CN-121981651-A - Data analysis decision-making method, system and medium based on mining area in electronic mall

CN121981651ACN 121981651 ACN121981651 ACN 121981651ACN-121981651-A

Abstract

The invention discloses a data analysis decision method, a system and a medium based on a mining area in an electronic mall, relates to the technical field of enterprise digital purchasing, and solves the problem of multi-system data isomerism by constructing a unified object model and carrying out main data fusion and caliber alignment. The discrete events are connected in series, the demand occurrence time is determined according to the priority rule, the real service time sequence is restored, and the demand time dislocation caused by the delay of the flow nodes is effectively avoided. By constructing the alternate relation map among materials and identifying the system section object, the adaptability of the system to complex scenes such as material shortage, policy change and the like is enhanced, and the flexibility and robustness of decision making are improved. And the demand prediction is carried out by fusing the demand event object, the substitution relation map and the system section object, and the prediction result is converted into a clear purchasing decision suggestion, so that the operability is remarkably improved. By establishing a feedback attribution and calibration updating mechanism, continuous learning and optimization of the system from an execution result are realized.

Inventors

YI YONGQIANG
ZHANG SHENGYU
HE WEN
ZHENG HONG
LIU YINGJUN
LI YINAN
Lv Weikun
YUAN ZAIXIN

Assignees

南方电网互联网服务有限公司

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. The data analysis decision-making method based on the mining area in the electronic mall is characterized by comprising the following steps: constructing a unified object model, namely butting the original data of a plurality of service systems, carrying out main data fusion and caliber alignment, and constructing the unified object model for representing the internal mining flow; generating a demand event object, namely based on the unified object model, concatenating discrete events in a purchasing link, determining demand occurrence time according to a priority rule, marking abnormal behaviors and generating a structured demand event object; Constructing a substitution relation map among materials based on the unified object model, and identifying a system section object representing structural change of a service environment; Generating a purchasing decision suggestion, namely carrying out demand prediction based on the demand event object, the substitution relation map and the system section object, and generating an executable purchasing decision suggestion based on a prediction result; And feeding back attribution and calibration updating, namely carrying out attribution analysis according to the actual execution result of the purchasing decision suggestion, and carrying out dynamic calibration on at least one of the unified object model, the substitution relation map, the system segment object or the purchasing decision suggestion based on the analysis result.
2. The method for data analysis and decision making based on the mining area in the electronic mall of claim 1, further comprising: And (3) aligning flow time delay modeling with executable time, namely respectively modeling approval, purchase generation and supply delivery to form an early-stage object, wherein the early-stage object is used for mapping the demand occurrence time to the order placing and arrival time.
3. The method for data analysis and decision-making based on the mining area in the electronic mall as set forth in claim 2, further comprising: and generating a demand decomposition feature, namely decomposing a demand event object into three driving types, namely project driving, seasonal and budget rhythm driving and abnormal disturbance driving, extracting time sequence features related to the driving types, and generating the demand decomposition feature in time alignment with the occurrence time of the demand.
4. The method for data analysis decision making based on the mining area in the electronic marketplace according to claim 3, wherein the purchasing decision suggestion generation comprises the following steps: Based on the demand decomposition characteristics, the early-stage objects and the system section objects, carrying out demand prediction, outputting a demand prediction interval and contribution degrees of different driving mechanisms, and carrying out characteristic migration prediction on the cold starting materials based on the substitution relation map; An executable purchasing decision suggestion is generated based on the demand forecast interval and the lead object.
5. The method for data analysis decision-making based on an e-mall in-collection area of claim 1, wherein the demand event object's demand occurrence time is determined from an associated plurality of business event timestamps according to predefined priority rules, the priority rules including at least demand reporting time prior to budget occupancy time prior to order creation time.
6. The method for data analysis and decision making based on the private area in the electronic mall, which is characterized in that the identification of the system section object comprises the steps of continuously monitoring key business indexes, and generating the system section object to identify business environment change when the change of the key business indexes exceeds a preset threshold value, wherein the key business indexes comprise alternative material use proportion, specific purchasing mode occupation ratio or average flow delay value.
7. The method for data analysis decision making based on the private area of electronic malls according to claim 1, wherein said executable purchasing decision advice includes at least information of advice delivery time, advice delivery quantity, advice safety stock level, and candidate replacement materials based on said replacement relational map.
8. The method for data analysis decision making based on the private area in the electronic marketplace according to claim 1, wherein the attribution analysis includes classifying deviations of the actual results from the purchasing decision advice as predicted deviations, time delay deviations, or execution deviations, attributing and processing the different types of deviations respectively, and the reasons of the execution deviations are structured as feedback information and written into a system log.
9. The data analysis decision-making system based on the mining area in the electronic mall is characterized by comprising the following components: the unified object model construction module is used for butting the original data of a plurality of service systems, carrying out main data fusion and caliber alignment, and constructing a unified object model for representing the internal acquisition flow; the demand event object generating module is used for serially connecting discrete events in the purchasing link based on the unified object model, determining demand occurrence time according to a priority rule and marking abnormal behaviors to generate a structured demand event object; The system comprises a system section object identification module, a system section object identification module and a system section object analysis module, wherein the system section object identification module is used for identifying a system section object representing structural change of a service environment; the intelligent decision module is used for fusing the demand event object, the substitution relation map and the system section object, carrying out demand prediction, and generating an executable purchasing decision suggestion based on a prediction result; And the feedback calibration module is used for carrying out attribution analysis according to the actual execution result of the purchasing decision suggestion, and carrying out dynamic calibration on at least one of the unified object model, the substitution relation map, the system section object or the purchasing decision suggestion based on the analysis result.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of a data analysis decision method based on a mining area in an electronic mall as claimed in any one of claims 1-8.

Description

Data analysis decision-making method, system and medium based on mining area in electronic mall Technical Field The invention relates to the technical field of enterprise digital purchasing, in particular to a data analysis decision-making method, a system and a medium based on a mining area in an electronic mall. Background In a modern enterprise digital purchasing system, a special area in an electronic mall is used as a key hub for connecting internal demands with external supplies, basic functions such as commodity display, online ordering, flow approval and financial reconciliation are generally integrated, and the purchasing operation efficiency and transparency are remarkably improved. However, with the continuous rise of the enterprise's fine operation requirements, the functions carried by the internal mining area are evolving from a "transaction execution platform" to an "intelligent decision center", with core challenges focused on how to achieve high-precision, executable demand forecasting and purchasing rhythm optimization. Current mainstream practice still relies primarily on historical orders or consumption data with stock assistance using moving averages, linear trend extrapolation, or safety inventory strategies based on limited business rules. Such methods remain basic in operation at an early stage of relatively smooth demand patterns with less external disturbances, and their design logic builds on the assumption of "repeatable history", i.e., it is believed that future demands can be deduced by simple statistical rules of past behavior. However, with the continuous increase of the production and management complexity of enterprises, the internal mining requirements show the characteristic of high unsteadiness, namely, on one hand, the requirements are generated to have strong project driving attribute, and often concentrated burst is performed around specific engineering nodes, development milestones or production plans, and on the other hand, the requirements and mechanisms of the same material in different organization units, different project stages and even different policy window periods are influenced by the institutional factors such as budget period, financial year, compliance policy adjustment and domestic replacement process. Further, the uncertainty of the external supply chain environment, such as supplier capacity fluctuation, logistic interruption, directory admission rule change, etc., also dynamically remodels the mapping relationship between demand and supply. Under the background, the prediction is carried out by simply relying on an order time sequence, multiple structural deviation is easy to fall into, firstly, the order generation time is often delayed from the real demand generation time, so that the prediction time sequence is misplaced, secondly, material specification iteration or code migration makes historical data incomparable, the existing system generally lacks structural modeling of a substitution relation and specification parameters, thirdly, a new project or a new material faces a 'cold start' dilemma due to lack of a history record, the traditional model cannot effectively migrate similar scene knowledge, fourthly, when policies or systems are switched (such as forced localization and compliance threshold lifting), the original statistical rule is totally invalid, the model parameters are systematically drifted, and the model parameters are recognized and responded by an inorganic system. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides a data analysis decision method, a system and a medium based on a special mining area in an electronic mall, and aims to solve the problems that the existing data analysis decision method of the special mining area in the electronic mall is low in demand prediction precision, ignores service flow time delay, is difficult to adapt to policy or system change and the like. The technical scheme adopted by the invention for solving the technical problems is that the data analysis decision method based on the mining area in the electronic mall comprises the following steps: constructing a unified object model, namely butting the original data of a plurality of service systems, carrying out main data fusion and caliber alignment, and constructing the unified object model for representing the internal mining flow; generating a demand event object, namely based on the unified object model, concatenating discrete events in a purchasing link, determining demand occurrence time according to a priority rule, marking abnormal behaviors and generating a structured demand event object; Constructing a substitution relation map among materials based on the unified object model, and identifying a system section object representing structural change of a service environment; Generating a purchasing decision suggestion, namely carrying out demand prediction based on the demand event object,