Sentence similarity calculation method, device and storage medium

Abstract

The invention discloses a calculation method of sentence similarity, which relates to the field of sentence analysis. The method comprises the following steps: acquiring a sentence pair to be calculated; calculating a semantic contribution degree of each word in the pair of sentences to be calculated to obtain a weight vector of the user statement and a weight vector of the standard statement; comparing each word in each sentence of the sentence pair to be calculated with the word of another sentence to obtain inter-word similarity, and calculating a similarity vector of the user sentence anda similarity vector of the standard sentence according to the inter-word similarity; calculating a sentence similarity between the user statement and the standard statement according to a weight vector and a similarity vector of the user statement and a weight vector and a similarity vector of the standard statement. The embodiment of the invention also provides a sentence similarity device and astorage medium, which can effectively improve the calculation efficiency of sentence similarity calculation and the credibility of the calculation result.

Description

technical field [0001] The invention relates to the field of sentence analysis, in particular to a calculation method, device and storage medium for sentence similarity. Background technique [0002] In the development of chat robots, obtaining the same or similar sentences according to the sentences entered by the user is one of the core processes, and the realization of this core process depends on the calculation method of sentence similarity. [0003] In the prior art, there are generally three different implementation methods for calculating sentence similarity, including a string-based method, a bag-of-words model-based method, and a knowledge-based method. Among them, the character string is used as an independent knowledge unit, and the sentence similarity is calculated according to the similarity of the characters; the bag-of-words model is used to learn through the neural network, and the text similarity is calculated according to the information obtained from the ...

AI Technical Summary

Problems solved by technology

[0005] 1. The string-based method lacks consideration of the importance of each component in the sentence, resulting in low credibility for the calculation results of sentence similarity;

[0006] 2. The method based on the bag-of-words model ignores the word order relationship in the sentence, and as the size of the sentence increases, it is easy to cause the disaster of dimensionality and the operation efficiency is low;

[0007] 3. The knowledge-based method relies on the construction of the knowledge base. Since the information in the knowledge base needs to be guaranteed to be effective and reliable, the update efficiency of the knowledge base is slow, resulting in low credibility for the calculation results of the similarity of sentences containing emerging vocabulary.

Images