CN-122022516-A - Method and device for sensing and predicting enterprise state based on resume data

CN122022516ACN 122022516 ACN122022516 ACN 122022516ACN-122022516-A

Abstract

The embodiment of the invention discloses an enterprise state sensing and predicting method and device based on resume data. The method comprises the steps of obtaining employee history resume data which are associated with a target enterprise and subjected to desensitization treatment from at least one third-party data platform, processing the resume data, extracting analysis indexes reflecting dynamic changes of employee groups of the target enterprise to generate an analysis index group, constructing and operating an enterprise state mixed deduction model according to the analysis index group to generate a predicted path group related to future states of the target enterprise, and generating an enterprise state analysis report group according to the predicted path group. The method and the device can improve the accuracy and the foresight of the future state prediction of the enterprise.

Inventors

CHEN FEI
DU YULONG
LIU DANRONG
SHEN DAWEI
GAO LEI

Assignees

北京阿尔法风控科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. An enterprise state sensing and predicting method based on resume data is characterized by comprising the following steps: Obtaining desensitized employee history resume data associated with the target enterprise from at least one third party data platform; processing the resume data, and extracting analysis indexes reflecting the dynamic changes of the employee groups of the target enterprise so as to generate an analysis index group; Constructing and running an enterprise state hybrid deduction model according to the analysis index set to generate a predicted path set about the future state of the target enterprise, wherein the predicted path comprises an evolution state sequence and corresponding risk assessment information; And generating an enterprise state analysis report group according to the predicted path group.
2. The method of claim 1, wherein processing the resume data to extract analysis indicators reflecting dynamic changes in the target enterprise employee group to generate an analysis indicator set comprises: processing the arbitrary time period information in the resume data to generate a group stability index group; Performing natural language processing on unstructured text information in the resume data to generate a group capacity activity index set; wherein the analysis index set includes the population stability index set and the population capacity activity index set.
3. The method of claim 2, wherein the performing natural language processing on unstructured text information in the resume data to generate a group of group capability activity indicators comprises: performing skill entity identification and standardization processing on the unstructured text information to obtain a standardized skill entity group; Constructing and maintaining a dynamically evolving enterprise-specific skill knowledge graph based on the standardized skill entity group; carrying out project topic mining on the unstructured text information to obtain a project topic group; And generating the group capacity activity index group according to the enterprise exclusive skill knowledge graph and the project theme group.
4. The method of claim 3, wherein said constructing and running an enterprise state hybrid-deduction model from said set of analysis indicators to generate a set of predicted paths for future states of said target enterprise comprises: Constructing an organization digital twin body of the target enterprise according to the analysis index set, wherein the organization digital twin body comprises an employee digital twin sub-model set and an organization network model; Initializing a qualitative simulation module based on the analysis index set and the organization digital twin, wherein the qualitative simulation module defines a qualitative variable set representing the macroscopic state of an enterprise and a state transition constraint rule among the qualitative variables; modeling an enterprise decision mechanism as a reinforcement learning agent, wherein an action space of the reinforcement learning agent comprises a preset strategic decision option; Taking the enterprise state at the current moment as an initial state, driving the reinforcement learning intelligent agent to explore and simulate in a state space defined by the qualitative simulation module, and outputting a multi-state evolution track as a candidate prediction path; and evaluating and screening the candidate predicted paths to generate the predicted path group.
5. The method of claim 4, wherein the organizational network model is constructed based on virtual collaboration relationships between employee digital twins models; The method further includes calculating impact weights of key employee nodes on network structural integrity based on the organizational network model to generate structural risk indicators.
6. The method according to claim 3 or 4, characterized in that the method further comprises: calculating an organization entropy value of a target enterprise in a specified time window according to the standardized skill entity group, wherein the organization entropy value is used for quantifying the concentration or the discrete degree of enterprise skill distribution; and generating digital gene vectors of the target enterprises at different time points according to the standardized skill entity group and the project theme group, and quantifying the strategic transformation amplitude of the enterprises by calculating the similarity between the digital gene vectors at different time points.
7. The method of claim 6, wherein the method further comprises: Performing linkage analysis on time sequence changes of the group stability index group, the group capacity activity index group and the tissue entropy value; Generating contradiction signal early warning information in response to detection of an index change mode meeting preset contradiction logic, wherein the preset contradiction logic comprises that the group capacity activity index group indicates strategic resource investment expansion, and the group stability index group simultaneously indicates core human resource loss exceeds a preset threshold.
8. The method of claim 1, wherein obtaining desensitized employee history resume data associated with the target enterprise from at least one third-party data platform comprises: under the federal learning framework, issuing a feature extraction model to each third-party data platform node; receiving encrypted intermediate feature information generated by each node based on the local desensitization resume data by utilizing the feature extraction model; And aggregating the encrypted intermediate characteristic information of each node to obtain basic data for generating the analysis index set.
9. An enterprise state sensing and predicting device based on resume data, comprising: A data acquisition unit configured to acquire desensitized employee history resume data associated with the target enterprise from at least one third party data platform; an index extraction unit configured to process the resume data, and extract analysis indexes reflecting dynamic changes of the target enterprise employee group, so as to generate an analysis index group; A model deduction unit configured to construct and run an enterprise state hybrid deduction model according to the analysis index set to generate a predicted path set about the future state of the target enterprise, wherein the predicted path includes an evolution state sequence and corresponding risk assessment information; and a report generating unit configured to generate an enterprise state analysis report group according to the predicted path group.
10. An electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon; When executed by the one or more processors, causes the one or more processors to implement the method of any of claims 1-8.

Description

Method and device for sensing and predicting enterprise state based on resume data Technical Field The embodiment of the disclosure relates to the technical field of computers, in particular to the field of big data analysis, artificial intelligence and enterprise risk management, and particularly relates to an enterprise state sensing and predicting method and device based on resume data. Background With the deep globalization and digital transformation of economy, the method accurately evaluates the business state of enterprises, predicts the future risk and development trend of the enterprises, and has important significance for credit decisions of financial institutions, due investigation of investment institutions and strategic management of the enterprises. Currently, mainstream evaluation methods rely heavily on financial statements regularly issued by enterprises, public market announcements, and third party credit reports. However, the prior art solutions described above have the significant limitation that, first, the financial data relied upon has a significant hysteresis, typically distributed quarterly or annually, and cannot reflect the dramatic changes in real-time dynamics and short term of enterprise operation. Secondly, there is a serious problem of information asymmetry, and enterprises as information providers may perform information disclosure management for financing, estimation or supervision, etc., resulting in difficulty in obtaining real, comprehensive and untangling internal information for external observers. More importantly, the dimensions of the existing assessment system are extremely single, focusing almost entirely on "wealth" (profit, cash flow, balance sheet) and "things" (assets, inventory, equipment), while systematically ignoring the most core, most active, most advanced risk or opportunity signal releasing element of the enterprise "people". Stability of enterprise employee groups, changes in skill structures, distribution of project experience, and collaborative patterns among each other are valuable data sources that are insights into their organizational health, technical competitiveness, innovation ability, and strategic direction, but have long been lacking in efficient technical means to externally quantify, dynamically, and systematically analyze them. With the popularity of internet recruitment platforms and professional social networks, a vast number of employees have left public, continuously updated personal resume data during the development of the job. The resume objectively and in detail records the work experience, mastering skills, participating projects and educational background of the individual spanning years, and forms 'digital footprint' big data reflecting the microcosmic human resource condition of the enterprise. However, how to convert the scattered, unstructured and semantic-rich resume data into deep insight into macroscopic states of organization and construct an intelligent model capable of simulating and predicting future evolution of enterprises through innovative technical means is a technical problem which is not solved in the field effectively. The prior attempts can only carry out simple keyword frequency statistics and trend description and lack deep semantic understanding and associated mining, or can only carry out post-early warning aiming at single indexes such as 'off-duty rate', and the like, and can not construct a causal analysis and deduction framework between 'personnel behavior-organization capacity-strategic decision-future state', so that the urgent demands of market on prospective, quantifiable and deep enterprise state perception and prediction can not be met. Disclosure of Invention The disclosure is in part intended to introduce concepts in a simplified form that are further described below in the detailed description. The disclosure is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. Some embodiments of the present disclosure provide methods and apparatuses for sensing and predicting enterprise states based on resume data, so as to solve the technical problems mentioned in the background section above. In a first aspect, some embodiments of the present disclosure provide a method for sensing and predicting enterprise states based on resume data, the method including obtaining employee history resume data associated with a target enterprise and subjected to desensitization processing from at least one third party data platform, processing the resume data, extracting analysis indicators reflecting dynamic changes of the target enterprise employee group to generate an analysis indicator set, constructing and running an enterprise state hybrid deduction model according to the analysis indicator set to generate a predicted path set about future states of the target enterprise, wherein a predicted pa