CN-122022713-A - Large language system for project sea area use demonstration report
Abstract
The invention discloses a large language system for a project sea and sea area use demonstration report, which comprises a data layer, an application layer, a policy question and answer service unit, a compliance checking service unit, a substantive checking service unit and a technical checking opinion generation and optimization unit, wherein the data layer is used for storing training data and checking results, the application layer comprises a policy question and answer service unit, the policy question and answer service unit is used for providing accurate policy question and answer service for training authority files such as domestic sea-related law regulations, policy specifications, technical standards and sea area planning, the compliance checking service unit is used for generating detailed compliance reports and accurately positioning problems, the substantive checking service unit is used for providing comprehensive substantive services, and the technical checking opinion generation and optimization unit is used for generating detailed and clear technical checking opinion reports. According to the invention, the accuracy and efficiency of the examination work are obviously improved, and a powerful support is provided for understanding laws, regulations and policies in the related field.
Inventors
- ZHANG WENJIA
Assignees
- 同济大学
Dates
- Publication Date
- 20260512
- Application Date
- 20251224
Claims (7)
- 1. A marine domain usage demonstration report large language system for a project, comprising: The data layer is used for storing training data and examination results; The application layer comprises a policy question-answer service unit, a compliance check service unit, a substantive examination service unit and a technical examination opinion generation and optimization unit; The policy question-answering service unit is used for providing accurate policy question-answering service for the trained authoritative files such as national sea-related law regulations, policy specifications, technical standards, sea area planning and the like; the compliance checking service unit is used for generating a detailed compliance report and accurately positioning the problem; The substantive censoring unit is used for providing comprehensive substantive censoring services by using core censoring elements in the argumentation report and combining the policy regulations and technical standards after carding; the technical review opinion generation and optimization unit is used for generating detailed and clear technical review opinion reports based on the analysis results of compliance checking and review elements.
- 2. A sea area usage demonstration report large language system for projects according to claim 1 wherein said data layer comprises a knowledge base comprising a regulatory knowledge base and a report knowledge base.
- 3. The large language system of project sea area usage demonstration report of claim 2, wherein the data layer collects national sea law regulations, policy specifications, technical standards and sea area planning files, performs text cleaning and structured labeling, and constructs a standardized training data set.
- 4. A sea area usage demonstration report large language system for projects according to claim 2 wherein the system performs corrective learning of error localization and cause interpretation by reporting typical error cases in historical reports in the knowledge base and collates the cases and their corresponding cause descriptions to provide systematic case reference resources for users.
- 5. The large language system of project sea area usage demonstration report of claim 1, wherein the compliance checking service unit comprises a development file analysis module which supports uploading and text extraction of PDF format demonstration report, carries out compliance scanning on the extracted text and chart information through regular expression and semantic analysis algorithm, realizes automatic detection through sequence comparison algorithm aiming at various problems of disordered chart sequence numbers and inconsistent formats, generates a detection report, and defines the position and the correction proposal of the problem.
- 6. The large language system of the project sea area use demonstration report of claim 1, wherein the substantial examination and maintenance unit carries out examination and maintenance element combing, the examination and maintenance element combing is used for extracting core examination and maintenance elements of the demonstration report according to sea area use demonstration technical guide and industry examination standards, and the core examination and maintenance element combing comprises hundred examination and maintenance elements such as demonstration level judgment, sea necessity, site selection rationality, sea surface integration, resource ecological influence analysis and the like, and a three-dimensional examination system of examination elements, policy basis and judgment logic is constructed.
- 7. An electronic device comprising at least one processor and a memory communicatively coupled to the at least one processor, wherein the memory stores a computer program executable by the at least one processor to enable the at least one processor to execute the project sea area usage demonstration report large language system of any one of claims 1-5.
Description
Large language system for project sea area use demonstration report Technical Field The invention relates to the technical field of sea area use demonstration examination, in particular to a large language system for a project sea area use demonstration report. Background The use demonstration examination of the sea area is a key link for guaranteeing the reasonable utilization of sea area resources and ecological protection. Currently, this auditing work is largely dependent on manual completion. The auditing mechanism needs to process a large number of demonstration reports with various forms and complex contents, and the auditing key points of the auditing mechanism are as many as hundreds of items. The method needs to coordinate a plurality of departments to conduct item-by-item manual comparison and auditing, and organize expert demonstration, so that the auditing period is long, the labor cost is high, and the normalization and consistency of the auditing result are difficult to guarantee. With the development of artificial intelligence technology, an intelligent solution capable of adapting to a complex sea management scenario is needed in the industry to convert a traditional manual-dependent inspection process into an efficient and accurate automatic or semi-automatic process so as to cope with the complexity and diversity challenges in the current inspection work. Disclosure of Invention Aiming at the defects existing in the prior art, the invention aims to provide a large language system for project sea and sea area use demonstration report, which not only remarkably improves the accuracy and efficiency of examination work, but also provides powerful support for law and regulation and policy understanding in the related fields. To achieve the above objects and other advantages and in accordance with the purpose of the present invention, there is provided a sea area usage demonstration report large language system for projects, comprising: The policy question-answering service unit develops a powerful question-answering engine by using a large language model technology of a demonstration report for the sea and sea area, and can provide accurate policy question-answering service based on authoritative files such as domestic sea-related law regulations, policy specifications, technical standards, sea area planning and the like for training. Through this function, the user can quickly query and understand relevant policy regulations, and reliable information support and policy interpretation are provided for the compiling and examining work of the sea area using the demonstration report. The question-answering service has high accuracy and timeliness, helps users to greatly improve the working efficiency, and reduces the dependence on manual searching and interpretation. The intelligent compliance checking service unit is used for developing a compliance checking function by using a large language model technology of a demonstration report in the sea and sea area for projects, and can generate a detailed compliance report and accurately position the problem. The basic checking function of the model covers wrongly written words, format specifications, name use accuracy and the like. Through the function, a user can quickly identify and solve the compliance problem existing in the report, a detailed compliance report is produced, and the efficiency and the accuracy of the examination work are greatly improved. And the technical examination opinion generation and optimization unit is used for generating a detailed and clear technical examination opinion report by using a large language model technology of the demonstration report for the sea and sea area of the project based on the analysis results of compliance examination and examination elements. The model not only can accurately position the compliance problem and the examination requirement problem, but also can analyze the cause of the problem in detail by combining with specific policy and regulation and examination standards, thereby providing practical revision advice for users. On the basis of generating the examination comments, the model also provides optimization suggestions, so that the user is helped to perfect the examination report better. And (3) case learning and knowledge base construction, namely continuously training the large language model technology of the project using the demonstration report in the sea and sea area, and continuously improving the suitability and accuracy of the model to examination work. The model can correct and learn error positioning and reason interpretation based on typical error cases in the history report, and collate the cases and corresponding reason descriptions, thereby providing systematic case reference resources for users. Meanwhile, a knowledge base is constructed based on report contents generated by the model, and comprises a checked report knowledge base, a national law and law regulation re