A 3-Layer Structural Analysis Framework for Chinese Text Understanding

How to make the machine better understand human language has always been a hot topic. It greatly influences the efficiency and accuracy of knowledge extraction, intelligent question answering and other nature language processing applications. For this reason, researchers put forward multilingual analysis and language modeling methods, for example, the distributed feature representation fitting for the neural network models and so on. This paper argues that language analysis and modeling should start with the characteristics of the language itself. In seeing this, we have thoroughly studied a series of basic NLP analysis techniques and present our own three-layer language processing and analysis model. This model takes into account the characteristics of language at all levels, and tries to preserve most of the attributes of the sentence. Such design deals with the analysis and representation of language from a brand new perspective, which may inspire many other applications in the field of NLP.