CN-121979982-A - Quick construction method and system for modular scheme based on NLP

CN121979982ACN 121979982 ACN121979982 ACN 121979982ACN-121979982-A

Abstract

The invention relates to the technical field of digital office scheme construction, in particular to a modular scheme rapid construction method and system based on NLP, comprising the following steps: and constructing a dynamic case knowledge base and a modularized scheme component base, wherein the dynamic case knowledge base stores a history scheme with multi-dimensional classification labels, and the modularized scheme component base stores components classified according to scheme structures and corresponding adaptation rules. According to the invention, by constructing the dynamic case knowledge base and the modularized scheme component base and combining with the semantic analysis, the accurate search of cases and the intelligent adaptation and combination of components required by the NLP technology, the basic base is continuously optimized through the feedback iteration mechanism, and the problem that the modularized components, the case knowledge base and the search engine in the prior art lack intelligent linkage adaptation is effectively solved.

Inventors

PAN TAO
LI WEN
ZHOU SHUANG
Xian Mengzhi
GAO CHENGZHI
XIAO YE
WANG TING
ZENG YUE

Assignees

成都浩睿企业管理咨询有限公司

Dates

Publication Date: 20260505
Application Date: 20260113

Claims (10)

1. The modular scheme rapid construction method based on NLP is characterized by comprising the following steps: step S1, a dynamic case knowledge base and a modularized scheme component base are constructed, wherein the dynamic case knowledge base stores a history scheme with multi-dimensional classification labels, and the modularized scheme component base stores components classified according to scheme structures and corresponding adaptation rules; Step S2, receiving scheme demand description input by service personnel, analyzing semantics of the scheme demand description through a multi-mode search engine, searching by combining with classification labels in the dynamic case knowledge base, and outputting matched historical cases; Step S3, extracting characteristics of the historical cases through an intelligent linkage adaptation module, converting the characteristics into component demand parameters, carrying out semantic matching on the component demand parameters and component adaptation rules in the modularized scheme component library, and generating a preliminarily adapted component list; s4, performing scene suitability verification on the preliminarily adapted component list, removing unmatched components, and generating a final adapted component combination; And S5, collecting the use feedback of service personnel on the final adapting assembly combination, and respectively updating the dynamic case knowledge base and the modularized scheme assembly base according to the use feedback.
2. The rapid construction method of a modular solution based on NLP according to claim 1, wherein the implementation of step S1 comprises the following specific steps: step S11, adding a multi-dimensional classification label for the historical scheme, and recording each updated content and iteration time of the scheme; step S12, storing the history scheme data with the labels in a layered structure to form the dynamic case knowledge base; s13, dividing the components into a frame class, a content class and a data class according to a scheme structure, and refining successful experience in a history scheme into an adaptation rule and a use instruction of each component; and S14, storing the classified components and the adaptation rules thereof to form the modularized scheme component library.
3. The rapid construction method of the modular solution based on NLP according to claim 1, wherein the implementation of step S2 comprises the following specific steps: s21, processing the scheme requirement description based on a BERT model, and extracting core semantic information; S22, calling tag data of the dynamic case knowledge base to obtain a history scheme index of a corresponding tag; step S23, associating industry knowledge graph, and carrying out scene filtering on the search result; and S24, analyzing historical retrieval and use records of service personnel, generating personalized retrieval recommendation, and outputting a historical case with highest matching degree.
4. The rapid construction method of the modular solution based on NLP according to claim 1, wherein the implementation of step S3 comprises the following specific steps: Step S31, receiving the retrieved historical case data, and extracting industry adaptation points and scene demand details in the cases; step S32, converting the extracted features into normalized component demand parameter vectors Wherein Represent the first The quantized values of the individual demand characteristics, Is a feature dimension; Step S33, calling the component adaptation rule vector provided by the modularized scheme component library Wherein Represent the first The first component pair An adaptive weight for each demand feature; Step S34, calculating semantic matching degree score And screening components with scores higher than a preset threshold value to form the preliminarily adapted component list.
5. The rapid construction method of NLP-based modular scheme according to claim 4, wherein in step S34, the preset threshold value The calculation formula of (2) is Wherein For the mean of all component matching scores, Is the standard deviation of the two-dimensional image, For adjusting the coefficients.
6. The rapid construction method of the modular solution based on NLP according to claim 1, wherein the implementation of step S4 comprises the following specific steps: S41, constructing a scene constraint condition set based on the characteristics of the historical cases; step S42, traversing each component in the preliminarily adapted component list, and checking whether the components meet the scene constraint condition set or not; and S43, removing the components which do not meet the constraint conditions, reserving the components which meet the constraint conditions, and generating the final adaptation component combination.
7. The rapid construction method of the modular solution based on NLP according to claim 1, wherein the implementation of step S5 comprises the following specific steps: step S51, recording service personnel use selection, adjustment content and effect evaluation after scheme use of the scheme component to form feedback data; Step S52, the feedback data is classified and processed, and is transmitted to the dynamic case knowledge base, and the adaptation effect label of the historical case is updated; and step S53, the feedback data is transmitted to the modularized scheme component library, and the adaptation rule and the use instruction of the optimization component are transmitted.
8. The rapid construction method of modular solution based on NLP according to claim 1, wherein in the step S2, the collaborative filtering subunit of the multimodal search engine constructs a user portrayal vector U based on the historical behavior data of business people and calculates the similarity of the user portrayal and the case feature vector C Where W is a learnable weight matrix for generating personalized search recommendations.
9. The modular solution quick construction method based on NLP according to claim 1, wherein the final adaptation component combination is presented in a visual form for business personnel to directly multiplex or adjust to complete the solution construction.
10. An NLP-based modular solution rapid construction system for implementing the NLP-based modular solution rapid construction method according to any one of claims 1 to 9, comprising: the dynamic case knowledge base is used for storing a history scheme with multi-dimensional classification labels; the modularized scheme component library is used for storing components classified according to scheme structures and corresponding adaptation rules thereof; The multi-mode search engine is used for analyzing the semantics of the scheme requirement description and searching the matched historical cases; The intelligent linkage adaptation module is used for converting the historical case characteristics into component demand parameters, carrying out semantic matching and adaptation verification with component adaptation rules, and generating a final adaptation component combination; And the feedback iteration module is used for collecting the use feedback and driving the updating of the dynamic case knowledge base and the modularized scheme component base.

Description

Quick construction method and system for modular scheme based on NLP Technical Field The invention relates to the technical field of digital office scheme construction, in particular to a modular scheme rapid construction method and system based on NLP. Background In a digital office scene, scheme writing is an important link of business development, and an intelligent scheme writing platform helps business personnel to improve scheme construction efficiency by integrating a knowledge base, a retrieval technology and modularized components, so that the business personnel has become an industry development trend. Such platforms typically rely on structured data management, semantic understanding, and component multiplexing techniques, allowing non-professional technicians to quickly complete project construction as well, reducing reliance on professional writing capabilities. The prior art discloses a modularized development method and system based on a large language model of a CN118626057A, which proposes a mode of generating an application by inquiring the large language model and a solidification process twice, thereby reducing an application development threshold. The patent core focuses on the flow simplification of application development, realizes multiplexing through the general unique identification code recording flow, but does not optimize the scene characteristic of scheme writing, and the modularized design focuses on the solidification of the development flow rather than the deep fusion of scheme components and case knowledge. The problem with the current technology is that there is a lack of intelligent linkage adaptation mechanisms between the modular components and the case knowledge base and the search engine. In addition, when a service person uses the platform, the searched historical cases are only presented as reference content, the components in the modularized component library are required to be manually screened and spliced, the selection of the components is not automatically associated with the characteristics of the industrial scene, the client type and the like of the search cases in a semantic level, meanwhile, the multiplexing of the components is not combined with the successful experience in the cases to carry out the adaptation adjustment, so that the service person needs to spend a large amount of time to match the components and the cases, the scheme construction efficiency is reduced, the problem that the components are not matched with the required scene easily occurs, and the scheme quality is influenced. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a method and a system for quickly constructing a modularized scheme based on NLP, which solve the problems of low scheme construction efficiency and poor scene suitability caused by the lack of intelligent linkage adaptation of modularized components, a case knowledge base and a search engine in the prior art and manual screening of the components. In order to achieve the purpose, the invention is realized by the following technical scheme that the modular scheme rapid construction method based on NLP comprises the following steps: step S1, a dynamic case knowledge base and a modularized scheme component base are constructed, wherein the dynamic case knowledge base stores a history scheme with multi-dimensional classification labels, and the modularized scheme component base stores components classified according to scheme structures and corresponding adaptation rules; Step S2, receiving scheme demand description input by service personnel, analyzing semantics of the scheme demand description through a multi-mode search engine, searching by combining with classification labels in the dynamic case knowledge base, and outputting matched historical cases; Step S3, extracting characteristics of the historical cases through an intelligent linkage adaptation module, converting the characteristics into component demand parameters, carrying out semantic matching on the component demand parameters and component adaptation rules in the modularized scheme component library, and generating a preliminarily adapted component list; s4, performing scene suitability verification on the preliminarily adapted component list, removing unmatched components, and generating a final adapted component combination; And S5, collecting the use feedback of service personnel on the final adapting assembly combination, and respectively updating the dynamic case knowledge base and the modularized scheme assembly base according to the use feedback. Further, the implementation of the step S1 includes the following specific steps: step S11, adding a multi-dimensional classification label for the historical scheme, and recording each updated content and iteration time of the scheme; step S12, storing the history scheme data with the labels in a layered structure to form the dynamic case knowledge base; s13, dividing