CN116485557A - Credit risk fusion prediction method and system based on knowledge graph - Google Patents
Credit risk fusion prediction method and system based on knowledge graph Download PDFInfo
- Publication number
- CN116485557A CN116485557A CN202310485135.0A CN202310485135A CN116485557A CN 116485557 A CN116485557 A CN 116485557A CN 202310485135 A CN202310485135 A CN 202310485135A CN 116485557 A CN116485557 A CN 116485557A
- Authority
- CN
- China
- Prior art keywords
- risk
- user
- behavior data
- result
- time interval
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000004927 fusion Effects 0.000 title claims abstract description 16
- 101100421536 Danio rerio sim1a gene Proteins 0.000 claims abstract description 18
- 101100495431 Schizosaccharomyces pombe (strain 972 / ATCC 24843) cnp1 gene Proteins 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 10
- 239000013598 vector Substances 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 6
- 230000006399 behavior Effects 0.000 claims 47
- 230000003542 behavioural effect Effects 0.000 claims 3
- 238000012546 transfer Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
- G06F18/256—Fusion techniques of classification results, e.g. of results related to same input data of results relating to different input data, e.g. multimodal recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
- G06N5/025—Extracting rules from data
-
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Technology Law (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a credit risk fusion prediction method and a credit risk fusion prediction system based on a knowledge graph, wherein the method is used for obtaining a knowledge graph early warning model through carrying out serialization processing on behavior data of a comparison user based on a preset time period and training; inputting behavior data characteristics of the current time of the target user into a knowledge graph early warning model to conduct risk prediction, and obtaining a first risk result; dividing according to the preset time period and the behavior data characteristics of the comparison user in the preset time periodDividing; calculating behavior data characteristic x of current time of target user b And x i Similarity sim1 between the target users, and calculating behavior data characteristics x of the current time of the target users b And x j The similarity sim2 between the two images, the similarity sim1 and the similarity sim2 are respectively compared with a set threshold value, and a second risk result is obtained; and fusing the first risk result and the second risk result through a pre-trained decision model to obtain a user risk prediction result, so that the safety of risk identification is improved.
Description
Technical Field
The invention relates to the field of risk prediction, in particular to a credit risk fusion prediction method and system based on a knowledge graph.
Background
In the current user fund and credit risk analysis method, personal behavior data is mostly directly input into a pre-trained neural network for recognition processing. For example, CN115631022a (publication date 20230120) discloses that when a first user receives a transfer of a second user, age information of the second user is acquired; acquiring transfer information of a plurality of users transferred to the first user in a preset time period when determining that the age of the second user exceeds a preset threshold according to the age information of the second user; and inputting the transfer information into a transfer risk identification model, and outputting a risk identification result of transferring to the first user. However, the method is only suitable for scenes with small data volume and simple frames, cannot be effectively applied, and cannot meet the recognition precision requirement under complex scenes because the accuracy and efficiency of risk recognition are not guaranteed due to large and complex data volume in the anomaly recognition aiming at big data risks.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims at solving the defects that the effectiveness and accuracy of behavior anomaly identification in the current user behavior data analysis technology are not guaranteed, and provides a credit risk fusion prediction method and a credit risk fusion prediction system based on a knowledge graph.
In order to achieve the aim of the invention, the invention adopts the following technical scheme:
in a first aspect, the present invention provides a credit risk fusion prediction method based on a knowledge graph, the method comprising the steps of:
step 1, collecting and comparing behavior data of a user in a preset time period; the preset time period is a time sequence formed by time intervals with the same step length;
step 2, carrying out serialization processing on the behavior data of the comparison user based on the preset time period, and training to obtain a knowledge graph early warning model reflecting the behavior data characteristics and the risk event relation; the nodes of the knowledge graph early warning model consist of three parts, namely a behavior data characteristic node, a time sequence node and a risk event node;
step 3, inputting behavior data characteristics of the current time of the target user into the knowledge graph early warning model to conduct risk prediction, and obtaining a first risk result;
step 4, dividing according to the preset time period and the behavior data characteristics of the comparison user in the preset time period: x is X H ＝{x 1 ,x 2 ,…,x h -a }; h is the total number of time interval segments of the same step size, h=1, 2,..h; x is x 1 Behavior data characteristics for a first time interval; x is x 2 Behavior data characteristics for a second time interval; x is x h Behavior data characteristics of the h time interval;
step 5, x i Is the behavior data characteristic of the ith time interval segment, which represents a risk-free state user; x is x j Behavior data features of the jth time interval, which represent users in a risk state;
step 6, calculating the behavior data characteristic x of the current time of the target user b And x i The degree of similarity between the two is s im1,
step 7, calculating the behavior data characteristic x of the current time of the target user b And x j The degree of similarity between the two is s im2,
step 8, comparing the similarity sim1 and the similarity sim2 with set thresholds respectively to obtain a second risk result;
and 9, effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result.
Further, the method further comprises the following steps: the behavior data features a behavior vector of a user, comprising: the credit performance behavior data and the user consumption behavior data generated by the user in each time interval are respectively obtained, the information extracted from the credit performance behavior data and the user consumption behavior data is subjected to digital processing to obtain user behavior sub-vectors of the user corresponding to each time interval, and then the user behavior sub-vectors of each time interval are spliced to obtain corresponding user behavior vectors.
Further, the method further comprises the following steps: step 3, inputting behavior data characteristics of the current time of the target user into the knowledge graph early warning model to perform risk prediction, and obtaining a first risk result, wherein the method specifically comprises the following steps: and taking the behavior data characteristics as risk characteristics, inputting the risk characteristics into the knowledge graph early warning model for calculation, and obtaining a risk score of the current time of the user.
Further, the method further comprises the following steps: and step 8, comparing the similarity sim1 and the similarity sim2 with set thresholds respectively to obtain a second risk result, wherein the method specifically comprises the following steps: and comparing the similarity sim1 and the similarity sim2 with corresponding preset thresholds respectively to judge whether the user behavior is abnormal or not, and obtaining corresponding risk levels.
Further, the method further comprises the following steps: step 9, effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result, which specifically includes: and carrying out weighted summation on the first risk result and the second risk result to obtain a final user risk prediction result.
In a second aspect, the present invention further provides a credit risk fusion prediction system based on a knowledge graph, where the system includes the following modules:
the acquisition module is used for acquiring and comparing behavior data of a user in a preset time period; the preset time period is a time sequence formed by time intervals with the same step length;
the training module is used for carrying out serialization processing on the behavior data of the comparison user based on the preset time period, and training to obtain a knowledge graph early warning model reflecting the behavior data characteristics and the risk event relation; the nodes of the knowledge graph early warning model consist of three parts, namely a behavior data characteristic node, a time sequence node and a risk event node;
the first risk prediction module is used for inputting the behavior data characteristics of the current time of the target user into the knowledge graph early warning model to perform risk prediction, so as to obtain a first risk result;
the segmentation module is used for carrying out segmentation and division according to the preset time period and the behavior data characteristics of the comparison user in the preset time period: x is X H ＝{x 1 ,x 2 ,…,x h -a }; h is the total number of time interval segments of the same step size, h=1, 2,..h; x is x 1 Behavior data characteristics for a first time interval; x is x 2 Behavior data characteristics for a second time interval; x is x h Behavior data characteristics of the h time interval;
x i is the behavior data characteristic of the ith time interval segment, which represents a risk-free state user; x is x j Behavior data features of the jth time interval, which represent users in a risk state;
a first calculation module for calculating behavior data characteristic x of the current time of the target user b And x i The degree of similarity between the two is s im1,
a second calculation module for calculating the behavior data characteristic x of the current time of the target user b And x j The degree of similarity between the two is s im2,
the second risk prediction module is used for comparing the similarity s im1 and the similarity s im2 with a set threshold value respectively to obtain a second risk result;
and the final prediction module is used for effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result.
Further, the method further comprises the following steps: the behavior vector acquisition module is characterized by behavior vectors of users, and comprises: the credit performance behavior data and the user consumption behavior data generated by the user in each time interval are respectively obtained, the information extracted from the credit performance behavior data and the user consumption behavior data is subjected to digital processing to obtain user behavior sub-vectors of the user corresponding to each time interval, and then the user behavior sub-vectors of each time interval are spliced to obtain corresponding user behavior vectors.
Further, the method further comprises the following steps: the first prediction module is configured to input behavior data characteristics of a current time of a target user into the knowledge graph early warning model to perform risk prediction, and obtain a first risk result, and specifically includes: and taking the behavior data characteristics as risk characteristics, inputting the risk characteristics into the knowledge graph early warning model for calculation, and obtaining a risk score of the current time of the user.
Further, the method further comprises the following steps: the second risk prediction module is configured to compare the similarity sim1 and the similarity sim2 with a set threshold value respectively to obtain a second risk result, and specifically includes: and comparing the similarity sim1 and the similarity sim2 with corresponding preset thresholds respectively to judge whether the user behavior is abnormal or not, and obtaining corresponding risk levels.
Further, the method further comprises the following steps: the final prediction module is configured to effectively fuse the first risk result and the second risk result through a pre-trained decision model, and obtain a final user risk prediction result, and specifically includes: and carrying out weighted summation on the first risk result and the second risk result to obtain a final user risk prediction result.
Compared with the prior art, the invention has the beneficial effects that: the invention discloses a credit risk fusion prediction method and a credit risk fusion prediction system based on a knowledge graph, which are used for providing a related knowledge graph early warning model for constructing behavior data feature nodes, time sequence nodes and risk event nodes, converting a problem of user behavior risk recognition into a knowledge graph relativity analysis problem, realizing risk recognition by utilizing the advantages of the knowledge graph, and effectively fusing a first prediction result obtained by the knowledge graph early warning model and a second prediction result obtained by a similarity calculation comparison method provided by the invention to obtain a final high-accuracy risk recognition result, thereby improving the safety of risk recognition.
Drawings
Fig. 1 is a schematic diagram of steps of a credit risk fusion prediction method based on a knowledge graph.
Detailed Description
The present application is described in further detail below with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be noted that, for convenience of description, only the portions related to the invention are shown in the drawings.
Example 1
As shown in fig. 1, the present embodiment provides a credit risk fusion prediction method based on a knowledge graph, which includes the following steps:
step 1, collecting and comparing behavior data of a user in a preset time period; the preset time period is a time sequence formed by time intervals with the same step length;
step 2, carrying out serialization processing on the behavior data of the comparison user based on the preset time period, and training to obtain a knowledge graph early warning model reflecting the behavior data characteristics and the risk event relation; the nodes of the knowledge graph early warning model consist of three parts, namely a behavior data characteristic node, a time sequence node and a risk event node;
step 3, inputting behavior data characteristics of the current time of the target user into the knowledge graph early warning model to conduct risk prediction, and obtaining a first risk result; the method specifically comprises the following steps: taking the behavior data characteristics as risk characteristics, inputting the risk characteristics into the knowledge graph early warning model for calculation, and obtaining a risk score of the current time of the user;
step 4, dividing according to the preset time period and the behavior data characteristics of the comparison user in the preset time period: x is X H ＝{x 1 ,x 2 ,…,x h -a }; h is the total number of time interval segments of the same step size, h=1, 2,..h; x is x 1 Behavior data characteristics for a first time interval; x is x 2 Behavior data characteristics for a second time interval; x is x h Behavior data characteristics of the h time interval;
step 5, x i Is the behavior data characteristic of the ith time interval segment, which represents a risk-free state user; x is x j Behavior data features of the jth time interval, which represent users in a risk state;
step 6, calculating the behavior data characteristic x of the current time of the target user b And x i The degree of similarity between the two is s im1,
step 7, calculating the behavior data characteristic x of the current time of the target user b And x j The degree of similarity between the two is s im2,
step 8, comparing the similarity sim1 and the similarity sim2 with set thresholds respectively to obtain a second risk result; the method specifically comprises the following steps: comparing the similarity sim1 and the similarity sim2 with preset corresponding thresholds respectively to judge whether the user behavior is abnormal or not, and obtaining corresponding risk levels;
step 9, effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result; the method specifically comprises the following steps: and carrying out weighted summation on the first risk result and the second risk result to obtain a final user risk prediction result.
In an alternative embodiment, the method further comprises: the behavior data features a behavior vector of a user, comprising: the credit performance behavior data and the user consumption behavior data generated by the user in each time interval are respectively obtained, the information extracted from the credit performance behavior data and the user consumption behavior data is subjected to digital processing to obtain user behavior sub-vectors of the user corresponding to each time interval, and then the user behavior sub-vectors of each time interval are spliced to obtain corresponding user behavior vectors.
Example two
Based on the same inventive concept as the first embodiment, the present embodiment provides a credit risk fusion prediction system based on a knowledge graph, which includes the following modules:
the acquisition module is used for acquiring and comparing behavior data of a user in a preset time period; the preset time period is a time sequence formed by time intervals with the same step length;
the training module is used for carrying out serialization processing on the behavior data of the comparison user based on the preset time period, and training to obtain a knowledge graph early warning model reflecting the behavior data characteristics and the risk event relation; the nodes of the knowledge graph early warning model consist of three parts, namely a behavior data characteristic node, a time sequence node and a risk event node;
the first risk prediction module is used for inputting the behavior data characteristics of the current time of the target user into the knowledge graph early warning model to perform risk prediction, so as to obtain a first risk result; the method specifically comprises the following steps: taking the behavior data characteristics as risk characteristics, inputting the risk characteristics into the knowledge graph early warning model for calculation, and obtaining a risk score of the current time of the user;
the segmentation module is used for carrying out segmentation and division according to the preset time period and the behavior data characteristics of the comparison user in the preset time period: x is X H ＝{x 1 ,x 2 ,…,x h -a }; h isTotal number of time interval segments of the same step size, h=1, 2,; x is x 1 Behavior data characteristics for a first time interval; x is x 2 Behavior data characteristics for a second time interval; x is x h Behavior data characteristics of the h time interval;
x i is the behavior data characteristic of the ith time interval segment, which represents a risk-free state user; x is x j Behavior data features of the jth time interval, which represent users in a risk state;
a first calculation module for calculating behavior data characteristic x of the current time of the target user b And x i The degree of similarity between the two is s im1,
a second calculation module for calculating the behavior data characteristic x of the current time of the target user b And x j The degree of similarity between the two is s im2,
the second risk prediction module is used for comparing the similarity s im1 and the similarity s im2 with a set threshold value respectively to obtain a second risk result; the method specifically comprises the following steps: comparing the similarity sim1 and the similarity sim2 with preset corresponding thresholds respectively to judge whether the user behavior is abnormal or not, and obtaining corresponding risk levels;
the final prediction module is used for effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result; the method specifically comprises the following steps: and carrying out weighted summation on the first risk result and the second risk result to obtain a final user risk prediction result.
In an alternative embodiment, the method further comprises: the behavior vector acquisition module is characterized by behavior vectors of users, and comprises: the credit performance behavior data and the user consumption behavior data generated by the user in each time interval are respectively obtained, the information extracted from the credit performance behavior data and the user consumption behavior data is subjected to digital processing to obtain user behavior sub-vectors of the user corresponding to each time interval, and then the user behavior sub-vectors of each time interval are spliced to obtain corresponding user behavior vectors.
The invention innovatively provides a related knowledge graph early warning model for constructing behavior data characteristic nodes, time sequence nodes and risk event nodes, the problem of risk identification of the user is converted into a knowledge graph relevance analysis problem, the risk identification is realized by utilizing the advantages of the knowledge graph, the first prediction result obtained by the knowledge graph early warning model and the second prediction result obtained by the similarity calculation comparison method provided by the invention are effectively fused, the final high-accuracy risk identification result is obtained, the safety of risk identification is improved, and the method can be suitable for complex abnormal user identification scenes.
The preferred embodiments of the invention disclosed above are intended only to assist in the explanation of the invention. The preferred embodiments are not exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best understand and utilize the invention. The invention is limited only by the claims and the full scope and equivalents thereof.
Claims (10)
1. The credit risk fusion prediction method based on the knowledge graph is characterized by comprising the following steps of:
step 1, collecting and comparing behavior data of a user in a preset time period; the preset time period is a time sequence formed by time intervals with the same step length;
step 2, carrying out serialization processing on the behavior data of the comparison user based on the preset time period, and training to obtain a knowledge graph early warning model reflecting the behavior data characteristics and the risk event relation; the nodes of the knowledge graph early warning model consist of three parts, namely a behavior data characteristic node, a time sequence node and a risk event node;
step 3, inputting behavior data characteristics of the current time of the target user into the knowledge graph early warning model to conduct risk prediction, and obtaining a first risk result;
step 4, dividing according to the preset time period and the behavior data characteristics of the comparison user in the preset time period: x is X H ＝{x 1 ,x 2 ,…,x h -a }; h is the total number of time interval segments of the same step size, h=1, 2,..h; x is x 1 Behavior data characteristics for a first time interval; x is x 2 Behavior data characteristics for a second time interval; x is x h Behavior data characteristics of the h time interval;
step 5, x i Is the behavior data characteristic of the ith time interval segment, which represents a risk-free state user; x is x j Behavior data features of the jth time interval, which represent users in a risk state;
step 6, calculating the behavior data characteristic x of the current time of the target user b And x i The similarity sim1 between the two,
step 7, calculating the behavior data characteristic x of the current time of the target user b And x j The similarity sim2 between the two images,
step 8, comparing the similarity sim1 and the similarity sim2 with set thresholds respectively to obtain a second risk result;
and 9, fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result.
2. The method of claim 1, wherein the behavioral data is characterized by a behavioral vector of the user, comprising: the credit performance behavior data and the user consumption behavior data generated by the user in each time interval are respectively obtained, the information extracted from the credit performance behavior data and the user consumption behavior data is subjected to digital processing to obtain user behavior sub-vectors of the user corresponding to each time interval, and then the user behavior sub-vectors of each time interval are spliced to obtain corresponding user behavior vectors.
3. The method according to claim 1, wherein the step 3 of inputting the behavior data feature of the current time of the target user into the knowledge-graph early-warning model to perform risk prediction, and obtaining the first risk result specifically includes: and taking the behavior data characteristics as risk characteristics, inputting the risk characteristics into the knowledge graph early warning model for calculation, and obtaining a risk score of the current time of the user.
4. The method according to claim 1, wherein the step 8 of comparing the similarity sim1 and the similarity sim2 with set thresholds respectively to obtain a second risk result specifically includes: and comparing the similarity sim1 and the similarity sim2 with corresponding preset thresholds respectively to judge whether the user behaviors are abnormal or not, and obtaining corresponding risk levels.
5. The method according to claim 1, wherein the step 9 of effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result specifically includes: and carrying out weighted summation on the first risk result and the second risk result to obtain a final user risk prediction result.
6. The credit risk fusion prediction system based on the knowledge graph is characterized by comprising the following modules:
the acquisition module is used for acquiring and comparing behavior data of a user in a preset time period; the preset time period is a time sequence formed by time intervals with the same step length;
the training module is used for carrying out serialization processing on the behavior data of the comparison user based on the preset time period, and training to obtain a knowledge graph early warning model reflecting the behavior data characteristics and the risk event relation; the nodes of the knowledge graph early warning model consist of three parts, namely a behavior data characteristic node, a time sequence node and a risk event node;
the first risk prediction module is used for inputting the behavior data characteristics of the current time of the target user into the knowledge graph early warning model to perform risk prediction, so as to obtain a first risk result;
the segmentation module is used for carrying out segmentation and division according to the preset time period and the behavior data characteristics of the comparison user in the preset time period: x is X H ＝{x 1 ,x 2 ,…,x h -a }; h is the total number of time interval segments of the same step size, h=1, 2,..h; x is x 1 Behavior data characteristics for a first time interval; x is x 2 Behavior data characteristics for a second time interval; x is x h Behavior data characteristics of the h time interval;
x i is the behavior data characteristic of the ith time interval segment, which represents a risk-free state user; x is x j Behavior data features of the jth time interval, which represent users in a risk state;
a first calculation module for calculating behavior data characteristic x of the current time of the target user b And x i The similarity sim1 between the two,
a second calculation module for calculating the behavior data characteristic x of the current time of the target user b And x j The similarity sim2 between the two images,
the second risk prediction module is used for comparing the similarity sim1 and the similarity sim2 with a set threshold value respectively to obtain a second risk result;
and the final prediction module is used for effectively fusing the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result.
7. The system of claim 6, wherein the behavior data is characterized by a behavior vector of the user, comprising: the credit performance behavior data and the user consumption behavior data generated by the user in each time interval are respectively obtained, the information extracted from the credit performance behavior data and the user consumption behavior data is subjected to digital processing to obtain user behavior sub-vectors of the user corresponding to each time interval, and then the user behavior sub-vectors of each time interval are spliced to obtain corresponding user behavior vectors.
8. The system of claim 6, wherein the first prediction module is configured to input the behavioral data characteristic of the current time of the target user into the knowledge-graph early-warning model to perform risk prediction, so as to obtain a first risk result, and specifically includes: and taking the behavior data characteristics as risk characteristics, inputting the risk characteristics into the knowledge graph early warning model for calculation, and obtaining a risk score of the current time of the user.
9. The system according to claim 6, wherein the second risk prediction module is configured to compare the similarity sim1 and the similarity sim2 with set thresholds, respectively, to obtain a second risk result, and specifically includes: and comparing the similarity sim1 and the similarity sim2 with corresponding preset thresholds respectively to judge whether the user behaviors are abnormal or not, and obtaining corresponding risk levels.
10. The system according to claim 6, wherein the final prediction module is configured to effectively fuse the first risk result and the second risk result through a pre-trained decision model to obtain a final user risk prediction result, and specifically includes: and carrying out weighted summation on the first risk result and the second risk result to obtain a final user risk prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310485135.0A CN116485557A (en) | 2023-04-28 | 2023-04-28 | Credit risk fusion prediction method and system based on knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310485135.0A CN116485557A (en) | 2023-04-28 | 2023-04-28 | Credit risk fusion prediction method and system based on knowledge graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116485557A true CN116485557A (en) | 2023-07-25 |
Family
ID=87219250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310485135.0A Withdrawn CN116485557A (en) | 2023-04-28 | 2023-04-28 | Credit risk fusion prediction method and system based on knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116485557A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117094817A (en) * | 2023-10-20 | 2023-11-21 | 国任财产保险股份有限公司 | Credit risk control intelligent prediction method and system |
-
2023
- 2023-04-28 CN CN202310485135.0A patent/CN116485557A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117094817A (en) * | 2023-10-20 | 2023-11-21 | 国任财产保险股份有限公司 | Credit risk control intelligent prediction method and system |
| CN117094817B (en) * | 2023-10-20 | 2024-02-13 | 国任财产保险股份有限公司 | Credit risk control intelligent prediction method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112199375B (en) | Cross-modal data processing method and device, storage medium and electronic device | |
| CN110378366B (en) | Cross-domain image classification method based on coupling knowledge migration | |
| Park et al. | Probabilistic representations for video contrastive learning | |
| CN111695415A (en) | Construction method and identification method of image identification model and related equipment | |
| Meng et al. | Leveraging concept association network for multimedia rare concept mining and retrieval | |
| CN109977895B (en) | Wild animal video target detection method based on multi-feature map fusion | |
| CN109034086B (en) | Vehicle weight identification method, device and system | |
| CN112199536A (en) | Cross-modality-based rapid multi-label image classification method and system | |
| CN116485557A (en) | Credit risk fusion prediction method and system based on knowledge graph | |
| CN113822315A (en) | Attribute graph processing method and device, electronic equipment and readable storage medium | |
| CN111177390A (en) | Accident vehicle identification method and device based on hybrid model | |
| CN117041017B (en) | Intelligent operation and maintenance management method and system for data center | |
| CN111402156B (en) | Restoration method and device for smear image, storage medium and terminal equipment | |
| CN113723426A (en) | Image classification method and device based on deep multi-flow neural network | |
| CN110414541B (en) | Method, apparatus, and computer-readable storage medium for identifying an object | |
| CN114881173A (en) | Resume classification method and device based on self-attention mechanism | |
| CN113449676B (en) | Pedestrian re-identification method based on two-way interaction-based disentanglement learning | |
| Chen et al. | Sample prior guided robust model learning to suppress noisy labels | |
| CN117078007A (en) | Multi-scale wind control system integrating scale labels and method thereof | |
| CN117095460A (en) | Self-supervision group behavior recognition method and system based on long-short time relation predictive coding | |
| KR20210018586A (en) | Method and apparatus for identifying video content using biometric features of characters | |
| CN112052674B (en) | Entity definition extraction method, system, storage medium and server | |
| CN114841588A (en) | Information processing method, device, electronic equipment and computer readable medium | |
| CN114676840A (en) | Mass data based causal event group extraction method and system and computer readable storage medium | |
| CN114494809A (en) | Feature extraction model optimization method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230725 |