Search

CN-122019881-A - Multi-dimensional data screening and aggregation method based on user problem intention

CN122019881ACN 122019881 ACN122019881 ACN 122019881ACN-122019881-A

Abstract

The invention discloses a multidimensional data screening and aggregation method based on user problem intentions, which relates to the technical field of data processing and comprises the steps of firstly receiving user business problems and carrying out semantic analysis, and extracting core operation intentions and intention feature sets; the method comprises the steps of determining multidimensional data requirements according to inherent corresponding relations between business operation behaviors and commercial elements, generating a business exclusive data requirement dimension set, defining association relations among dimensions, sequentially generating a business exclusive initial data set and a verification data set through basic data screening integration and data consistency verification, performing multidimensional association operation on the verification data set, removing redundant data to obtain a business exclusive target aggregate data set, and finally presenting logic conversion data according to business decisions to generate a business exclusive structural answer and feeding the business exclusive structural answer back to a user side.

Inventors

  • ZHAO HANG

Assignees

  • 成都映潮科技股份有限公司

Dates

Publication Date
20260512
Application Date
20260203

Claims (8)

  1. 1. The multi-dimensional data screening and aggregation method based on the user problem intention is characterized by comprising the following steps of: step 100, receiving a commercial and trade problem input by a user, executing semantic analysis on the commercial and trade problem, and extracting a core operation intention and a corresponding intention feature set from an analysis result; Step 200, determining multidimensional data requirements based on inherent correspondence between business actions in the business operation field and business elements, generating a business exclusive data requirement dimension set, and determining dimension association relations among dimensions in the business exclusive data requirement dimension set; Step 300, screening matched commercial basic data from a preset structured data system based on the commercial exclusive data demand dimension set, and carrying out integration processing on the commercial basic data to generate a commercial exclusive initial data set; Step 400, classifying the dimension of the exclusive initial data set of the trade based on the exclusive data requirement dimension set of the trade, executing data consistency check on the classified exclusive initial data set of the trade, eliminating unqualified data, and generating an exclusive verification data set of the trade; Step 500, based on the business exclusive data demand dimension set and the dimension association relation, performing multidimensional association operation on the business exclusive verification data set, and removing redundant data generated in the operation process to generate a business exclusive target aggregation data set; Step 600, based on the core operation intention and the business exclusive data requirement dimension set, performing data conversion processing on the business exclusive target aggregate data set according to a business decision presentation logic, generating a business exclusive structural answer, and feeding back to a user side.
  2. 2. The method for multi-dimensional data screening and aggregation based on user question intents according to claim 1, wherein the step 100 comprises the sub-steps of: step 101, constructing a special feature word library of commerce and trade, wherein the feature word library is constructed in a classified manner according to a commercial main body, a business behavior and commercial elements; And step 102, performing semantic analysis on the received commercial questions based on the commercial exclusive feature word stock, and extracting core operation intention and corresponding intention feature set from the analysis result.
  3. 3. The method for multi-dimensional data screening and aggregation based on user question intents according to claim 1, further comprising the following sub-steps after step 200 and before step 300: A substep 201, performing decision type nesting division on a core operation intention, wherein the core operation intention comprises a single main decision type or a plurality of main decision types, and a multi-level business decision type system is formed; Step 202, performing data demand complementation processing on weak feature dimensions in a dimension set based on associated data of main feature dimensions in the data demand dimension set exclusive to commerce; And 203, sorting and optimizing the completed business exclusive data demand dimension sets according to the nested relation of the multi-level business decision type system to generate optimized business exclusive data demand dimension sets, and updating the dimension association relation among the dimensions in the optimized business exclusive data demand dimension sets.
  4. 4. The method for multidimensional data screening and aggregation based on user question intents according to claim 3, wherein when the core business intents include a plurality of main decision types, the following are included after the sub-step 201 and before the sub-step 202: Performing business logic decoupling processing on the multiple main decision types, and splitting independent requirement dimensions and cross-decision associated requirement dimensions of each main decision type in the optimized exclusive data requirement dimension set; sequencing a plurality of main decision types according to the priority of business operation decisions, configuring basic weights for independent demand dimensions, and executing weight sharing processing on cross-decision associated demand dimensions; And executing dimension weight dynamic hedging according to the business decision priority and the weight configuration rule, simultaneously executing cross-decision type cross-checking on cross-decision association requirement dimensions, and synchronously updating the dimension association relation among the dimensions in the optimized business exclusive data requirement dimension set.
  5. 5. The method for multi-dimensional data screening and aggregation based on user question intents according to claim 1, wherein the step 300 comprises the sub-steps of: Step 301, based on a business exclusive data requirement dimension set, extracting a business data pool with corresponding dimension from a preset structured data system, and dividing the business data pool into a core data layer and an auxiliary data layer according to influence degree on business operation decision; a sub-step 302 of executing full-scale aging verification on the core data layer, executing sampling aging verification on the auxiliary data layer, setting aging verification standards according to the level, and eliminating commercial and trade data exceeding the standards; A substep 303, performing commercial element feature anchoring processing on unstructured and semi-structured commercial data, extracting core element features matched with the commercial-dedicated data requirement dimension set, and associating the core element features with a preset structured field; in a substep 304, the aging-verified structured commerce data and the feature-anchored unstructured, semi-structured commerce data are integrated to generate a commerce-specific initial data set.
  6. 6. The method for multi-dimensional data screening and aggregation based on user question intents according to claim 1, wherein the step 400 comprises the sub-steps of: Step 401, based on the business exclusive data demand dimension set, conducting dimension classification filing on the business exclusive initial data set, conducting rasterization partition on the target business area according to the business value, and matching corresponding area grid labels for the classified and filed business exclusive initial data set; A substep 402, performing consistency check of data statistics caliber and data unit on the exclusive initial data set of business with regional grid label, and eliminating data with abnormal format and overrun value; step 403, setting differentiated region adaptation verification standards according to the region grid labels, configuring corresponding amateur saturation and consumption capability verification thresholds, and executing region adaptation verification; In step 404, based on the business-specific data requirement dimension set, determining element attributes of the checked data missing dimension, performing missing tolerance processing on the non-core element data, performing missing mark processing on the core element data, integrating the checked qualified data, the missing tolerance processed data and the missing mark processed data, and generating a business-specific verification data set.
  7. 7. The method for multi-dimensional data screening and aggregation based on user question intents according to claim 1, wherein the step 500 comprises the sub-steps of: in the substep 501, identifying pseudo-relevant data of the commercial-trade exclusive verification data set according to a preset rule of the commercial-trade exclusive data requirement dimension set, and eliminating the identified pseudo-relevant data; Step 502, constructing a dimension weight dynamic hedging rule, and dynamically adjusting the basic weights of all dimensions in a business exclusive data demand dimension set by combining the target area business ecological real-time data to realize weight balance among all dimensions; step 503, based on the dynamically adjusted dimension weight and dimension association relation, executing multidimensional association fusion operation on the business exclusive verification data set from which the pseudo related data is removed, and generating a business association operation intermediate result; In step 504, non-decision related redundant feature extraction is performed on the intermediate result of the business association operation, the extracted redundant feature is removed, and optimization processing is performed on the intermediate result of the business association operation after the redundant feature is removed, so as to generate a business exclusive target aggregation data set.
  8. 8. The method for multi-dimensional data screening and aggregation based on user question intents according to claim 1, wherein the step 600 comprises the sub-steps of: In the substep 601, hierarchical combing is performed on the exclusive target aggregate data set of the commerce and trade based on the element priority of the exclusive data demand dimension set of the commerce and trade, so as to generate a characteristic matrix of the commerce and trade data; step 602, a preset exclusive decision simulation model is called, and a characteristic matrix of commercial data is input into the exclusive decision simulation model to simulate and generate a plurality of groups of commercial operation decision results; sub-step 603, performing operation risk level division on the multiple groups of business operation decision results, matching corresponding decision feasibility assessment for the business operation decision results of different risk levels, and extracting core decision indexes from the business operation decision results of each risk level and the corresponding feasibility assessment; In step 604, the feature matrix of the business data, the multiple sets of business operation decision results, the feasibility assessment of each level and the core decision index are integrated in a modularized manner, and the integrated content is sequenced according to the reading logic of the business operation decision, so that a structured answer dedicated to the business is generated and fed back to the user side.

Description

Multi-dimensional data screening and aggregation method based on user problem intention Technical Field The invention relates to the technical field of data processing, in particular to a multi-dimensional data screening and aggregation method based on user problem intention. Background In a business operation decision scene, for supporting decision making, related data are screened, integrated and processed from massive business data based on business questions input by users, and finally a structured answer is generated for decision reference. At present, after receiving the business problem of the user in the prior art, only extracting keywords through simple semantic recognition, screening matching data according to preset general data dimensions, and sequentially carrying out operations such as data integration, verification, operation, conversion and the like. However, the data requirements are defined only by the keywords and the general dimensions, so that definition of the data requirements lacks pertinence and accuracy, a data range truly matched with the core business intention cannot be locked, the screened data and the core business intention are not sufficiently associated, and the finally generated structured answer is difficult to fit with the actual requirements of business decisions. Disclosure of Invention The invention provides a multidimensional data screening and aggregation method based on user question intention, which aims to solve the technical problem that the structured answer generated in the prior art is difficult to attach to the actual demand of business operation decisions. The technical scheme adopted by the invention is that the multidimensional data screening and aggregation method based on the intention of the user problem comprises the following steps: step 100, receiving a commercial and trade problem input by a user, executing semantic analysis on the commercial and trade problem, and extracting a core operation intention and a corresponding intention feature set from an analysis result; Step 200, determining multidimensional data requirements based on inherent correspondence between business actions in the business operation field and business elements, generating a business exclusive data requirement dimension set, and determining dimension association relations among dimensions in the business exclusive data requirement dimension set; Step 300, screening matched commercial basic data from a preset structured data system based on the commercial exclusive data demand dimension set, and carrying out integration processing on the commercial basic data to generate a commercial exclusive initial data set; Step 400, classifying the dimension of the exclusive initial data set of the trade based on the exclusive data requirement dimension set of the trade, executing data consistency check on the classified exclusive initial data set of the trade, eliminating unqualified data, and generating an exclusive verification data set of the trade; Step 500, based on the business exclusive data demand dimension set and the dimension association relation, performing multidimensional association operation on the business exclusive verification data set, and removing redundant data generated in the operation process to generate a business exclusive target aggregation data set; Step 600, based on the core operation intention and the business exclusive data requirement dimension set, performing data conversion processing on the business exclusive target aggregate data set according to a business decision presentation logic, generating a business exclusive structural answer, and feeding back to a user side. The invention has the beneficial effects that at least one of the following is adopted: according to the method, after the core business intention is extracted by analyzing the business problems, the multidimensional data requirements are determined based on the inherent corresponding relation between the business behaviors and the business elements in the business management field, and the association relation between the exclusive data requirement dimension set and each dimension of the business is generated, so that the limitation of the conventional generalized data dimension is eliminated, the defined data requirements are more attached to the actual attribute of the business scene, and the data inclusion irrelevant to the core business intention can be effectively reduced. The steps of data screening, integration, verification, operation and conversion are developed on the basis of a business exclusive data demand dimension set and dimension association relation, the problem of dislocation caused by lack of uniform basis in each link in the prior art is solved, the processed data is more matched with the core operation intention, and the generated business exclusive structured answer is more matched with the actual demand of business operation decision. Through the guidance of the