A Graph based Keyword Extraction from Twitter using Node and Edge Weight

Abstract

With the advent of mobile and internet technology, social media has experienced a tremendous growth over the past few years. Social networking site such as Twitter provides the platform for content generation, information dissemination and communications. Summarizing the textual contents or extracting the influential words is a challenging task. Therefore an unsupervised graph based approach is proposed for automatic Keyword Extraction from the collection of Tweets using Node and Edge Weight (KETNEW). The node weight and edge weight depends on various parameters such as tweet frequency, position of node, clustering coefficient, co-occurrence frequency and shared neighbors. The important aspect of our approach is that it is a hybrid approach i.e. a graph based approach (structural approach) which depends upon statistical as well as linguistic features. Apart from this, it also depends on the position of the word in the text. These numerous features made it perform better than existing techniques.