Construction and application of industrial knowledge atlas (2): representation and modeling of commodity knowledge

This chapter first introduces the evolution of knowledge representation technology ; Then introduce the common ways of industry knowledge modeling ; Then a practical case of commodity knowledge modeling in industrial application scenarios is given , And focus on the knowledge system related to the knowledge map in the e-commerce field ; Finally, the application scenarios and future prospects of commodity knowledge modeling are introduced .

2.1 Introduction to knowledge representation

Machines only have a lot of knowledge , Especially common sense knowledge , In order to achieve real human like intelligence . Knowledge representation represents all kinds of knowledge in the real world into a structure that can be stored and calculated by computer . The change of knowledge representation technology can be roughly divided into three stages ：1） Knowledge representation and reasoning based on symbolic logic , Logical representation is close to human natural language , It is one of the earliest knowledge representation methods ;2） With the concept of Semantic Web proposed , Knowledge representation technology of World Wide Web content is gradually emerging , At present, most of the large-scale applications in industry are the representation methods based on resource description framework triples ;3） With the field of natural language processing word vector （Word Embedding）[1] Equivector representation [2]（Embedding） The emergence of Technology , The research of using continuous vector to express knowledge is gradually rising , And symbolic logic as the basic knowledge representation method , It has become a research hotspot of knowledge representation [1].

2.1.1 Knowledge representation method based on symbolic logic

1. First order predicate logic

First order predicate logic （First Order Logic） It is based on mathematical logic , Pass proposition 、 Logical connectives 、 Knowledge representation composed of predicates and Quantifiers , It is close to natural language in form and convenient to store in computer , It is a representation method that was first applied to artificial intelligence .

First order