Method and device for matching time series record link data based on attribute association

By building a time correlation model and dynamically adjusting similarity weights, the problem of record link inaccuracy caused by changes in entity attributes is solved, and higher recall and comprehensive performance are achieved, which is suitable for entity matching of multiple data sources.

CN116680325BActive Publication Date: 2025-09-05HANGZHOU DIANZI UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310747487.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-06-25
Publication Date
2025-09-05
Estimated Expiration
2043-06-25

AI Technical Summary

Technical Problem

In the prior art, when processing the change of entity attributes over time, the recall and comprehensive performance of the recording link are insufficient, making it difficult to accurately match the changed entity data.

Method used

Through the timing record linking method based on attribute correlation, the XGBoost algorithm is used to build a time correlation model, explore the correlation between attributes, predict the probability of attribute change over time, and dynamically adjust the similarity weight to achieve matching of entity attributes.

Benefits of technology

It significantly improves the accuracy and recall of record linkage, can automatically mine attribute associations, is applicable to large data sets, and is universal and scalable.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116680325B_ABST
    Figure CN116680325B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and device for matching time-series record link data based on attribute relevance. The method first performs data preprocessing to clean the data source data of each link and the data used to train the time association model. Secondly, the time association model is trained to train a model that predicts the probability of each link attribute changing over time. Finally, data linking is performed to realize entity recognition from multiple data source data. The device includes a data preprocessing module, a training module and a linking module. The present invention can accurately predict the probability of entity attributes changing over time to adjust the similarity between attributes, and significantly improve the accuracy of link time records, with universality and scalability.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Time sequence knowledge graph construction method and system

    CN115858822A

  • Method for real-time cohort creation based on entity attributes derived from partially observable location data

    US20210256037A1