An Automated Question Answering Method Based on Variable Binding and Relation Activation

Abstract

The invention belongs to the field of natural language processing, and particularly relates to an inference method based on variable binding and relationship activation. In order to solve the problemthat in the prior art, it is difficult to imitate the human brain to infer results, the method comprises the steps that a text and an interrogative sentence are encoded, and the semantic information of the text and the inquire statement of the interrogative sentence are obtained; an entity in the text is recognized, mapped and tracked, a first action network is utilized to understand the semanticinformation, and the position of the entity in a logic rule is determined; a second action network is utilized to activate an entity relationship, the entity relationship is written into the logic rule, and an image schema is obtained; the image schema is retrieved by using the inquire statement, and answer words are generated according to a retrieved result. In the method, the position of the entity and the entity relationship are written into the logic rule, then the image schema is generated, the image schema can exist as prior knowledge, variables similar to the human brain and the relationship sequence can be generated without the dependence on a lot of data, and the right answer is generated.

Description

technical field [0001] The invention belongs to the field of natural language processing, and in particular relates to an automatic question answering method based on variable binding and relationship activation. Background technique [0002] Automatic question answering is an important part of natural language processing. Its manifestation is usually by reading the text, understanding the narrative paragraphs contained in the text, and answering the questions given in the paragraph content. Automatic question answering needs to be able to accurately understand the text in the text. concepts and relationships, and the ability to reason. The current automatic question answering method is to use the existing named entity recognition, semantic role labeling, and syntactic analysis tools to obtain the sentence meaning representation, substitute the sentence meaning representation into logical rules, encode all the sentences in the paragraph and store them directly. The force me...

AI Technical Summary

Problems solved by technology

[0010] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem that the prior art relies too much on the amount of data and lacks prior knowledge of conceptual categories and opposite relations, which makes it difficult to simulate the accurate reasoning results of the human brain, the present invention provides An automatic question answering method based on variable binding and relation activation, including:

Images