Discovering region features based on user’s comments

Abstract

With the development of web 2.0 technology, people can not only have access to information on the internet but can express their opinions, engage in on-line discussion and interact within the network’s platform. By analyzing user comment from the same region, we can understand the implied region features and trending topics in that region. Region features can be categorized as an event or topic therefore it can be labeled based on the user’s comment. In this paper, we propose the discovery of similar topics based on semantics and level or extent of attention focusing on the user’s comment data. Semantics represents the user’s comment while level of attention represents the amount of user’s comment on a news topic, therefore, semantics and the level of attention reveals the user’s comment behavior. This paper uses the LDA and K-means clustering algorithm to analyze similar topics in a region and proposes methods to determine region features. By analyzing the region features and the similar region topics, the labeled region topics can be used for advertisement, improve business strategies, and as a reference for regional administration and planning which has a practical significance.