Efficient algorithms for Thai tweet summarization

Nowadays, Twitter is one of the most popular microblogging services. A user may follow many people who may post a 140 character status (tweet) often. Thus, if the user does not continuously read tweets, users may find an excessive number of unread tweets. Such incident causes a burden on the user to find the relevant tweet. It is one of the reasons why Twitter can lead users to feel overloaded with information. This article implemented and evaluated six automatic summarization algorithms for finding similar Thai tweets. The experimental results showed that TextRank algorithm performed the best because this algorithm selected the tweets with the highest scores. On the other hand, Hybrid TF-IDF algorithm could detect similar tweets the least because this algorithm calculated the score by taking the sum frequency of words in a tweet instead of considering the similarity in the level of sentences.