Automatic Text Summarization for Sinhala

Abstract

According to the people life style, people are surrounded with vast amounts of information albeit with less and less time or ability to make sense of it. Automatic summarization first started as early as in the 1950s.In the modern world due to lack of the time, generating an accurate and intelligent summary for a long document or text pieces has become a popular research as well as an industry problem. This research is carried out to find the suitable approaches to address the above mention issue with minimum linguistic resources. This research proposes a solution for summarizing Sinhala text by identifying the most important and relevant sentences based on linguistic and statistical features of a given text, using an unsupervised extractive summarization approach. In order to generate a better summary, keyword and sentence extraction is manipulated by using a graph based TextRank algorithm. The proposed method was evaluated by comparing the machine generated summaries and human generated summaries based on the assumption that human generated summaries are perfect. The critical evaluation was done by using ROUGE-n and F1 Score to ensure the proposed method usability, performance and efficiency. According to the ROUGE-n values it gives more than 60% of recall rate and more than 42% of precision rate. Further, this research provides a benchmark for future research on Sinhala automatic text summarization.