Concept Interpretation by Semantic Knowledge Harvesting

Abstract

Text with limited context is referred as short text. Short text understanding means detecting the concepts mentioned in a short text. Understanding short texts is important to many applications, but challenges abound. The reasons are, firstly, short texts do not always observe the syntax of a written language. Therefore, traditional natural language processing tools cannot be easily implemented. Secondly, short texts usually do not contain sufficient statistical signals to support many state-of-the-art approaches for text mining such as topic modeling. Thirdly, short texts are generated in a huge volume, which increases the difficulty to handle them. The main goal of the paper is to explore the semantics from short text and utilize it for proper decision making. A community blog is implemented in which the registered users in the community can text and the concept of the short text makes it possible to cluster the users. Online and Offline processing is performed. The instance ambiguity scoring and locating substrings in a text which are similar to terms contained in a predefined vocabulary in the offline processing increase the accuracy of the proposed system.