Convolutional neural network-based document automatic question and answer system construction method

Abstract

The invention discloses a convolutional neural network-based document automatic question and answer system construction method. The method comprises the following steps of: 1, constructing a theme document library; 2, constructing a word vector model; 3, carrying out theme matching; 4, constructing a word vector matrix; 5, carrying out semantic matching on the basis of a semantic model of a convolutional neural network, wherein the semantic model of the convolutional neural network is divided into three layers, the first layer is a convolutional neural network layer, the second layer is an attention layer and the third layer is a full connection layer; and 6, an answer selection process: selecting a matched answer. According to the method, a synonym dictionary does not need to be manuallyconstructed so that plenty of manpower and time cost are saved, semantic meanings of word contexts can be purposely sampled in a model training process, and an attention mechanism is added in the network so that the contribution degrees, to semantic meanings of whole sentences, of certain representative words can be enhanced.

Description

technical field [0001] The present invention relates to the fields of natural language processing and artificial intelligence. Using a convolutional neural network algorithm, a method for semantic modeling and semantic matching of questions and answers is proposed under the background of large-scale application of deep learning algorithms to natural language processing. Program. Background technique [0002] The most important thing in automatic question answering is the sentence semantic matching technology. Most of the traditional methods are based on HowNet (HowNet), large-scale dictionaries, and Harbin Institute of Technology Synonyms Cilin tools, and construct a combination of rules that meet the scene to achieve the purpose of calculation. The advantage of this type of method is that it can quickly establish a calculation model, effectively use the semantics of the word itself, and quickly adjust the model. The disadvantage is that this method cannot effectively use th...

AI Technical Summary

Problems solved by technology

The advantage of this type of method is that it can quickly establish a calculation model, effectively use the semantics of the word itself, and quickly adjust the model. The disadvantage is that this method cannot effectively use the semantics of the word context, or even the semantics of the entire sentence or paragraph.

From this point of view, the traditional method is easy to cause the semantic loss of the word context, and the obtained results cannot be accurately used to calculate the matching degree between sentences

[0003] Some traditional semantic matching methods for sentences are not as good as the current popular deep learning methods in terms of matching effect due to the inability to effectively use the semantics of word context and the high demand for manpower and time costs, and it is difficult to adapt to the Internet with explosive growth in data volume Under the background of the economic era, the demand of enterprises for automatic question answering technology

Images