more quickly and effectively.