Machine Learning Algorithms for Natural Language Semantics and Cognitive Computing

We present elegant machine learning algorithms to efficiently learn natural language semantics (MLANLP), thus enabling much better Natural Language Computing (NLC) and Cognitive Computing (CC). Our algorithms use human brain-like learning approach and achieve very good generalization on natural language (mainly text) data. Existing machine learning algorithms performs well on numerical data and cannot easily learn semantics of natural language. Such algorithms, however, can address well some specific problems of natural language, like Name Entity Recognition where data can be easily represented by numbers and semantics between words (name and entity) are simple. Besides, the generalization capabilities of existing machine learning algorithms are limited, especially for complex data. The generalization capability for learning semantics of natural language should be very good to ensure reliable NLC and CC. Our MLANLP has good generalization capability, and can also derive new semantics and knowledge, very much needed for NLC and CC.