Composing Indonesian Paragraph for Biography Domain using Extracive Summarization

A main problem for Wikipedia and other wiki-based websites is the need of human resource to create new articles and update existing articles. Furthermore, the biography domain is one of the most active domains as each person's article need to be updated as time goes by. In this paper, we investigate an approach to update Indonesian Wikipedia articles, specifically in person biography domain, automatically by practicing extractive summarization of information drawn from online news portal. We use news articles about a person to compose candidate paragraphs for updating the person's Wikipedia article. The candidate paragraphs would be filtered to reduce the possibility of redundancy and given a section recommendation based on the existing sections in the person's current Wikipedia article. Given a person's name, we show that our model can find and extract relevant information about the person as reflected by the ROUGE score for summarization and human evaluation for the overall result.