Government service guidance data processing method and system based on domain feature construction

By constructing text feature representations containing domain features in the government service guide data processing, and combining attention mechanism and LSTM model, the problem of insufficient quality and diversity of data sets is solved, and more accurate text theme recognition and multi-domain text classification are achieved.

CN119149748BActive Publication Date: 2025-03-28山东省大数据中心
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202411594963.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-11-11
Publication Date
2025-03-28
Estimated Expiration
2044-11-11

AI Technical Summary

Technical Problem

In the existing government service guide data processing methods, the data sets are not of high quality and lack of diversity, which leads to the model performing well in specific scenarios but not in other scenarios, and the complex, multi-step or cross-domain problems cannot be correctly identified and handled.

Method used

The data processing method based on domain feature construction is adopted. By obtaining text data from multiple domains, extracting domain feature words, calculating the average distance between characters and domain feature words, constructing text feature representations containing domain features, and inputting them into the attention mechanism to enhance the attention of feature representations, and finally classifying predictions are made by fusing text-dependent features and domain feature features.

Benefits of technology

The model's understanding of text structure and meaning is improved, the attention to long-term dependence and key words is enhanced, and more accurate text theme recognition and multi-field text classification are achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119149748B_ABST
    Figure CN119149748B_ABST
Patent Text Reader

Abstract

The present invention provides a government service guide data processing method and system based on domain feature construction, which relates to the field of data processing technology, including obtaining text data in multiple fields and preprocessing; extracting domain feature words in the preprocessed text data, obtaining the interaction between the domain feature words and the text data, and calculating the average distance between each character in the text data and each character of the domain feature words; performing function mapping on the calculated average distance result to construct a text feature representation containing domain features; inputting the text feature representation into an attention mechanism to enhance the attention of the text feature representation containing domain features and obtain the final domain feature features; extracting text dependency features of text data in multiple fields, fusing the text dependency features with the domain feature features, and using the fused features to complete the prediction output of the text category, obtaining multi-field text classification results, and realizing fast and automatic classification of government service guide data sets.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Keyword fuzzy matching method and device

    CN105550298A

  • Cross-domain emotion classification system based on attention mechanism fusion

    CN110874411A