Intelligent semantic matching method and device based on depth feature dimension changing mechanism

Abstract

The invention discloses an intelligent semantic matching method and device based on a depth feature dimension changing mechanism, and belongs to the technical field of artificial intelligence and natural language processing. The technical problem to be solved by the invention is how to capture more semantic context information and interaction information between sentences. Intelligent semantic matching of sentences is realized; the adopted technical scheme is as follows: the method comprises the following steps: constructing and training a sentence matching model consisting of an embedding layer, a depth feature variable-dimension coding layer, a convolution matching layer and a prediction layer; according to the method, deep feature variable-dimension coding representation of the sentences is realized, so that more semantic context information and interaction information between the sentences are obtained, and meanwhile, a convolution matching mechanism is realized, so that the purpose of intelligent semantic matching of the sentences is achieved. The device comprises a sentence matching knowledge base construction unit, a training data set generation unit, a sentence matching model construction unit and a sentence matching model training unit.

Description

technical field [0001] The invention relates to the technical fields of artificial intelligence and natural language processing, in particular to an intelligent semantic matching method and device based on a deep feature variable dimension mechanism. Background technique [0002] In recent years, semantic matching methods for sentences have received more and more attention in the field of natural language processing. The reason is that many natural language processing tasks are based on the semantic matching of sentences, which can be regarded as an extension of sentence semantic matching tasks to a certain extent. For example, the task of "automatic question answering" can be processed by calculating the matching degree of "question" and "candidate answer"; the task of "information retrieval" can be regarded as calculating the matching degree of "query sentence" and "matching document". Because of this, semantic matching of sentences plays a crucial role in the field of na...

AI Technical Summary

Problems solved by technology

[0003] Through analysis and research, we found that most of the existing methods are based on convolutional neural network model or recurrent neural network model, and the characteristics and limitations of these two models make it impossible to completely solve this problem.

For example, although the convolutional neural network is good at capturing and representing local features with different kernel functions, it ignores the sequence information in the text, and due to the characteristics of the text data itself, only 1D can be selected when using the convolutional neural network model. The convolution kernel, which greatly affects its application in the field of text; although cyclic neural networks can process sequence information, most of them only encode text data at a specific angle, such as only hyphenating text data Make it into a set of single characters before encoding, or only perform word segmentation to make it into a set of words before encoding. This way of encoding data from a single angle may lead to the loss of some important hierarchical encoding information.

For the semantic matching task of sentences, the order of words in sentences and the hierarchical information of sentences are very important. Therefore, it is almost impossible to obtain satisfactory result

Images