CN-121598924-B - Purchasing file generation and evaluation method based on process mining analysis

CN121598924BCN 121598924 BCN121598924 BCN 121598924BCN-121598924-B

Abstract

The invention provides a purchasing file generation and evaluation method based on process mining analysis, which relates to the technical field of enterprise management and comprises the steps of collecting event log data in an enterprise purchasing system, constructing a purchasing process hypergraph model through a process hypergraph discovery algorithm, generating a variant association topological structure, constructing a self-adaptive rule mapping network according to the purchasing process hypergraph model and the variant association topological structure, inputting specific purchasing demand data into an evolving template by utilizing a dynamic evolving purchasing file template of the network, filling file content by using a semantic intervention filling mechanism, and generating a complete purchasing file.

Inventors

LIANG XIN
DU JUNGANG
DONG PENG
WANG KEWEN
LU WEI

Assignees

中国人民解放军海军工程大学

Dates

Publication Date: 20260508
Application Date: 20260130

Claims (8)

1. A purchasing file generation and evaluation method based on process mining analysis is characterized by comprising the following steps: Sp1, collecting event log data in an enterprise purchasing system, constructing a purchasing process hypergraph model through a process hypergraph discovery algorithm, and generating a variant association topological structure; extracting purchasing event sequence data from an enterprise purchasing system through a multisource event aggregation interface, constructing a purchasing process hypergraph model by utilizing a process hypergraph discovery algorithm fusion embedding representation technology, and converting a variation path into an associated topological structure through a topology extraction component; sp2, constructing a self-adaptive rule mapping network according to the purchasing process hypergraph model and the variant association topological structure generated by Sp1, and dynamically evolving a purchasing file template by utilizing the network; the method comprises the steps of (1) inputting purchase demand data into a template of Sp2 evolution, filling file contents by using a semantic intervention filling mechanism to generate a complete purchase file, mapping the purchase demand data input by a user, including material properties, transaction conditions and performance terms, into template fields of Sp2 by using a semantic analysis component, and filling contract elements, negotiation terms and risk distribution contents by using a generation and identification iterative process of the semantic intervention filling mechanism; Sp4, evaluating the purchase file generated by Sp3 by using a process mining anti-facts path checking technology, calculating the matching of the file content and the anti-facts structure of the process model, and generating an evaluation intervention report; sp5, reconstructing a purchasing process hypergraph model, a variation association topological structure and a self-adaptive rule mapping network according to the evaluation intervention report of Sp4, and realizing closed-loop self-organizing iteration of file generation and evaluation.
2. The method for generating and evaluating the purchasing file based on the process mining analysis of claim 1, wherein in Sp2, according to a purchasing process hypergraph model and a variant association topological structure of Sp1, a graph neural network is used for capturing hyperedge characteristics, and characteristics are fused through a dynamic aggregation mechanism of an adaptive rule mapping network, and a purchasing file template comprising branch constraint, sequence dependence and entity association is generated through evolution.
3. The method of claim 1, wherein in Sp4, a process mining-based back facts playback technique is applied to convert Sp3 purchase files into virtual event traces, wherein the virtual event traces refer to simulated event sequences back-pushed from the purchase files, map file clauses to potential execution paths, map to Sp1 purchase process hypergraph models, calculate back facts path matches by a structural intervention algorithm, and generate an evaluation report containing deviation paths and intervention adjustments.
4. The method for generating and evaluating a purchasing file based on process mining analysis according to claim 1, wherein in Sp5, according to an evaluation report of Sp4, a self-organizing reconstruction mechanism is used for adjusting node connection of a purchasing process hypergraph model, side weight of a variant association topological structure and cluster parameters of an adaptive rule mapping network, so that collaborative reconstruction of the model and the network is realized.
5. The method for generating and evaluating a purchasing file based on process mining analysis according to claim 1, wherein in Sp1, a distributed consensus verification mechanism is further integrated, cross-system purchasing event sequence data is processed through an event verifier, and integrity of a purchasing process hypergraph model and reliability of a variant association topological structure are ensured.
6. The method for generating and evaluating the purchasing file based on the process mining analysis of claim 2, wherein a privacy isolation component is further embedded in Sp2, and the sensitive superside of the adaptive rule mapping network is packaged and protected to ensure data isolation in the template evolution process.
7. The method for generating and evaluating a purchasing file based on process mining analysis as recited in claim 3, wherein in Sp3, the purchasing demand data is associated and matched with the previous virtual event trace by combining a historical intervention path comparison mechanism, and structural consistency of semantic intervention filling is optimized.
8. The method for generating and evaluating a purchasing file based on process mining analysis as recited in claim 3, wherein a hierarchical counterfactual framework is introduced into Sp4, wherein a time sequence intervention verification component is used to check the path compliance of the virtual event trace to the purchasing process hypergraph model, and wherein an intervention adjustment path is structured and output.

Description

Purchasing file generation and evaluation method based on process mining analysis Technical Field The invention relates to the technical field of enterprise management, in particular to a purchasing file generation and evaluation method based on process mining analysis. Background With acceleration of enterprise digital transformation and complexity of supply chain management, purchasing process has become a key link affecting enterprise operation efficiency, cost control and risk prevention. Particularly in the context of globalization of purchasing, multi-provider collaboration, and increasingly stringent compliance requirements, there is a growing need for automated generation, standardization, and intelligent evaluation of purchasing files. For example, large manufacturing enterprises need to quickly generate raw material contracts containing complex terms, the service industry needs to accommodate frequently changing IT service agreements, retail enterprises need to deal with flexible terms of seasonal commodity purchases, and public institutions emphasize transparency and compliance of bidding documents. Currently, process mining techniques and natural language processing techniques have been widely used in the fields of business process optimization and document automation. However, the existing purchase file generation and evaluation technology still faces the following key technical problems when meeting the above requirements: The purchasing process model has insufficient expression capability, and the traditional process mining mostly adopts the models such as a Petri network or a simple directed graph, and the like, only can capture a linear sequence or a basic branch relation, and cannot effectively represent complex correlations such as multiparty interaction, group decision, high-order dependence and the like in purchasing activities, so that the deviation between the model found from a historical event log and an actual purchasing process is larger, and the accuracy and the adaptability of a follow-up file template are further influenced; The dynamic matching mechanism of the file generation and the process model is absent, the existing method is mostly dependent on a fixed rule template or simple keyword filling, the self-adaptive adjustment of process variation and demand change is absent, semantic level intervention of terms and historical path comparison cannot be realized, the problems of term conflict, missing of risk distribution content or files inconsistent with an actual execution path are easily generated, and the execution efficiency and legal effectiveness of a contract are seriously affected; The purchase file evaluation lacks prospective risk identification capability, the traditional evaluation is mainly based on post-hoc compliance inspection or manual audit, path deviation under a potential intervention scene is difficult to simulate, anti-fact risks (such as consequences caused by supplier change or delivery delay) cannot be quantified, potential deviation path detection rate is low, the evaluation report lacks structural intervention suggestion, and closed loop optimization and continuous improvement are difficult to support. What is needed is a method of procurement file generation and evaluation based on process mining analysis that addresses the above-described issues. Disclosure of Invention Technical problem to be solved Aiming at the defects of the prior art, the invention provides a purchasing file generation and evaluation method based on process mining analysis, which solves the problems of the prior art. Technical proposal In order to achieve the purpose, the invention is realized by the following technical scheme that the purchasing file generation and evaluation method based on process mining analysis comprises the following steps: sp1, collecting event log data in an enterprise purchasing system, constructing a purchasing process hypergraph model through a process hypergraph discovery algorithm, and generating a variant association topological structure; sp2, constructing a self-adaptive rule mapping network according to the purchasing process hypergraph model and the variant association topological structure generated by Sp1, and dynamically evolving a purchasing file template by utilizing the network; Sp3, inputting specific purchasing demand data into a Sp2 evolution template, and filling file contents by using a semantic intervention filling mechanism to generate a complete purchasing file; Sp4, evaluating the purchase file generated by Sp3 by using a process mining anti-facts path checking technology, calculating the matching of the file content and the anti-facts structure of the process model, and generating an evaluation intervention report; sp5, reconstructing a purchasing process hypergraph model, a variation association topological structure and a self-adaptive rule mapping network according to the evaluation intervention report of Sp4, and realizi