Analisis Sentimen Tokoh Politik pada Situs Berita Menggunakan NER. Studi Kasus: IMMC

Abstract

In the political world, decisions made by the media can be a measuring instrument of the image of a character. In a news created by a news site, it can be categorized as positive and negative news. Currently there are only a few applications that can see and record the news of a character on a news site. Percentage count can be representative of news sites, it can show positive and negative results from rticle on those sites. To classify a news site, need sentiment analysis of each article on the news site. The sentiment analysis results of each article will affect the percentage count. In general, the analysis is done using preprocessing text which is compared with the word sentiment. However, the preprocessing process and the sentiment word are not appropriate if used to analyze the sentiments of an article with using bahasa. Named Entitity Recognition (NER) is part of the word extracted from a collection of texts. NER can be used to extract positive and negative words. In this study, each article from a news site will be analyzed using Named Entity Recognation (NER). The results of sentiment analysis are validated by users. In this study, from 10 test data (articles), the accuracy of sentiment analysis with NER was 90%. While the sentiment analysis using sentiment word and Preprocessing is only 80%.