An Internationalized Time-Aware Recall Method Based on Text Temporal Semantic Parsing

By employing an adaptive temporal semantic extraction strategy, combined with rules, domain-specific models, and large language models, the problem of memory systems being unable to understand natural language temporal expressions has been solved. This enables support for multilingual and cross-time zone conversions, enhances the ability to express time, resolves time perception issues, supports multilingual and cross-time zone conversions, and is deeply integrated with vector databases to improve retrieval accuracy.

CN122087092APending Publication Date: 2026-05-26LIAOCHENG DONGLI ZHIYUAN DIGITAL TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610325711.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-03-17
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing memory systems cannot understand time expressions in natural language. Time extraction tools are disconnected from memory systems, lack an adaptive time semantic extraction architecture, and have insufficient support for internationalization and time zones.

Method used

It adopts an adaptive temporal semantic extraction strategy, combining rules, domain-specific models and large language models to achieve intelligent extraction of natural language temporal expressions, supports multilingual and cross-time zone conversion, and is deeply integrated with vector databases.

Benefits of technology

It achieves end-to-end time awareness capabilities, supports accurate extraction and retrieval of various time expression formats, reduces computational costs, provides internationalization and time zone support, and improves retrieval accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122087092A_ABST
    Figure CN122087092A_ABST
Patent Text Reader

Abstract

This invention discloses an internationalized time-aware recall method based on text time semantic parsing, comprising the following steps: S1. Receiving user input; S2. Complexity classification; S3. Adaptive time semantic extraction: an adaptive time semantic extraction strategy extracts time expressions from natural language text; S4. Time normalization: converting the time expressions into standardized time interval representations; S5. Time zone processing: converting the time interval representation into a timestamp of a standard time zone according to the user's time zone parameters; S6. Time-constrained memory retrieval: performing time-constrained memory retrieval in a vector database based on the time interval representation; S7. Outputting memory results: returning the memory results to the user. This invention supports multiple expression forms such as absolute time, relative time, fuzzy time, and time range, supports multilingual extensions including Chinese, English, and Japanese, supports global time zone conversion, and achieves end-to-end time-aware memory recall, significantly improving user query efficiency and user experience.
Need to check novelty before this filing date? Find Prior Art