Finding Core Topics: Topic Extraction with Clustering on Tweet

• Published in: 2012 International Conference on Cloud and Green Computing pp. 777 - 782
• Main Authors: Sungchul Kim, Sungho Jeon, Jinha Kim, Young-Ho Park, Hwanjo Yu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2012
• Subjects: Clustering algorithms; document clustering; Encyclopedias; Internet; social network; topic extraction; Twitter; Vectors
• ISBN: 1467330272; 9781467330275
• DOI: 10.1109/CGC.2012.120

Twitter is one of the most popular microblogging services that lets users post short text called Tweet. Tweet is distinguished from conventional text data in that it is typically composed of short and informal message, and it makes typical text analysis methods do not work well. Accordingly, extracting meaningful topics from tweets brings up new challenges. In this work, we propose a simple and novel method called Core-Topic-based Clustering (CTC), which extracts topics and cluster tweets simultaneously based on the clustering principles: minimizing the inter-cluster similarity and maximizing the intra-cluster similarity. Experimental results show that our method efficiently extracts meaningful topics, and the clustering performance is better than K-means algorithm.