Intelligent data feature derivation method and system based on large model and MCP protocol

Abstract

The invention relates to the technical field of internet and big data, in particular to an intelligent data feature derivation method and system based on a large model and an MCP protocol, and the method comprises the steps: receiving a feature engineering task inputted by a user, the task comprising a feature target, an industry background and a constraint condition; the task is analyzed, a structured feature derivation strategy is generated based on a special cue word template for feature engineering, and the cue word template at least comprises a business target field, a data type field and a constraint condition field; based on the feature derivation strategy, a feature tool library is scheduled through an MCP protocol to execute feature processing operation, and the feature processing operation comprises basic data processing, data derivation, data splicing and data verification; performing multi-dimensional quality evaluation on the processed feature set, and generating a feature generation log; and outputting the evaluated feature set. According to the method, the performance and the landing value of a machine learning model constructed based on the features are improved.

Description

Technical Field

[0001] This invention relates to the fields of Internet and big data technology, and in particular to an intelligent data feature derivation method and system based on a large model and the MCP protocol. Background Technology

[0002] Currently, feature engineering, as a core component of the machine learning process, aims to extract and construct high-value features from raw data, directly impacting model performance and the quality of business decisions. Existing technologies primarily rely on four paths: manual design, automated tools, traditional model learning, and brute-force combination. However, these methods suffer from a series of fundamental common flaws: the quality and efficiency of feature engineering are highly dependent on the individual knowledge reserves of technical personnel (including business domain, statistical methods, and algorithm principles), leading to significant fluctuations in feature quality and an inability to guarantee consistency; automated tools struggle to understand and integrate complex business logic and constraints, often generating features with weak business relevance and poor interpretability; simultaneously, brute-force combination methods are prone to the curse of dimensionality, generating a large number of features without business meaning, resulting in a severe waste of computing and storage resources. Therefore, the industry urgently needs a new feature engineering solution that can reduce reliance on individual knowledge, intelligently integrate business logic, and ensure feature interpretability and resource efficiency. Summary of the Invention

[0003] This invention provides an intelligent data feature derivation method and system based on a large model and the MCP protocol, which solves the problems of existing technologies that rely too much on human experience, are difficult to integrate into business logic, and have waste of resources and poor interpretability.

[0004] The objective of this invention can be achieved through the following technical solutions: The first aspect of this invention is to provide an intelligent data feature derivation method based on a large model and the MCP protocol, comprising: Receive a feature engineering task input by the user, the task including feature objectives, industry background and constraints; The task is parsed, and a structured feature derivation strategy is generated based on a feature engineering-specific prompt word template. The prompt word template includes at least a business target field, a data type field, and a constraint field. Based on the aforementioned feature derivation strategy, feature processing operations are executed by scheduling the feature tool library through the MCP protocol. The feature processing operations include basic data processing, data derivation, data concatenation, and data verification. Perform multidimensional quality assessment on the processed feature set and generate a feature generation log; Output the feature set that passes the evaluation.

[0005] Furthermore, the structured feature derivation strategy based on feature engineering-specific prompt word templates includes the following specific steps: The parsed task elements are injected into the parameterized prompt word template; The large language model is used to reason about the constructed prompt words and generate machine-readable feature derivation strategies. The feature derivation strategy includes tool call sequences, parameter mapping relationships, and business constraint rules.

[0006] Furthermore, the communication process of scheduling the feature tool library via the MCP protocol includes: Initiate a tool availability query and metadata retrieval to the tool registry center; Request execution permission from the state manager and make load balancing decisions; The call request containing the business constraint fields is routed to the corresponding feature tool instance; Asynchronously receive processing results and update tool status.

[0007] Furthermore, the processed feature set undergoes a multi-dimensional quality assessment, wherein the assessment dimensions include: Technical dimensions: Evaluate the importance, stability, and collinearity among features; Business dimension: Assess the interpretability, business compliance, and business value of the features; Engineering dimension: Evaluate the computational efficiency, storage overhead, and incremental update support capabilities of the features.

[0008] Furthermore, the method also includes a verification feedback step: if the multidimensional quality assessment fails, the feature derivation strategy regeneration mechanism is automatically triggered, the tool call sequence or parameters are adjusted, and the feature processing operation is re-executed until the output features meet all constraints.

[0009] A second aspect of this invention is to provide an intelligent data feature derivation system based on a large model and the MCP protocol, comprising: The intelligent agent control module is used to receive and parse user tasks, generate derivative strategies, and verify the results. The large model cognitive module is connected to the intelligent agent control module and is used for policy reasoning based on prompt word templates; The MCP protocol interface module is connected to the intelligent agent control module to realize standardized communication and tool scheduling; The MCP Server feature tool library, connected to the MCP protocol interface module, is used to perform feature processing operations. It includes a basic data processing module, a data derivation module, a data splicing module, and a data verification module.

[0010] Furthermore, the intelligent agent control module includes: The task parsing unit is used to parse the natural language task input by the user into a structured task description; The strategy generation unit is used to construct prompt words and call the large model cognitive module; The tool scheduling unit is used to convert strategies into a sequence of tool invocation instructions; The results verification unit is used to perform three-dimensional quality assessment and closed-loop feedback of technology, business, and engineering.

[0011] Furthermore, the MCP protocol interface module includes a tool registry center, a discovery service, and a state manager to enable dynamic registration, semantic discovery, load balancing, and lifecycle management of tools.

[0012] Furthermore, the MCP Server feature toolkit includes: The basic data processing module is used to perform data cleaning and standardization; The data derivation module is used to generate statistical features, business indicators, and time series features based on business constraints. The data splicing module is used to realize intelligent feature combination and control; The data verification module is used to perform feature generation log recording and multidimensional quality assessment.

[0013] A third aspect of the present invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the intelligent data feature derivation method based on a large model and the MCP protocol.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: It receives a feature engineering task input by a user, the task including feature objectives, industry background, and constraints; it parses the task and generates a structured feature derivation strategy based on a feature engineering-specific prompt word template, the prompt word template including at least a business objective field, a data type field, and a constraint field; through carefully designed prompt word engineering, it formalizes business knowledge, guiding a large model to generate feature strategies that conform to the business scenario, thus solving the core pain point of missing business logic in automated tools; based on the feature derivation strategy, it schedules a feature tool library to execute feature processing operations via the MCP protocol, the feature processing operations including basic data processing... The system processes, derivates, stitches, and validates data; through intelligent feature combination strategies and control mechanisms, it avoids the dimensional disaster and resource waste caused by brute-force combinations, significantly improving feature generation efficiency; it performs multi-dimensional quality assessment on the processed feature sets and generates feature generation logs; through a three-dimensional verification system and complete feature generation logs, it ensures the technical reliability, business compliance, and engineering feasibility of each feature, meeting regulatory audit requirements; it outputs a feature set that passes the evaluation; it solves the problems of existing technologies relying too much on human experience, being difficult to integrate into business logic, and having wasteful resources and poor interpretability, improving the performance and practical value of machine learning models built based on these features. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This invention provides a flowchart illustrating the steps of an intelligent data feature derivation method based on a large model and the MCP protocol. Figure 2 This invention provides a schematic diagram of the module flow of an intelligent data feature derivation system based on a large model and the MCP protocol. Figure 3 A schematic diagram of the process for deriving intelligent data features. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] To address the problems existing in the background technology, this paper studies and designs an intelligent data feature derivation method and system based on a large model and the MCP protocol, which has important practical significance.

[0020] like Figure 1 As shown, the first aspect of this invention is to provide an intelligent data feature derivation method based on a large model and the MCP protocol, comprising the following steps: Step S1: Receive the feature engineering task input by the user, the task including feature target, industry background and constraints.

[0021] It's important to note that the purpose of this operation is to transform vague, unstructured business requirements into standardized task descriptions that are understandable and processable by machines. Its core function is to lay the foundation for subsequent automated processing: by forcibly defining the three key dimensions of "goal, context, and constraints," it ensures that the system can accurately understand the user's true intent, transforming abstract business problems (such as "predicting user churn") into controlled technical problems. This guides the large model to generate executable strategies that conform to the business scenario and meet resource constraints, fundamentally avoiding the errors in feature derivation or resource waste caused by misunderstandings of requirements in traditional methods.

[0022] Step S2: Parse the task and generate a structured feature derivation strategy based on the feature engineering-specific prompt word template. The prompt word template includes at least a business target field, a data type field, and a constraint field.

[0023] The structured feature derivation strategy based on feature engineering-specific prompt word templates includes the following steps: The parsed task elements are injected into the parameterized prompt word template; The large language model is used to reason about the constructed prompt words and generate machine-readable feature derivation strategies. The feature derivation strategy includes tool call sequences, parameter mapping relationships, and business constraint rules.

[0024] It's important to note that this operation aims to act as the system's "intelligent brain," transforming the standardized task description parsed in the previous step into a specific, executable "feature engineering formula." Its core function lies in leveraging the generalized knowledge and reasoning capabilities of the large language model to precisely map business requirements into technical actions. Through carefully designed prompt templates (containing three main fields: business objective, data type, and constraints), the system guides the large model to think creatively within clear boundaries, outputting a structured strategy. This strategy not only includes the sequence of tool modules to be invoked but also specifies the concrete parameters of each tool and the business rules that must be followed. This achieves a reliable conversion of business logic into automated technical execution, ensuring the high relevance and value of the derived features.

[0025] Step S3: Based on the feature derivation strategy, the feature tool library is scheduled to perform feature processing operations through the MCP protocol. The feature processing operations include basic data processing, data derivation, data concatenation, and data verification.

[0026] The basic data processing includes intelligent missing value imputation, outlier detection and correction, and data type standardization. Missing value imputation based on distribution characteristics and outlier handling combining statistical thresholds and business rules ensure input data quality and lay the foundation for subsequent feature derivation.

[0027] Among these, data derivation enables the derivation of statistical features (mean / variance / quantiles, etc.), business indicators (such as RFM model features), and time-series features (rolling statistics / trend features). Meaningless derivation is avoided by embedding business constraints, thereby enhancing the business value of features.

[0028] The data stitching process enables intelligent feature combination control, including feature cross-referencing (selecting cross-referencing objects based on business relevance), horizontal stitching (multi-source data association), and vertical stitching (time series expansion). This combination strategy prevents the curse of dimensionality and ensures the business relevance of feature combinations.

[0029] Among them, data verification involves a three-dimensional feature quality assessment, which ensures feature interpretability and engineering feasibility through a three-dimensional verification system; the three dimensions include technical, business, and engineering dimensions.

[0030] The method also includes a verification feedback step: if the multidimensional quality assessment fails, the feature derivation strategy regeneration mechanism is automatically triggered, the tool call sequence or parameters are adjusted and the feature processing operation is re-executed until the output features meet all constraints.

[0031] The MCP protocol in the "Scheduling Feature Tool Library via MCP (ModelContext Protocol)" step adopts a communication specification based on JSON-RPC 2.0 (JavaScript Object Notation Remote Procedure Call 2.0) extensions, and its communication process includes: Initiate a tool availability query and metadata retrieval to the tool registry center; Request execution permission from the state manager and make load balancing decisions; The call request containing the business constraint fields is routed to the corresponding feature tool instance; Asynchronously receive processing results and update tool status.

[0032] It's important to note that the purpose of this operation is to translate intelligent decisions (strategies) into concrete computational actions in the physical world, acting as the "execution arm" of the entire system. Its core function is to decouple and efficiently collaborate between intelligent agents and the tool library through standardized protocols: the MCP protocol acts as a "universal translator" and "traffic commander," ensuring that abstract instructions in the strategy are accurately routed to the corresponding modules (basic processing, derivation, concatenation, and verification) in the feature tool library, along with business constraints. This drives these distributed tools to perform high-quality feature processing and quality inspection according to predetermined processes, ultimately achieving automated, pipeline-style feature production and guaranteeing reliable strategy implementation and system scalability.

[0033] Step S4: Perform multidimensional quality assessment on the processed feature set and generate a feature generation log; output the feature set that passes the assessment.

[0034] The evaluation dimensions include: Technical dimensions: Evaluate the importance, stability, and collinearity among features; Business dimension: Assess the interpretability, business compliance, and business value of the features; Engineering dimension: Evaluate the computational efficiency, storage overhead, and incremental update support capabilities of the features.

[0035] It's important to note that the purpose of this operation is to build a closed-loop quality control and reliability assurance system, ensuring that the final output feature set is not only effective but also reliable. Its core function is to provide quantifiable quality assurance and decision-making basis for the results of automated feature engineering through quantitative evaluation across three dimensions: technology, business, and engineering. This multi-dimensional quality assessment acts like a rigorous "quality inspection station," comprehensively screening from feature importance and business interpretability to computational efficiency, filtering out invalid, unstable, or non-compliant features. Simultaneously, detailed feature generation logs record the "past and present" of each feature, forming a complete traceable chain. This not only greatly enhances the credibility and auditability of the features, meeting compliance and regulatory requirements, but also provides crucial data support for subsequent model iteration and problem investigation. Ultimately, what is delivered to the user is a high-quality, highly available, and completely transparent data asset. The intelligent data feature derivation process is illustrated in the diagram below. Figure 3 As shown.

[0036] This concludes the embodiment.

[0037] like Figure 2 As shown, a second aspect of the present invention is to provide an intelligent data feature derivation system based on a large model and the MCP protocol, comprising: The intelligent agent control module 101 is used to receive and parse user tasks, generate derivative strategies, and verify results. The large model cognitive module 102 is connected to the intelligent agent control module and is used for strategy reasoning based on prompt word templates; The MCP protocol interface module 103 is connected to the intelligent agent control module and is used to realize standardized communication and tool scheduling. MCP Server Feature Tool Library 104 is connected to the MCP protocol interface module and is used to perform feature processing operations. It includes a basic data processing module, a data derivation module, a data splicing module, and a data verification module.

[0038] The intelligent agent control module 101 includes: Task parsing unit 101-01 is used to parse the natural language task input by the user into a structured task description; The strategy generation unit 101-02 is used to construct prompt words and call the large model cognitive module; Tool scheduling unit 101-03 is used to convert strategies into tool invocation instruction sequences; Result verification unit 101-04 is used to perform three-dimensional quality assessment and closed-loop feedback of technology, business, and engineering.

[0039] The MCP protocol interface module includes a tool registry, discovery service, and state manager to enable dynamic registration, semantic discovery, load balancing, and lifecycle management of tools.

[0040] Among them, the MCP Server Feature Tool Library 104 contains: The basic data processing module 104-01 is used to perform data cleaning and standardization. The data derivation module 104-02 is used to generate statistical features, business indicators and time series features based on business constraints. The data splicing module 104-03 is used to realize intelligent feature combination and control; The data verification module 104-04 is used to perform feature generation log recording and multidimensional quality assessment. This concludes the embodiment.

[0041] A third aspect of the present invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements an intelligent data feature derivation method based on a large model and the MCP protocol.

[0042] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0043] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, systems, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0044] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0045] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for intelligent data feature derivation based on a large model and the MCP protocol, characterized in that, include: Receive a feature engineering task input by the user, the task including feature objectives, industry background and constraints; The task is parsed, and a structured feature derivation strategy is generated based on a feature engineering-specific prompt word template. The prompt word template includes at least a business target field, a data type field, and a constraint field. Based on the aforementioned feature derivation strategy, feature processing operations are performed by scheduling the feature tool library through the MCP protocol. The feature processing operations include basic data processing, data derivation, data concatenation, and data verification. Perform multidimensional quality assessment on the processed feature set and generate a feature generation log; Output the feature set that passes the evaluation.

2. The intelligent data feature derivation method based on a large model and MCP protocol according to claim 1, characterized in that, The structured feature derivation strategy based on feature engineering-specific prompt word templates includes the following steps: The parsed task elements are injected into the parameterized prompt word template; The large language model is used to reason about the constructed prompt words and generate machine-readable feature derivation strategies. The feature derivation strategy includes tool call sequences, parameter mapping relationships, and business constraint rules.

3. The intelligent data feature derivation method based on a large model and MCP protocol according to claim 1, characterized in that, The communication process of scheduling the feature tool library via the MCP protocol includes: Initiate a tool availability query and metadata retrieval to the tool registry center; Request execution permission from the state manager and make load balancing decisions; The call request containing the business constraint fields is routed to the corresponding feature tool instance; Asynchronously receive processing results and update tool status.

4. The intelligent data feature derivation method based on a large model and MCP protocol according to claim 1, characterized in that, The process involves performing a multi-dimensional quality assessment on the obtained feature set, wherein the assessment dimensions include: Technical dimensions: Evaluate the importance, stability, and collinearity among features; Business dimension: Assess the interpretability, business compliance, and business value of the features; Engineering dimension: Evaluate the computational efficiency, storage overhead, and incremental update support capabilities of the features.

5. The intelligent data feature derivation method based on a large model and MCP protocol according to claim 1, characterized in that, The method also includes a verification feedback step: if the multidimensional quality assessment fails, the feature derivation strategy regeneration mechanism is automatically triggered, the tool call sequence or parameters are adjusted and the feature processing operation is re-executed until the output features meet all constraints.

6. An intelligent data feature derivation system for implementing the method according to any one of claims 1-5, characterized in that, include: The intelligent agent control module is used to receive and parse user tasks, generate derivative strategies, and verify the results. The large model cognitive module is connected to the intelligent agent control module and is used for policy reasoning based on prompt word templates; The MCP protocol interface module is connected to the intelligent agent control module to achieve standardized communication and tool scheduling. The MCP Server feature tool library, connected to the MCP protocol interface module, is used to perform feature processing operations. It includes a basic data processing module, a data derivation module, a data splicing module, and a data verification module.

7. The intelligent data feature derivation system based on a large model and the MCP protocol according to claim 6, characterized in that, The intelligent agent control module includes: The task parsing unit is used to parse the natural language task input by the user into a structured task description; The strategy generation unit is used to construct prompt words and call the large model cognitive module; The tool scheduling unit is used to convert strategies into a sequence of tool invocation instructions; The results verification unit is used to perform three-dimensional quality assessment and closed-loop feedback of technology, business, and engineering.

8. The intelligent data feature derivation system based on a large model and the MCP protocol according to claim 6, characterized in that, The MCP protocol interface module includes a tool registration center, a discovery service, and a state manager to enable dynamic registration, semantic discovery, load balancing, and lifecycle management of tools.

9. The intelligent data feature derivation system based on a large model and the MCP protocol according to claim 6, characterized in that, The MCP Server feature tool library includes: The basic data processing module is used to perform data cleaning and standardization; The data derivation module is used to generate statistical features, business indicators, and time series features based on business constraints. The data splicing module is used to realize intelligent feature combination and control; The data verification module is used to perform feature generation log recording and multidimensional quality assessment.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the intelligent data feature derivation method based on a large model and the MCP protocol as described in any one of claims 1-5.

Images