CN111753928B - Customs detection rule generation method based on knowledge graph and tree model construction - Google Patents
Customs detection rule generation method based on knowledge graph and tree model construction Download PDFInfo
- Publication number
- CN111753928B CN111753928B CN202010741825.4A CN202010741825A CN111753928B CN 111753928 B CN111753928 B CN 111753928B CN 202010741825 A CN202010741825 A CN 202010741825A CN 111753928 B CN111753928 B CN 111753928B
- Authority
- CN
- China
- Prior art keywords
- customs
- parameters
- rule
- tree model
- sample data
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Image Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a customs detection rule generation method based on a knowledge graph and a tree model structure, which is characterized in that a plurality of effective index features are extracted from a customs detection traditional expert rule engine, feature information is generally captured through the following two parts, the first part is used for extracting the feature information based on a plurality of CART decision trees to construct a random forest, the other part is used for learning a CART decision tree model based on a correlation relation between knowledge graph learning customs report data, and the learned correlation relation is used for learning the CART decision tree model.
Description
Technical Field
The invention relates to the technical field of customs detection, in particular to a customs detection rule generation method based on knowledge graph and tree model construction.
Background
With the development of economy, people have more demands on articles across the country, and import and export trade volume in the across-country field is obviously increased, which brings great pressure to customs departments for detecting import and export articles. The international trade bill is as high as 400 or more, and the traditional expert rule-based engine can not meet the hard indexes of customs, such as improvement of control efficiency, check rate and the like.
In the traditional customs inspection method, an expert rule engine is most widely applied. The expert rule engine effectively utilizes the related knowledge of the expert field to construct a rule system for detecting the articles. The expert system builds hundreds of rules and tens of thousands of rules, but the common rules only occupy a few of the overall rules, that is, the expert rule system only applies a small part of the rules in all the information of the articles, and many other related information is not utilized. Whereas the unused information often has a correlation with the outcome of the item detection that is difficult to find.
At present, some detection methods based on the tree model automatic generation rules are presented, however, the decision process of the decision tree often has the problem of single standard, and the decision process is separated from the original expert rule system, so that the knowledge of the expert rule system is lost although all relevant information of the articles can be utilized, and the knowledge waste of the expert information is caused.
Disclosure of Invention
In view of the above, the invention provides a method for generating a customs inspection rule based on knowledge graph and tree model construction, which is used for improving customs management and control efficiency and improving the acquisition rate.
The invention provides a customs detection rule generation method based on knowledge graph and tree model construction, which comprises the following steps:
s1: extracting a customs clearance list of a history detection article from a customs detection database, and cleaning data of the customs clearance list;
s2: taking the most common rule in a customs inspection expert system as a preliminary rule;
s3: carrying out characteristic engineering treatment on the customs clearance report after data cleaning, analyzing customs report parameters contained in the customs report after the characteristic engineering treatment, and extracting non-zero value parameters from the customs report parameters;
s4: extracting sample data from the extracted non-zero value parameters in a replacement sampling mode, performing primary screening on the extracted sample data by using the primary rule, and training the CART decision tree model by using the sample data after the primary screening to obtain a trained CART decision tree model;
s5: returning to the step S4, and repeatedly executing the step S4; after repeated times, weighting and fusing the trained CART decision tree models in a superposition mode to obtain a random forest;
s6: based on related knowledge in the field of customs detection experts, establishing a customs clearance sheet containing relationship comprising a customs clearance sheet type containing relationship and a customs clearance sheet parameter containing relationship;
s7: respectively treating each customs clearance sheet and each customs clearance sheet parameter as a graph node;
s8: judging whether a containing relation exists among the customs notes, the customs notes parameters and the customs notes parameters or not; if yes, establishing an edge between two nodes with inclusion relation, wherein the edge points to an included person from the included person, and the weight of the edge is a probability value from the included person to the included person; if not, no communication relationship is established between the two nodes with no inclusion relationship; traversing all customs declaration and all customs declaration parameters to obtain a knowledge graph directed probability graph model;
s9: setting the customs clearance sheet and customs clearance sheet parameters after the preliminary rule screening as initial nodes n 1 ,n 2 …n k Setting a customs clearance sheet of a target object as a final node m 1 ,m 2 ,…m l Calculating the maximum probability value max P (m 1 ,m 2 ,…m l |n 1 ,n 2 …n k )；
S10: and carrying out weighted fusion on the maximum probability value of the knowledge graph directed probability graph model and the maximum probability value output by the random forest to generate a final customs detection rule.
In a possible implementation manner, in the method for generating customs inspection rules provided by the present invention, in step S3, a non-zero value parameter is time information, and in step S3, feature engineering processing is performed on a customs clearance report, and parameters of the customs report included in the customs report after feature engineering processing are analyzed, and the non-zero value parameter is extracted therefrom, which specifically includes the following steps:
s31: analyzing the time stamp in the customs clearance list, extracting the date from the time stamp, and judging whether the date is the p-th day of each week, the q-th day of each month or not, and whether the date is a holiday or not; p=1, 2, …,7; q=1, 2, …,31;
s32: marking holidays at different levels, and respectively representing the legal holidays, the long holidays and the short holidays by different numbers;
s33: judging whether the number of days from the next holiday is greater than a threshold alpha; if yes, the method is regarded as far away from legal holidays; if not, the method is regarded as being close to the legal holiday;
s34: the threshold α is adjusted.
In a possible implementation manner, in the method for generating a customs inspection rule provided by the present invention, in step S4, sample data is extracted from the extracted non-zero value parameter by adopting a put-back sampling manner, the extracted sample data is initially screened by using the preliminary rule, and a CART decision tree model is trained by using the initially screened sample data, so as to obtain a trained CART decision tree model, which specifically includes the following steps:
s41: randomly extracting a part of data from each customs clearance report as sample data, and learning the extracted sample data by using a CART decision tree model as a base classifier;
s42: and the learned sample data is put back into the original data.
According to the customs inspection rule generation method based on the knowledge graph and the tree model structure, a plurality of effective index features are extracted from a customs inspection traditional expert rule engine, feature information is generally captured through the following two parts, the first part is used for extracting the feature information based on a plurality of CART decision trees to construct a random forest, the other part is used for learning the CART decision tree model based on the correlation relation between knowledge graph learning customs report data, and the learned correlation relation is used for learning the CART decision tree model. The invention fully utilizes the interrelationship between the customs notes, and through the learning of the customs note parameters and the construction of the relationship of the knowledge graph, the relationship information between the data is discovered from different angles, so that the newly learned rule system is more complete and has higher accuracy.
Drawings
FIG. 1 is a schematic flow chart of a customs inspection rule generation method based on knowledge graph and tree model construction;
fig. 2 is a flowchart of a customs inspection rule generation method based on knowledge graph and tree model construction.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is apparent that the described embodiments are merely examples and are not intended to limit the present invention.
The invention provides a customs detection rule generation method based on knowledge graph and tree model construction, as shown in fig. 1 and 2, comprising the following steps:
s1: extracting customs clearance list of history detection articles from a customs detection database, and cleaning data of the customs clearance list;
s2: taking the most common rule in a customs inspection expert system as a preliminary rule;
s3: carrying out characteristic engineering treatment on the customs clearance report after data cleaning, analyzing customs report parameters contained in the customs report after the characteristic engineering treatment, and extracting non-zero value parameters from the customs report parameters;
s4: extracting sample data from the extracted non-zero value parameters by adopting a replacement sampling mode, carrying out preliminary screening on the extracted sample data by utilizing a preliminary rule, and training a CART decision tree model by utilizing the sample data after the preliminary screening to obtain a trained CART decision tree model;
s5: returning to the step S4, and repeatedly executing the step S4; after repeated times, weighting and fusing the trained CART decision tree models in a superposition mode to obtain a random forest;
s6: based on related knowledge in the field of customs detection experts, establishing a customs clearance sheet containing relationship comprising a customs clearance sheet type containing relationship and a customs clearance sheet parameter containing relationship;
s7: respectively treating each customs clearance sheet and each customs clearance sheet parameter as a graph node;
s8: judging whether a containing relation exists among the customs notes, the customs notes parameters and the customs notes parameters or not; if yes, establishing an edge between two nodes with inclusion relation, wherein the edge points to an included person from the included person, and the weight of the edge is a probability value from the included person to the included person; if not, no communication relationship is established between the two nodes with no inclusion relationship; traversing all customs declaration and all customs declaration parameters to obtain a knowledge graph directed probability graph model;
s9: setting the customs declaration form and customs declaration form parameters after preliminary rule screening as initial node n 1 ,n 2 …n k Setting a customs clearance sheet of a target object as a final node m 1 ,m 2 ,…m l Calculating the maximum probability value max P (m 1 ,m 2 ,…m l |n 1 ,n 2 …n k )；
S10: and carrying out weighted fusion on the maximum probability value of the knowledge graph directed probability graph model and the maximum probability value output by the random forest to generate a final customs detection rule.
The specific implementation of the customs inspection rule generating method provided by the invention is described in detail below by a specific embodiment.
Example 1:
since the customs clearance is secret data, the original customs clearance data cannot be used for flow explanation, and therefore, the feasibility and effectiveness of the customs detection rule generation method provided by the invention are fully explained by taking import and export declaration forms (table 1) in a single window website of China international trade as an example.
Table 1 shows the report of import and export
| Type of newspaper delivery | Time of delivery | E-commerce platform code | Electronic port numbering | ..... | Article-related information |
| C1 | C14 | C..50 | C18 | ... | ... |
The first step: and extracting a reporting form of the history detection article from the customs detection database, and cleaning the data of the reporting form. For example, if the item-related information in table 1 contains a large number of missing values, the item-related information may not be used, that is, the item-related information in table 1 may be deleted because the item-related information is an insignificant indicator.
And a second step of: the most common rules in the customs inspection expert system are taken as preliminary rules.
And a third step of: and carrying out characteristic engineering treatment on the report after data cleaning, analyzing report parameters contained in the report after the characteristic engineering treatment, and extracting non-zero value parameters from the report parameters. The invention mainly decomposes time information, and discovers relevant information as far as possible from holidays on a time period. The specific method comprises the following steps:
(1) Analyzing the time stamp in the report after the data cleaning, extracting the date from the time stamp, and judging whether the date is the p-th day of each week, the q-th day of each month or not; p=1, 2, …,7; q=1, 2, …,31;
(2) Marking holidays at different levels, and respectively representing the legal holidays, the long holidays and the short holidays by different numbers;
(3) Judging whether the number of days from the next holiday is greater than a threshold alpha; if yes, the method is regarded as far away from legal holidays; if not, the method is regarded as being close to the legal holiday;
(4) The threshold α is adjusted. Specifically, the threshold α needs to be adjusted according to the training result.
Fourth step: extracting sample data from the extracted non-zero value parameters by adopting a replacement sampling (BootStrap) mode, carrying out preliminary screening on the extracted sample data by utilizing a preliminary rule, and training a CART decision tree model by utilizing the sample data after the preliminary screening to obtain a trained CART decision tree model, wherein the specific method comprises the following steps of:
(1) Randomly extracting a part of data from each report form as sample data, and learning the extracted sample data by using a CART decision tree model as a base classifier;
(2) And the learned sample data is put back into the original data.
Fifth step: returning to the fourth step, and repeatedly executing the fourth step; and after repeating for a plurality of times, weighting and fusing the trained CART decision tree models in a superposition (stacking) mode to obtain a random forest.
Sixth step: and establishing report form containing relations based on related knowledge of the customs detection expert field, wherein the report form containing relations comprise report form type containing relations and report form parameter containing relations.
Seventh step: and respectively treating each report form and each report form parameter as a graph node. Judging whether a containing relation exists among the reporting forms, the reporting form parameters and the reporting form parameters; if yes, establishing an edge between two nodes with inclusion relation, wherein the edge points to an included person from the included person, and the weight of the edge is a probability value from the included person to the included person; if not, no communication relation is established between the two nodes containing the relation, and the two nodes do not participate in subsequent calculation; and traversing all the reporting forms and all reporting form parameters to obtain a knowledge graph directed probability graph model.
Eighth step: setting a reporting form and reporting form parameters after preliminary rule screening as an initial node n 1 ,n 2 …n k Setting a reporting form and reporting form parameters of the target object as a final node m 1 ,m 2 ,…m l Calculating the maximum probability value max P (m 1 ,m 2 ,…m l |n 1 ,n 2 …n k )。
Ninth step: and carrying out weighted fusion on the maximum probability value of the knowledge graph directed probability graph model and the maximum probability value output by the random forest to generate a final customs detection rule.
The invention provides an algorithm model capable of automatically learning rules, which has different learning strategies in aspects of feature selection and feature learning. The introduction of the knowledge graph can enable the customs clearance information to have related information, break the independence between the original parameters, make decisions according to the correlation learned by the knowledge graph, and combine with the original expert system to synthesize a more powerful rule engine.
According to the customs inspection rule generation method based on the knowledge graph and the tree model structure, a plurality of effective index features are extracted from a customs inspection traditional expert rule engine, feature information is generally captured through the following two parts, the first part is used for extracting the feature information based on a plurality of CART decision trees to construct a random forest, the other part is used for learning the CART decision tree model based on the correlation relation between knowledge graph learning customs report data, and the learned correlation relation is used for learning the CART decision tree model. The invention fully utilizes the interrelationship between the customs notes, and through the learning of the customs note parameters and the construction of the relationship of the knowledge graph, the relationship information between the data is discovered from different angles, so that the newly learned rule system is more complete and has higher accuracy.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (3)
1. A customs detection rule generation method based on knowledge graph and tree model construction is characterized by comprising the following steps:
s1: extracting a customs clearance list of a history detection article from a customs detection database, and cleaning data of the customs clearance list;
s2: taking the most common rule in a customs inspection expert system as a preliminary rule;
s3: carrying out characteristic engineering treatment on the customs clearance report after data cleaning, analyzing customs report parameters contained in the customs report after the characteristic engineering treatment, and extracting non-zero value parameters from the customs report parameters;
s4: extracting sample data from the extracted non-zero value parameters in a replacement sampling mode, performing primary screening on the extracted sample data by using the primary rule, and training the CART decision tree model by using the sample data after the primary screening to obtain a trained CART decision tree model;
s5: returning to the step S4, and repeatedly executing the step S4; after repeated times, weighting and fusing the trained CART decision tree models in a superposition mode to obtain a random forest;
s6: based on related knowledge in the field of customs detection experts, establishing a customs clearance sheet containing relationship comprising a customs clearance sheet type containing relationship and a customs clearance sheet parameter containing relationship;
s7: respectively treating each customs clearance sheet and each customs clearance sheet parameter as a graph node;
s8: judging whether a containing relation exists among the customs notes, the customs notes parameters and the customs notes parameters or not; if yes, establishing an edge between two nodes with inclusion relation, wherein the edge points to an included person from the included person, and the weight of the edge is a probability value from the included person to the included person; if not, no communication relationship is established between the two nodes with no inclusion relationship; traversing all customs declaration and all customs declaration parameters to obtain a knowledge graph directed probability graph model;
s9: setting the customs clearance sheet and customs clearance sheet parameters after the preliminary rule screening as initial nodes n 1 ,n 2 …n k Setting a customs clearance sheet of a target object as a final node m 1 ,m 2 ,…m l Calculating the maximum probability value max P (m 1 ,m 2 ,…m l |n 1 ,n 2 …n k )；
S10: and carrying out weighted fusion on the maximum probability value of the knowledge graph directed probability graph model and the maximum probability value output by the random forest to generate a final customs detection rule.
2. The customs inspection rule generating method according to claim 1, wherein in step S3, the non-zero value parameter is time information, and in step S3, the feature engineering processing is performed on the customs inspection sheet after the data cleaning, the customs inspection sheet parameter included in the customs inspection sheet after the feature engineering processing is analyzed, and the non-zero value parameter is extracted therefrom, specifically comprising the steps of:
s31: analyzing the time stamp in the customs clearance list, extracting the date from the time stamp, and judging whether the date is the p-th day of each week, the q-th day of each month or not, and whether the date is a holiday or not; p=1, 2, …,7; q=1, 2, …,31;
s32: marking holidays at different levels, and respectively representing the legal holidays, the long holidays and the short holidays by different numbers;
s33: judging whether the number of days from the next holiday is greater than a threshold alpha; if yes, the method is regarded as far away from legal holidays; if not, the method is regarded as being close to the legal holiday;
s34: the threshold α is adjusted.
3. The customs inspection rule generating method according to claim 1, wherein in step S4, sample data is extracted from the extracted non-zero value parameters by a put-back sampling method, the extracted sample data is subjected to preliminary screening by using the preliminary rule, and the CART decision tree model is trained by using the preliminarily screened sample data, so as to obtain a trained CART decision tree model, and specifically comprising the following steps:
s41: randomly extracting a part of data from each customs clearance report as sample data, and learning the extracted sample data by using a CART decision tree model as a base classifier;
s42: and the learned sample data is put back into the original data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010741825.4A CN111753928B (en) | 2020-07-29 | 2020-07-29 | Customs detection rule generation method based on knowledge graph and tree model construction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010741825.4A CN111753928B (en) | 2020-07-29 | 2020-07-29 | Customs detection rule generation method based on knowledge graph and tree model construction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111753928A CN111753928A (en) | 2020-10-09 |
| CN111753928B true CN111753928B (en) | 2023-05-16 |
Family
ID=72712319
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010741825.4A Active CN111753928B (en) | 2020-07-29 | 2020-07-29 | Customs detection rule generation method based on knowledge graph and tree model construction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111753928B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116776831B (en) * | 2023-03-27 | 2023-12-22 | 阿里巴巴（中国）有限公司 | Customs, customs code determination, decision tree construction method and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108073673A (en) * | 2017-05-15 | 2018-05-25 | 北京华宇元典信息服务有限公司 | A kind of legal knowledge map construction method, apparatus, system and medium based on machine learning |
| CN110609906A (en) * | 2019-09-16 | 2019-12-24 | 金色熊猫有限公司 | Knowledge graph construction method and device, storage medium and electronic terminal |
| CN111428051A (en) * | 2020-03-23 | 2020-07-17 | 郑州大学 | Method and system for constructing self-adaptive learning knowledge graph fused with multivariate intelligence |
| CN111444410A (en) * | 2020-03-27 | 2020-07-24 | 民生科技有限责任公司 | Associated transaction mining and identifying method and device based on knowledge graph |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7139764B2 (en) * | 2003-06-25 | 2006-11-21 | Lee Shih-Jong J | Dynamic learning and knowledge representation for data mining |
-
2020
- 2020-07-29 CN CN202010741825.4A patent/CN111753928B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108073673A (en) * | 2017-05-15 | 2018-05-25 | 北京华宇元典信息服务有限公司 | A kind of legal knowledge map construction method, apparatus, system and medium based on machine learning |
| CN110609906A (en) * | 2019-09-16 | 2019-12-24 | 金色熊猫有限公司 | Knowledge graph construction method and device, storage medium and electronic terminal |
| CN111428051A (en) * | 2020-03-23 | 2020-07-17 | 郑州大学 | Method and system for constructing self-adaptive learning knowledge graph fused with multivariate intelligence |
| CN111444410A (en) * | 2020-03-27 | 2020-07-24 | 民生科技有限责任公司 | Associated transaction mining and identifying method and device based on knowledge graph |
Non-Patent Citations (1)
| Title |
|---|
| 融合知识图谱及贝叶斯网络的智能推荐方法;侯位昭;张欣海;宋凯磊;司佳;齐幸辉;张博;;中国电子科学研究院学报(第05期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111753928A (en) | 2020-10-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110188198B (en) | Anti-fraud method and device based on knowledge graph | |
| CN108629413B (en) | Neural network model training and transaction behavior risk identification method and device | |
| WO2021164382A1 (en) | Method and apparatus for performing feature processing for user classification model | |
| US11699160B2 (en) | Method, use thereof, computer program product and system for fraud detection | |
| CN110147911B (en) | Social influence prediction model and prediction method based on content perception | |
| CN109949152A (en) | A kind of personal credit's violation correction method | |
| CN111882446A (en) | Abnormal account detection method based on graph convolution network | |
| CN110413319A (en) | A kind of code function taste detection method based on deep semantic | |
| CN109635010B (en) | User characteristic and characteristic factor extraction and query method and system | |
| CN111260568B (en) | Peak binarization background noise removing method based on multi-discriminator countermeasure network | |
| CN114756686A (en) | Knowledge reasoning and fault diagnosis method based on knowledge graph | |
| CN111753928B (en) | Customs detection rule generation method based on knowledge graph and tree model construction | |
| CN112465226A (en) | User behavior prediction method based on feature interaction and graph neural network | |
| CN114117029B (en) | Solution recommendation method and system based on multi-level information enhancement | |
| CN111047428A (en) | Bank high-risk fraud client identification method based on small amount of fraud samples | |
| CN109446229B (en) | Big data and deep learning based identification method and robot system | |
| CN116503026A (en) | Operation and maintenance risk assessment method, system and storage medium for science and technology items | |
| Singh | Comparison between artificial neural network and linear model prediction performance for FDI disparity and the growth rate of companies in Hungarian counties | |
| CN107402984B (en) | A kind of classification method and device based on theme | |
| CN115604025A (en) | Network intrusion detection method based on PLI4DA | |
| CN114648118A (en) | Rule derivation method based on decision tree model and iterative elimination | |
| CN111754195B (en) | Information processing method and device, electronic equipment and computer readable storage medium | |
| Heydari et al. | Graph Representation Learning Towards Patents Network Analysis | |
| UTAMA et al. | Optimizing decision tree criteria to identify the released factors of COVID-19 patients in south korea | |
| Ramesh et al. | UPLIFT: Usage of Poverty Level Indicators to Facilitate Transformation |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |