A corpus-based lexicon building in Indonesian political context through Indonesian online news media

Considering public opinion has always been a necessity for most people including governments and politicians. This information provides more direct means in determining public views which important for their decision-making process. With technology and the Internet nowadays, people are able to assess public opinion by using opinion mining or sentiment analysis. There are several known methods for this technology, for instance is lexicon-based method which is inherited from sentiment classification approach. This method uses lexicon in determining sentiment of particular object within related data sets. This paper solely concentrates on building the lexicon for the method. By focusing on Indonesian politic, we create a corpus-based approach to build a contextual lexicon which uses news articles as corpora. We determine the initial seed words and have it validated by domain experts for our experiment Based on the tests that we have done, we find that 51.79 per cent of the terms in our lexicon are relevant to our research domain. We use this finding to evaluate and improve our method as we continue the research to obtain more relevant result.