Academic institution name entity alignment method based on text features

Abstract

The invention discloses an academic institution name entity alignment method based on text features. The academic institution name entity alignment method comprises the following five steps: converting English abbreviations into English full names; correcting the error correspondence of the English abbreviation and the English full name; translating to complement English full names and Chinese names; correcting wrong Chinese names; and performing academic institution combination based on the text features. According to the method, entity alignment is carried out by using academic institution data extracted from Chinese and English text data; wherein each part of institution data contains English abbreviation, English full name, Chinese name, geographic position, content loss and a small amount of errors, and finally obtaining a plurality of different names corresponding to the same institution by complementing the missing data, correcting the error data and combining the data of the same institution. According to the method, the organization name text features and the geographic position information are combined for academic organization name entity alignment, the pre-marked organization name corresponding relation and the context semantic information of the name are not needed, and a good entity alignment effect is obtained with low complexity.

Description

technical field [0001] The technical fields involved in the present invention include entity alignment, entity disambiguation, construction of knowledge graph, data preprocessing technology and search algorithm, especially relates to the method of entity alignment for academic institution names and construction of academic knowledge graph, and relates to a text-based The academic institution name entity alignment method for the trait. Background technique [0002] In recent years, with the development of computers and networks and the accumulation of data, there have been more and more electronic data assisting computers to complete more tasks. In order to understand the relationship between common items in life, so that the computer can learn more knowledge, a knowledge graph can be constructed for the entities in life. Each item corresponds to an entity point on the graph, and the relationship between items corresponds to the graph. Edges connecting entities, this method ...

AI Technical Summary

Problems solved by technology

[0006] In view of the shortcomings and existing difficulties and challenges of the prior art, the purpose of the present invention is to provide a method for entity alignment of academic institution names based on text features for the characteristics of academic institution names. The relationship between full name, Chinese and English, geographic location and institution solves the problems that the existing technical solutions cannot align academic institution entities without context, the algorithm implementation complexity is high, and there is not enough labeled data to train the model. Only use a small number of known correspondences between abbreviations and full names, and place name data to perform alignment tasks on academic institution entities, without the need for relevant papers and author information, and achieve better results

Images