Method, system, and equipment for disambiguating characters with duplicate names that improve the efficiency of article-by-article filing

Abstract

The invention belongs to the technical field of disambiguation of the same name, and discloses a method, system and device for disambiguating characters with the same name that improves the efficiency of file-by-article disambiguation, divides name groups according to names; obtains text sets corresponding to name groups; splits texts based on rules Set, divide text group; Calculate the similarity between the text groups; The homonym disambiguation system of the efficiency of described article-by-article filing includes: name group division module, text collection acquisition module, large group splitting module, storage module, training set Building blocks, model training and optimization modules, sub-model discriminator building blocks, prediction modules, ID selector building blocks, ID selector training and optimization modules, classification prediction modules. The present invention adopts the method of rule + model to improve the calculation efficiency of the whole model; reduces the number of candidate matches in units of groups to improve calculation efficiency; uses multi-model fusion technology, and each sub-model uses different features to improve the robustness and prediction of the model precision.

Description

technical field [0001] The invention belongs to the technical field of character duplicate name disambiguation, and in particular relates to a method, system and device for character duplicate name disambiguation that improve the efficiency of filing articles one by one. Background technique [0002] At present, the phenomenon of people with the same name is common in all countries in the world, so it is difficult to carry out big data analysis based on personnel. Even the simplest work of counting the number of patent inventors in a certain city cannot be easily carried out. Based on the cluster analysis of personnel, it is more difficult to implement the analysis of personnel flow, so the work of disambiguation of duplicate names in the literature database is particularly important. [0003] Currently commonly used technical means such as: rule-based disambiguation of characters with duplicate names, dynamic programming-based disambiguation of characters with duplicate nam...

AI Technical Summary

Problems solved by technology

[0004] Through the above analysis, the existing problems and defects of the existing technology are: the existing methods for disambiguation of characters with the same name, there are problems of low feature collection efficiency and low calculation efficiency in the process of incremental file-by-article archiving

Images