CN113642826A - Supplier default risk prediction method - Google Patents
Supplier default risk prediction method Download PDFInfo
- Publication number
- CN113642826A CN113642826A CN202110614810.6A CN202110614810A CN113642826A CN 113642826 A CN113642826 A CN 113642826A CN 202110614810 A CN202110614810 A CN 202110614810A CN 113642826 A CN113642826 A CN 113642826A
- Authority
- CN
- China
- Prior art keywords
- supplier
- data
- default risk
- knowledge graph
- risk prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000005516 engineering process Methods 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 14
- 238000013058 risk prediction model Methods 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 10
- 238000007493 shaping process Methods 0.000 claims description 10
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 4
- 238000012384 transportation and delivery Methods 0.000 claims description 4
- 230000009193 crawling Effects 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 abstract description 11
- 238000004519 manufacturing process Methods 0.000 description 23
- 238000011161 development Methods 0.000 description 8
- 238000011160 research Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000007726 management method Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000002950 deficient Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013068 supply chain management Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- Manufacturing & Machinery (AREA)
- Animal Behavior & Ethology (AREA)
- Databases & Information Systems (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a supplier default risk prediction method, which is characterized in that a knowledge graph technology is used for constructing a supplier knowledge graph, partial missing information in the knowledge graph with time sequence characteristics is restored by utilizing a knowledge graph prediction link technology, cross-domain data consideration is integrated, and the supplier default risk is predicted by adopting a neural network.
Description
Technical Field
The invention belongs to the technical field of supply chain management in manufacturing industry, and particularly relates to a supplier default risk prediction method.
Background
Compared with other industries, the manufacturing industry has various and complex manufacturing and assembling procedures, a single enterprise cannot carry out all work, and most of the manufacturing enterprises currently outsource non-core business of the enterprises and carry out production and manufacturing by purchasing parts. The parts suppliers have more and more initiative and technical strength, so that the supply chain efficiency is improved under more and more intense competitive environment, and the method is an operating strategy for realizing rapid development of enterprises. Suppliers play a very important role in the supply chain as a pre-stage link in the supply chain. The interruption or pause of any section of the industrial supply chain can cause the stop of the whole supply chain, and the scientific management of the supply chain is a key factor for realizing the strategic planning and the goal of an enterprise. The suppliers submit parts of warranty quantities that are very important to the manufacturing enterprises on the premise of improving the design ability and the development level, but with the current invariance of economic forms, the risk of default of the parts suppliers becomes more and more complicated, and once a problem occurs, the supply chain is likely to be interrupted, thereby bringing irreparable loss to all parties. It is therefore necessary to effectively predict risk to suppliers.
In terms of avoiding supplier risk, a manufacturing enterprise can predict current supplier default and predict current and future development trends of the supplier according to past supplier data. The management status of the enterprise, cash flow, resources, industry chain and other important information can be reflected from the supplier default risk prediction data. With the continuous development of internet technology, the purchase decision process of consumers starts to be switched from off-line to on-line, and the sales data of different dimensions and different domains are recorded, so that a richer research view is provided for the default prediction and selection of the part suppliers in the manufacturing industry. The influence of consumer evaluation, internet attention, manufacturer stock price information and macroscopic economic performance on the purchase demand of manufacturing enterprises is gradually enlarged, the type of the sales data is no longer single, and the superposition of multiple factors causes the sales data to become vital. Therefore, the enterprise can combine the data of the sales domain mastered by the enterprise with the information of different domains on the Internet, and guide the production decision of the manufacturing enterprise through data fusion and even through cross-domain data.
The supplier default risk prediction can be used as a novel enterprise management method to provide help for selecting suppliers for manufacturing enterprises through research analysis and prediction of the supplier sales data. For a manufacturing enterprise, accurate supplier default risk prediction can enable the enterprise to buy and use commodities with half the effort, and from another perspective, the added value of parts can be improved. From the aspect of industry development, the development of companies mastering market dynamics is bound to be more prosperous, so that the prosperous development of various industries in the society is promoted.
In conclusion, the prediction of the wind default risk of the supplier has important significance, and the good prediction of the wind default risk of the supplier determines the development prospect and direction of the company, so that the method is a direction worthy of research.
Disclosure of Invention
Aiming at the problems of lack of default risk prediction research of suppliers in the manufacturing industry, single data dimension and the like, the invention provides a supplier default risk prediction method which is carried out based on a knowledge graph and an RLSTM algorithm technology, so that the stability of an enterprise supply chain system is improved, the default risk of the suppliers can be judged as fast as possible, and the technical effects of reducing the economic and time losses of enterprises are achieved.
The invention is realized by adopting the following technical scheme:
a supplier default risk prediction method is provided, which comprises the following steps: acquiring historical data of a supplier and crawling related data of the supplier on the Internet; constructing a supplier knowledge graph based on the historical data and the relevant data; predicting missing information of the supplier knowledge graph based on a knowledge graph link prediction technology, and constructing a supplier time sequence knowledge graph; determining parameter indexes of supplier supply risks of a supplier, and analyzing and shaping input data of a supplier default risk prediction model based on the parameter indexes; constructing and training a supplier default risk prediction model based on a knowledge graph; and obtaining a supplier default risk prediction result based on the supplier default risk prediction model.
Further, constructing a knowledge graph of the supplier specifically comprises: constructing an ontology, and defining concept categories, attributes and relationships in the knowledge graph; data acquisition, namely identifying the historical data and the related data based on the definition in the ontology construction step; and extracting and fusing knowledge, namely performing entity extraction, relation extraction and attribute extraction on the identified knowledge, and completing entity alignment of supplier data by adopting a named entity attribute-based relation similarity comparison method to form a knowledge graph.
Further, analyzing and shaping the input data specifically includes: and (3) analysis: constructing a 3 x 3 influence factor matrix by taking default processing records, product delivery conditions, product operation conditions, product acceptance and service, consumer evaluation indexes, Internet attention indexes, enterprise stock market stock price drop and rise, and national macro economy GDP and CPI as nine kinds of influence factor indexes according to the related information of a supplier of a single part, wherein the product operation condition is a matrix center; shaping: and selecting a 3 x 3 influence factor matrix of eight suppliers with the closest operation condition of the supplier products of the single parts, and expanding the 3 x 3 influence factor matrix corresponding to the suppliers of the single parts in the analysis step into a 9 x 9 related supplier influence factor matrix, wherein the 3 x 3 influence factor matrices of the eight suppliers are arranged around the 3 x 3 influence factor matrix on the supply of the single parts from high to low according to the customer evaluation indexes.
Further, a supplier default risk prediction model constructed and based on the knowledge graphThe type specifically includes: by using
Normalizing the input data; where t is time, xtRepresenting the current input data at time t, xt(max)And xt(min)Respectively representing the maximum and minimum values of data in time series data, yt(max)And yt(min)Respectively representing the maximum value and the minimum value of the data after mapping; constructing an RLSTM neural network model; the RLSTM neural network model is trained using the SGD algorithm.
Further, the method further comprises: by using
Carrying out data reduction on the supplier default risk prediction result; where t is time, xtDenotes the actual predicted value at time t, xt(max)And xt(min)Respectively representing the maximum and minimum values of data in the time series data of the whole training, yt(max)And yt(min)Respectively representing the maximum and minimum values of the data after mapping.
Compared with the prior art, the invention has the advantages and positive effects that: the supplier default risk prediction method provided by the invention has the advantages that the knowledge graph technology is used for constructing the supplier knowledge graph, partial missing information in the knowledge graph with time sequence characteristics is restored by utilizing the knowledge graph prediction link technology, cross-domain data consideration is integrated, the supplier default risk is predicted by adopting the neural network, and compared with a single-domain data prediction mode of using historical sales or user feedback evaluation and the like in the traditional supplier default risk prediction, the risk of supplier supply in the prediction period can be effectively judged, the selection of future suppliers and the safe operation of a supply chain are facilitated to be analyzed, and the enterprise operation risk is reduced.
Other features and advantages of the present invention will become more apparent from the detailed description of the embodiments of the present invention when taken in conjunction with the accompanying drawings.
Drawings
FIG. 1 is a flow chart of a method for predicting a default risk of an enterprise provider according to the present invention;
FIG. 2 is a schematic diagram of knowledge graph construction according to the present invention;
FIG. 3 is a schematic diagram of an influencing factor matrix in the analysis of input data according to the present invention;
FIG. 4 is a schematic diagram of an influence factor matrix in input data shaping according to the present invention;
FIG. 5 is a schematic diagram of the training of the RLSTM neural network of the present invention.
Detailed Description
The following describes embodiments of the present invention in further detail with reference to the accompanying drawings.
The invention aims to provide a supplier default risk prediction method, based on five principles of a risk index system and combined with the characteristics of the manufacturing industry field, influence factors of the supplier default risk of the manufacturing industry enterprise are systematically analyzed, and the influence factors are quantified by using related indexes. Aiming at the problems of supplier default and the like of manufacturing enterprises, the RLSTM neural network structure is adopted to predict the supplier default risk, the supplier information is integrated into a knowledge graph of the supplier, the knowledge graph link prediction technology is used for completing the knowledge graph of the supplier, the information of related suppliers close to parts is input into an RLSTM model through shaping, the convolution operation of the RLSTM model is fully utilized, the high-dimensional logic connection between data of different suppliers in different domains is mined, and the prediction result is obtained.
Compared with the traditional prediction algorithm, the method has the advantages that reference data are more comprehensive by combining enterprise archived data and internet data, the time-sequence knowledge graph of the supplier is established by the knowledge graph link prediction technology, model prediction has better data support by data shaping, and the RLSTM has a more accurate prediction method.
Specifically, as shown in fig. 1, the supplier default prediction method provided by the present invention includes the following steps:
step S11: historical data of the suppliers is obtained, and relevant data of the suppliers on the Internet is crawled.
The supplier's historical data such as historical transaction data, historical default handling records, product delivery conditions, product operation conditions, product acceptance and service data, etc.
Relevant data is obtained from the internet based on crawler technology, for example: part data of different suppliers and likes and dislikes data of users on the microblog for the product. Through character analysis, distinguishing and calculating whether the evaluation of the user on the product is positive evaluation or negative evaluation; the times of searching the manufacturer by the user in the hundred-degree search engine; collecting the percentage of the stock price fluctuation range of the corresponding manufacturer in the current month through stock trading data; collecting the GDP and CPI values of the supplier in the month through the relevant data of the national statistical bureau; the method includes the steps of collecting previous shopping data of a business, and obtaining data when a transaction is conducted with a supplier.
Step S12: a provider knowledge graph is constructed based on historical data and related data.
The knowledge graph is composed of a mode layer structure and a data layer structure, and based on the structure, as shown in FIG. 2, the method for constructing the supplier knowledge graph comprises the following steps:
1) constructing an ontology: .
In the embodiment of the invention, the body is fully collected before designing, the related knowledge is comprehensively known by browsing the manufacturing industry pages of each large website, the related concepts are listed, and then the expert of the purchasing department of the enterprise is consulted according to the application scene. And finally, defining concept categories, attributes and relationships in the knowledge graph of the suppliers by combining the current supplier selection indexes of the current manufacturing enterprises and the default risk prediction indexes of the suppliers in the industry.
2) And (6) acquiring data.
Identifying the historical data and the related data obtained in the step S11 based on the definition in the ontology construction: the main task of the data layer is to identify concept categories, attributes, relationships and the like defined by the body layer from the acquired historical data and related data.
3) And extracting and fusing knowledge.
And performing entity extraction, relationship extraction and attribute extraction on the identified knowledge: entities and attributes of suppliers are respectively extracted by adopting named entity identification based on BilSTM-CRF (Bi-directional Long Short-Term Memory-conditional random field algorithm) and attribute extraction based on the BilSTM-CRF. And finishing entity alignment of the supplier data by adopting a named entity attribute relationship similarity comparison method, forming a knowledge graph and storing the knowledge graph.
Step S13: and predicting missing information of the knowledge graph of the supplier based on a knowledge graph link prediction technology, and constructing a time sequence knowledge graph of the supplier.
Although the data acquisition direction of the knowledge graph is wide, the semantic richness and the quality precision are slightly deficient, some structured facts are effective only in a specific time, the evolution of the facts follows a time sequence, and therefore, the missing links between entities are automatically predicted according to the facts existing in the knowledge graph by using a TransE-TAE model knowledge graph link prediction technology.
Step S14: and determining parameter indexes of supplier supply risks of the suppliers, and analyzing and shaping input data of the supplier default risk prediction model based on the parameter indexes.
Analyzing the default reasons of the suppliers, summarizing the influence factors of the default risks of the manufacturers according to five principles established by a risk index system, and determining the parameter indexes of the supply risks of the suppliers.
In the embodiment of the invention, the input data of the supplier default risk prediction model is analyzed based on the parameter indexes, and the method comprises the following steps: constructing a 3 x 3 influence factor matrix by taking default processing records, product delivery conditions, product operation conditions, product acceptance and service, consumer evaluation indexes, Internet attention indexes, enterprise stock market stock price drop and rise, and national macro economy GDP and CPI as nine kinds of influence factor indexes according to the related information of a supplier of a single part, wherein the product operation condition is a matrix center; as shown in fig. 3.
In the embodiment of the invention, the input data of the supplier default risk prediction model is shaped based on the parameter indexes, and the method comprises the following steps: in the following implementation of the invention, RLSTM neural network model is adopted to predict supplier default risk, and for the input of RLSTM model, the convolution effect of 3-3 influence source matrix is not ideal enough, so in the embodiment of the invention, the input data is expanded to 9 relevant supplier influence factor matrix, specifically, 3-3 influence factor matrix of eight suppliers with the closest supplier product operation condition of the single part is selected, the 3-3 influence factor matrix corresponding to the supplier of the single part in the analysis step is expanded to 9-9 relevant supplier influence factor matrix, wherein the 3-3 influence factor matrix of the eight suppliers is arranged on the supply of the single part from high to low according to consumer evaluation index The surroundings are shown in fig. 4.
Step S15: and constructing and training a supplier default risk prediction model based on the knowledge graph.
In the design of the model network, for non-single cross-domain data, the convolutional neural network can learn the high-dimensional logical relationship of the data through convolutional pooling operation, so in the embodiment of the invention, an RLSTM model is adopted, and a model for predicting the default risk of the cross-domain provider is constructed by combining the default risk data of the cross-domain provider.
The RLSTM neural network is a further improvement of the RLTM, the traditional LSTM adopts a full-connection form to transmit information during the state conversion of the neurons, the full-connection form before the input information enters each gate is replaced by a convolution form by the RLSTM, and the spatial information of the input information is extracted by utilizing a convolution structure. The LSTM neural network that RLSTM neural network can be fine with one-dimensional data input extends multidimensional data input, simultaneously, has introduced the peeking for the information transfer of neuron and has been connected to every gate, lets the gate layer can receive the state output at overflow moment, and the formula of RLSTM information transfer is as follows:
wherein, denotes the convolution calculation, xtRepresents the input of the neuron at time t; ctRepresenting the information data state of the neuron at time t; h istIndicating the status information and { i } passed to a later time at time tt，ot，ftAn input gate, an output gate and a forgetting gate represented by }; wx、WhAnd
representing the weights of the input, hidden and output layers, respectively, bf、bi、bc、boBoth represent an offset.
For the RLSTM model, the input data is three-dimensional data with time marks, a two-dimensional related supplier default risk influencing factor matrix needs to be expanded, and a time label is added to form three-dimensional matrix input.
In order to obtain a better training model and prevent the problem of gradient disappearance, related optimization preprocessing needs to be carried out on original data, therefore, normalization processing is carried out on input data of the RLSTM model, the numerical range of data with different dimensionalities can influence the accuracy of predicted data, data analysis can be facilitated through the normalization preprocessing, and the model training speed can be improved
Normalizing the input data; where t is time, xtRepresenting the current input data at time t, xt(max)And xt(min)Respectively representing the maximum and minimum values of data in time series data, yt(max)And yt(min)Respectively representing the maximum value and the minimum value of the data after mapping; in consideration of the fact that the supplier-related data does not exist less than 0, in the normalization process of the present model, 1 and 0 can be set here, respectively.
In the embodiment of the invention, the RLSTM neural network is constructed based on the tensorflow framework, a selected method and parameters need to be set in a model code written based on the tensorflow before the data input model starts to train, and the RLSTM neural network is trained by adopting an SGD algorithm.
The specific training process is as shown in fig. 5, and the training is automatically exited after the corresponding set exit condition is reached, and a learning curve is output; the specific training process and the related data such as loss function and parameters can be realized by those skilled in the art according to the application means of the conventional RSLTM application network, and the application of the present invention is not particularly limited.
In the embodiment of the invention, the input data of the model is normalized, so that the output result of the RLSTM neural network is not the final result and needs to be subjected to data reduction to obtain the prediction result of the model, and specifically, the method adopts
Carrying out data reduction on the supplier default risk prediction result; where t is time, xtDenotes the actual predicted value at time t, xt(max)And xt(min)Respectively representing the maximum and minimum values of data in the time series data of the whole training, yt(max)And yt(min)Respectively representing the maximum and minimum values of the data after mapping.
Step S16: and obtaining a supplier default risk prediction result based on the supplier default risk prediction model.
Based on the supplier default prediction model obtained by training in step S15, the supplier default prediction result can be obtained by inputting the supplier knowledge graph data obtained in steps S11 to S14 as a model.
The invention provides a manufacturing industry supplier risk prediction method based on a knowledge graph with universal significance, aiming at the problems of lack of research on the supplier risk violation prediction of the manufacturing industry and single data dimension based on a knowledge graph correlation technique and an RLSTM algorithm model, compared with the prior art, the invention uses the knowledge graph technique to construct the supplier knowledge graph, aiming at the problems that the semantic richness and the quality refinement of the knowledge graph are slightly deficient, some structured facts are only effective in specific time, the evolution of the facts follows the current situation of a time sequence, a TransE-TAE model is used for completing the knowledge graph, and missing links between entities are automatically predicted according to the existing facts in the knowledge graph; the invention provides cross-domain data provider default risk prediction, combines the historical sales of products, search attention indexes of the products on the Internet, public praise indexes of the products on a social network platform and the change conditions of manufacturing enterprise stocks, integrates comprehensive consideration of cross-domain data, predicts the provider risk through a neural network, can effectively judge the risk of provider supply in the prediction period, is beneficial to analyzing the selection of future providers, ensures the safe operation of a supply chain and reduces the enterprise operation risk.
It should be noted that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art should also make changes, modifications, additions or substitutions within the spirit and scope of the present invention.
Claims (5)
1. A supplier default risk prediction method, comprising the steps of:
acquiring historical data of a supplier and crawling related data of the supplier on the Internet;
constructing a supplier knowledge graph based on the historical data and the relevant data;
predicting missing information of the supplier knowledge graph based on a knowledge graph link prediction technology, and constructing a supplier time sequence knowledge graph;
determining parameter indexes of supplier supply risks of a supplier, and analyzing and shaping input data of a supplier default risk prediction model based on the parameter indexes;
constructing and training a supplier default risk prediction model based on a knowledge graph;
and obtaining a supplier default risk prediction result based on the supplier default risk prediction model.
2. The supplier default risk prediction method according to claim 1, wherein constructing a supplier knowledge graph specifically comprises:
constructing an ontology, and defining concept categories, attributes and relationships in the knowledge graph;
data acquisition, namely identifying the historical data and the related data based on the definition in the ontology construction step;
and extracting and fusing knowledge, namely performing entity extraction, relation extraction and attribute extraction on the identified knowledge, and completing entity alignment of supplier data by adopting a named entity attribute-based relation similarity comparison method to form a knowledge graph.
3. The supplier default risk prediction method of claim 1, wherein analyzing and shaping the input data specifically comprises:
and (3) analysis: constructing a 3 x 3 influence factor matrix by taking default processing records, product delivery conditions, product operation conditions, product acceptance and service, consumer evaluation indexes, Internet attention indexes, enterprise stock market stock price drop and rise, and national macro economy GDP and CPI as nine kinds of influence factor indexes according to the related information of a supplier of a single part, wherein the product operation condition is a matrix center;
shaping: and selecting a 3 x 3 influence factor matrix of eight suppliers with the closest operation condition of the supplier products of the single parts, and expanding the 3 x 3 influence factor matrix corresponding to the suppliers of the single parts in the analysis step into a 9 x 9 related supplier influence factor matrix, wherein the 3 x 3 influence factor matrices of the eight suppliers are arranged around the 3 x 3 influence factor matrix on the supply of the single parts from high to low according to the customer evaluation indexes.
4. The supplier default risk prediction method according to claim 1, wherein constructing and training a knowledge-graph-based supplier default risk prediction model specifically comprises:
by using
Normalizing the input data; where t is time, xtRepresenting the current input data at time t, xt(max)And xt(min)Respectively representing the maximum and minimum values of data in time series data, yt(max)And yt(min)Respectively representing the maximum value and the minimum value of the data after mapping;
constructing an RLSTM neural network model;
the RLSTM neural network model is trained using the SGD algorithm.
5. The supplier default risk prediction method of claim 4, further comprising:
by using
Carrying out data reduction on the supplier default risk prediction result; where t is time, xtDenotes the actual predicted value at time t, xt(max)And xt(min)Respectively representing the maximum and minimum values of data in the time series data of the whole training, yt(max)And yt(min)Respectively representing the maximum and minimum values of the data after mapping.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110614810.6A CN113642826A (en) | 2021-06-02 | 2021-06-02 | Supplier default risk prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110614810.6A CN113642826A (en) | 2021-06-02 | 2021-06-02 | Supplier default risk prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113642826A true CN113642826A (en) | 2021-11-12 |
Family
ID=78415894
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110614810.6A Pending CN113642826A (en) | 2021-06-02 | 2021-06-02 | Supplier default risk prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113642826A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114254916A (en) * | 2021-12-20 | 2022-03-29 | 国网江苏省电力有限公司物资分公司 | Multi-dimensional supplier risk real-time early warning method with power grid service characteristics |
| CN114648260A (en) * | 2022-05-24 | 2022-06-21 | 深圳装速配科技有限公司 | Building material purchasing method and building material purchasing platform |
| CN114707883A (en) * | 2022-04-18 | 2022-07-05 | 工银瑞信基金管理有限公司 | Bond default prediction method, device, equipment and medium based on time sequence characteristics |
| CN115099504A (en) * | 2022-06-29 | 2022-09-23 | 中南民族大学 | Cultural relic security risk element identification method based on knowledge graph complement model |
| CN115964503A (en) * | 2021-12-28 | 2023-04-14 | 北方工业大学 | Safety risk prediction method and system based on community equipment facilities |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657837A (en) * | 2018-11-19 | 2019-04-19 | 平安科技（深圳）有限公司 | Default Probability prediction technique, device, computer equipment and storage medium |
| EP3619712A1 (en) * | 2018-07-11 | 2020-03-11 | Illumina, Inc. | DEEP LEARNING-BASED FRAMEWORK FOR IDENTIFYING SEQUENCE PATTERNS THAT CAUSE SEQUENCE-SPECIFIC ERRORS (SSEs) |
| US20200118423A1 (en) * | 2017-04-05 | 2020-04-16 | Carnegie Mellon University | Deep Learning Methods For Estimating Density and/or Flow of Objects, and Related Methods and Software |
| CN112256887A (en) * | 2020-10-28 | 2021-01-22 | 福建亿榕信息技术有限公司 | Intelligent supply chain management method based on knowledge graph |
-
2021
- 2021-06-02 CN CN202110614810.6A patent/CN113642826A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200118423A1 (en) * | 2017-04-05 | 2020-04-16 | Carnegie Mellon University | Deep Learning Methods For Estimating Density and/or Flow of Objects, and Related Methods and Software |
| EP3619712A1 (en) * | 2018-07-11 | 2020-03-11 | Illumina, Inc. | DEEP LEARNING-BASED FRAMEWORK FOR IDENTIFYING SEQUENCE PATTERNS THAT CAUSE SEQUENCE-SPECIFIC ERRORS (SSEs) |
| CN109657837A (en) * | 2018-11-19 | 2019-04-19 | 平安科技（深圳）有限公司 | Default Probability prediction technique, device, computer equipment and storage medium |
| CN112256887A (en) * | 2020-10-28 | 2021-01-22 | 福建亿榕信息技术有限公司 | Intelligent supply chain management method based on knowledge graph |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114254916A (en) * | 2021-12-20 | 2022-03-29 | 国网江苏省电力有限公司物资分公司 | Multi-dimensional supplier risk real-time early warning method with power grid service characteristics |
| CN115964503A (en) * | 2021-12-28 | 2023-04-14 | 北方工业大学 | Safety risk prediction method and system based on community equipment facilities |
| CN115964503B (en) * | 2021-12-28 | 2023-07-07 | 北方工业大学 | Safety risk prediction method and system based on community equipment facilities |
| CN114707883A (en) * | 2022-04-18 | 2022-07-05 | 工银瑞信基金管理有限公司 | Bond default prediction method, device, equipment and medium based on time sequence characteristics |
| CN114648260A (en) * | 2022-05-24 | 2022-06-21 | 深圳装速配科技有限公司 | Building material purchasing method and building material purchasing platform |
| CN115099504A (en) * | 2022-06-29 | 2022-09-23 | 中南民族大学 | Cultural relic security risk element identification method based on knowledge graph complement model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kraus et al. | Deep learning in business analytics and operations research: Models, applications and managerial implications | |
| CN113642826A (en) | Supplier default risk prediction method | |
| Chen et al. | Strategic decisions using the fuzzy PROMETHEE for IS outsourcing | |
| Barak et al. | Outsourcing modelling using a novel interval-valued fuzzy quantitative strategic planning matrix (QSPM) and multiple criteria decision-making (MCDMs) | |
| Mehralian | Identifying and Explaining the Effective Factors of Digital Marketing Strategies in Consumers' Emotional States and Sales Rates: A Mixed Methods Research | |
| Rožanec et al. | Knowledge graph-based rich and confidentiality preserving Explainable Artificial Intelligence (XAI) | |
| Chen et al. | A soft-computing based rough sets classifier for classifying IPO returns in the financial markets | |
| Gulzat et al. | RESEARCH ON PREDICTIVE MODEL BASED ON CLASSIFICATION WITH PARAMETERS OF OPTIMIZATION. | |
| Saura et al. | Data-driven strategies in operation management: mining user-generated content in Twitter | |
| Agarwal et al. | Machine Learning and Natural Language Processing in Supply Chain Management: A Comprehensive Review and Future Research Directions. | |
| Li et al. | Multi‐criteria probabilistic dual hesitant fuzzy group decision making for supply chain finance credit risk assessments | |
| Türkbayrağí et al. | Artificial intelligence based prediction models: sales forecasting application in automotive aftermarket | |
| El Koufi et al. | Research on precision marketing based on big data analysis and machine learning: case study of Morocco | |
| Spreitzenbarth et al. | Artificial intelligence and machine learning in purchasing and supply management: A mixed-methods review of the state-of-the-art in literature and practice | |
| Wang et al. | Modeling of individual customer delivery satisfaction: An AutoML and multi-agent system approach | |
| Tian et al. | An end-to-end deep learning model for solving data-driven newsvendor problem with accessibility to textual review data | |
| Wang et al. | Risk Prevention of E-Commerce Supply Chain Financial Market Based on High-Performance Computing | |
| Chang | Evaluation model of enterprise lean management effect based on data mining | |
| Mondal et al. | Fact-based expert system for supplier selection with ERP data | |
| Ghosh et al. | Panic Selling Analysis in Stock Market | |
| Nath et al. | Predicting Financial Distress in Enterprises by Applying Multilevel Ensemble Technique | |
| Kaur et al. | Assessing efficacy of association rules for predicting global stock indices | |
| Agarwal et al. | Applications of Machine Learning Techniques in Supply Chain Management | |
| Kvist | Artificial Intelligence as Assistance in Warranty Claim Handling | |
| Zhao et al. | Multi-angle perception and convolutional neural network for service quality evaluation of cross-border e-commerce logistics enterprise |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information |
Inventor after: Gao Xiaoyan Inventor after: Yu Shusong Inventor after: Guo Baoqi Inventor after: Yang Ning Inventor after: Ding Xiangqian Inventor after: Shi Shuo Inventor after: Hou Ruichun Inventor after: Gong Huili Inventor before: Yu Shusong Inventor before: Gao Xiaoyan Inventor before: Guo Baoqi Inventor before: Yang Ning Inventor before: Ding Xiangqian Inventor before: Shi Shuo Inventor before: Hou Ruichun Inventor before: Gong Huili |
|
| CB03 | Change of inventor or designer information |