Concept graph based semantic matching of articles

Abstract

Natural Language Processing (NLP) is a luring area to explore. It allows machines to directly “understand” natural language, therefore operation based on text can be processed without further disposal, and human orders can be taken and implemented by machines without further programming, which enhance the user-friendliness for many industries. Past years have seen a rapid improvement of NLP. In current NLP technology, keyword detection is widely used for matching articles. However, this method overlooked the semantics of articles. On the other hand, the existing models targeting at semantic analysis take up large computational capacity. In this project, Concept Interaction Graph (CIG), a model generating semantic graphs from articles, was studied.