Statement similarity calculation method and device, and computer equipment

Abstract

The invention provides a statement similarity calculation method and a device, and computer equipment. The method comprises the steps of inputting at least two statements of which the similarity is tobe calculated into a semantic vector calculation model group, the semantic vector calculation model group comprising at least two semantic vector calculation models, the at least two semantic vectorcalculation models sharing one set of model parameters, and different statements being correspondingly input into different semantic vector calculation models; utilizing any semantic vector calculation model to respectively encode the received sentences into semantic vectors of preset dimensions and inputting the semantic vectors into a similarity calculation model; and converting the semantic vectors of any two preset dimensions into a binary classification vector by utilizing a similarity calculation model, and obtaining the similarity of any two statements according to the binary classification vector. The semantic vector calculation model group is a multi-input model, different statements can be correspondingly input into different semantic vector calculation models, the statement splicing step does not need to be executed, and the statement similarity calculation efficiency is improved.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, and more specifically, relates to a sentence similarity calculation method, device and computer equipment. Background technique [0002] In the technical field of natural language processing, the application of calculating the similarity between two sentences is very extensive. In the currently used model-based sentence similarity calculation method, the models used are all single-input models. Before the statement similarity calculation, it is necessary to splice the sentences to be calculated to obtain a spliced ​​sentence. Sentences are input into the model for semantic analysis and then calculation of similarity. Therefore, before calculating the similarity between a certain target sentence and multiple candidate sentences, it is necessary to splice the target sentence with each candidate sentence once, which takes a long time and results in low efficiency of sent...

AI Technical Summary

Problems solved by technology

Therefore, before calculating the similarity between a target sentence and multiple candidate sentences, it is necessary to splicing the target sentence with each candidate sentence once, which takes a long time and results in low efficiency of sentence similarity calculation.

Images