面向大语言模型的软件信息获取方法、电子设备及存储介质
By parsing source code using parsing tools and type resolvers that disable type binding, a structured code index is built, providing a query tool for the large language model. This solves the dependency problem of the large language model when obtaining software information and achieves fast and accurate code understanding.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN POWEROAK NEWENER CO LTD
- Filing Date
- 2026-04-30
- Publication Date
- 2026-07-17
AI Technical Summary
Existing large language models rely on a complete compilation environment and JAR packages when acquiring software information, making them difficult to apply to unknown projects, unable to obtain complete call chains, and lacking the ability to uniformly process and integrate multi-source heterogeneous information, thus limiting the depth of understanding of complex projects.
By disabling type binding through parsing tools, the source code is parsed using a type resolver to obtain basic node code, detailed information such as variables, methods, and call relationships, a structured code index is built, and it is encapsulated into a query tool for use by large language models.
It does not rely on the project's build and compilation environment, but deeply understands the code of the software project, quickly and accurately obtains software information in the source code, avoids the dependencies and computational overhead brought by vector retrieval, and is low-cost and highly interpretable.
Smart Images
Figure CN122111447B_ABST