Historical error correction method and system for preventing overexsplit of scholar paper library

Abstract

The invention provides a historical error correction method and system for preventing overexsplit of a scholar paper library. The method comprises the following steps: reconstructing a scholar name; directly matching the target scholar paper library with the paper cluster to be distributed according to the information capable of uniquely determining the author; for the paper cluster which is not successfully matched, identifying author related information of the paper and entity information in the abstract through a BERT-Bi-LSTM-CRF model; respectively calculating the matching degree of the information of the institutions to which authors belong and the periodical information included in the papers to be matched; respectively calculating similarity characteristics of each candidate alignment paper cluster and the target scholar paper library, and judging whether each candidate alignment paper cluster is aligned with the target scholar paper library or not; and combining the candidate aligned paper clusters which are judged to be aligned by the integrated learning model, and carrying out manual annotation on the paper clusters which are not aligned. According to the method, the over-splitting error generated in the disambiguation process can be solved, and the over-splitting error correction speed, accuracy and recall rate are improved.

Description

technical field [0001] The present application relates to the technical field of information processing, and in particular, to a method and system for correcting historical errors for preventing over-splitting of a scholar's dissertation database. Background technique [0002] At present, the over-splitting of scholar papers is a common problem encountered in the operation of scholar papers libraries. This problem originates from the scenario where the cold-start disambiguation of the paper and the incremental disambiguation algorithm of the paper are designed to eliminate the ambiguity of different authors with the same name, and the algorithm is over-split in the process of running. The over-splitting scenario generates a large number of fragment clusters, that is, multiple scholar libraries of the same scholar. The fragment clusters will continue to increase with the increase of systematic papers, which will lead to a great reduction in the recall and accuracy of the incr...

AI Technical Summary

Problems solved by technology

For example, in the incremental disambiguation algorithm, the INC algorithm directly merges all the over-split fragments by checking whether the similarity between the over-split fragments is greater than the specified threshold, but this method is prone to erroneous merging

The MINDi algorithm specifically studies fragment merging. When multiple candidate authors meet the specified conditions, it calculates the distance between the candidate authors and merges the two closest candidate authors. However, the accuracy of this method and Insufficient recall

[0004] Therefore, the above-mentioned solution to prevent over-splitting by improving the name disambiguation algorithm, under the principle of purity first, multiple papers by the same author will inevitably be divided into multiple clusters, resulting in over-splitting errors, over-splitting Less accurate error correction

Images