Building material classification and extraction method and system based on multi-agent cooperation

By employing a multi-agent collaborative method for classifying and extracting building materials, utilizing semantic cleansing and multi-path fusion retrieval, and combining a large language model and a multi-source knowledge base, the problem of standardized management of building material data is solved. This achieves efficient and accurate material classification and structured extraction, supporting intelligent decision-making and continuous optimization.

CN121786201BActive Publication Date: 2026-06-19云筑信息科技(成都)有限公司
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
云筑信息科技(成都)有限公司
Filing Date
2026-03-04
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

In the construction industry, the standardized management of material data faces bottlenecks such as low standardization, high data processing costs, limited effectiveness of traditional methods, and a lack of intelligent application development. Existing technologies struggle to handle fuzzy inputs and lack self-optimization capabilities.

Method used

A multi-agent collaborative method for classifying and extracting building materials is adopted. Through semantic cleansing, multi-path fusion retrieval, adversarial intent analysis, and iterative decision-making, combined with a large language model and a multi-source knowledge base, the method achieves accurate classification and structured extraction of building materials.

Benefits of technology

It achieves efficient and accurate classification of fuzzy inputs, reduces manual intervention, improves the efficiency and accuracy of data processing, supports intelligent decision-making and continuous optimization, and ensures that the output meets industry standards.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121786201B_ABST
    Figure CN121786201B_ABST
Patent Text Reader

Abstract

This invention discloses a multi-agent collaborative method and system for classifying and extracting building materials, addressing the standardization challenges caused by multiple codes for a single item and multiple items for a single code in building material data. The method performs semantic cleansing on building material description text, filtering out non-material interference information; it obtains a preliminary candidate material set from vector and relational databases through multi-path fusion retrieval; based on this set, it extracts material definitions and sample knowledge from the relational database and obtains historical correction knowledge by matching with a Badcase case knowledge base; it acquires external supplementary knowledge through network retrieval with adversarial intent analysis and multi-stage reflective verification mechanisms; after fusing the above multi-source knowledge, it determines the final material classification through confidence-driven iterative decision-making, and extracts corresponding attribute rules based on the classification results, completing structured attribute extraction and output. This invention improves the accuracy and automation of material classification.
Need to check novelty before this filing date? Find Prior Art