A method and system for deep mining of targets based on custom association relationships

Abstract

The present application relates to the field of data mining technology, and provides a method and system for deep mining of targets based on custom association relationships, the method comprising: obtaining mining target information; performing pan-target detection on the massive data existing in the target network through big data to obtain pan-target detection results; performing correlation indicator parameter analysis based on the pan-target detection results and mining target information to determine custom association relationships; performing key network detection on the pan-target detection results to obtain the actual traffic of the key network; performing target traffic detection and analysis on the actual traffic of the key network using custom association relationships to obtain high-value traffic clues; performing target re-identification based on the high-value traffic clues and mining target information to obtain target clue identification results. The use of this method can solve the technical problem of low target data acquisition rate and accuracy caused by excessive data volume in the target network.

Description

Technical Field

[0001] The present application relates to the field of data mining technology, and in particular to a method and system for deep mining of targets based on user-defined association relationships. Background Art

[0002] Data mining is a hot topic in current research in the fields of artificial intelligence and databases. Data mining refers to the nontrivial process of uncovering hidden, previously unknown, and potentially valuable information from large amounts of data in a database. Data mining is a decision-support process primarily based on artificial intelligence, machine learning, and databases. It uses highly automated data analysis to make inductive inferences and extract valuable data.

[0003] With the advent of the big data era, data is rapidly expanding and becoming larger. How to use the massive data in the network for data mining and application is a key issue. Due to the huge amount of data in the target network, the traditional target data acquisition method can no longer meet people's requirements.

[0004] In summary, the prior art has a technical problem in that the speed and accuracy of target data acquisition are low due to the excessive amount of data in the target network. Summary of the Invention

[0005] Based on this, it is necessary to provide a target deep mining method and system based on customized association relationships to address the above technical issues.

[0006] A target deep mining method based on custom association relationships, the method is applied to a target deep mining system, the method comprising: obtaining mining target information; based on the mining target information, performing pan-target detection on massive data existing in a target network through big data to obtain pan-target detection results; based on the pan-target detection results and the mining target information, performing correlation indicator parameter analysis to obtain a relevant indicator set, and determining custom association relationships based on the relevant indicator set; performing key network detection on the pan-target detection results to obtain actual traffic of the key network; using the custom association relationship to perform target traffic detection and analysis on the actual traffic of the key network to obtain high-value traffic clues; based on the high-value traffic clues and the mining target information, performing target re-identification to obtain target clue identification results.

[0007] In one embodiment, it also includes: based on the relevant indicator set, performing indicator confidence analysis with the mining target information to determine the confidence of each indicator; based on the confidence of each indicator, performing indicator screening to determine the target indicator; based on the target indicator, performing indicator relationship analysis to obtain the target indicator relationship; based on the target indicator relationship, performing custom relationship setting to obtain the custom association relationship.

[0008] In one embodiment, it also includes: setting a custom rule list, wherein the custom rule list includes indicator proportions and indicator correlations; determining the number of indicators and the cross-correlation of each indicator based on the target indicator relationship; determining the indicator combination relationship based on the number of indicators, the target indicator relationship, and the indicator proportion; calculating the combination correlation of the indicator combination relationship based on the cross-correlation of each indicator to determine the indicator combination correlation; matching the indicator combination correlation based on the indicator correlation in the custom rule list to obtain the custom correlation.

[0009] In one embodiment, it also includes: constructing a fitness function for mining target information and related indicator sets based on the target indicator relationship; determining the indicator requirements for each level of relationship based on the number of indicators and the proportion of indicators; adding the indicator requirements for each level of relationship as constraints to the fitness function; and based on the fitness function, using a global optimization algorithm to perform iterative optimization to obtain the indicator combination relationship corresponding to each level of relationship.

[0010] In one embodiment, it also includes: obtaining a multi-layer association relationship based on the target association relationship; using the multi-layer association relationship to perform target traffic detection and analysis on the actual traffic of the key network to obtain a traffic clue set; constructing a multi-level clue map based on the multi-layer association relationship and the traffic clue set; determining the tag value at each level based on the multi-level clue map, performing an overlap rate analysis on each clue to determine the overlap rate, and determining the overlap tag value based on the overlap rate; screening high-value traffic clues based on the multi-level tag value, the overlapping tag value and the preset weight to obtain the high-value traffic clues.

[0011] In one embodiment, it also includes: inputting the high-value traffic clues into a feature recognition model, performing feature recognition extraction, and obtaining a clue feature set; obtaining core features based on the influence of the clue feature set and the semantic features of the high-value traffic clues; using the core features and the mining target information to perform influence analysis to obtain clue relevance; based on the clue relevance, all core features are screened according to a preset clue quantity threshold to obtain target mining features; and fusion analysis is performed based on the target mining features to obtain the target clue recognition results.

[0012] A target deep mining system based on custom association relationships, comprising:

[0013] A mining target information obtaining module, wherein the mining target information obtaining module is used to obtain mining target information;

[0014] A pan-target detection module, configured to perform pan-target detection on the massive data in the target network based on the mined target information through big data to obtain pan-target detection results;

[0015] A custom association relationship determination module is used to perform an analysis of associated indicator parameters based on the pan-target detection results and target information mining to obtain a set of related indicators, and determine a custom association relationship based on the set of related indicators;

[0016] A key network detection module, configured to perform key network detection on the general target detection result to obtain actual traffic of the key network;

[0017] A target traffic detection and analysis module, which is used to perform target traffic detection and analysis on the actual traffic of the key network by using the custom association relationship to obtain high-value traffic clues;

[0018] The target clue identification result acquisition module is used to mine target information based on the high-value traffic clues, perform target relevance re-identification, and obtain target clue identification results.

[0019] The above-mentioned method and system for deep target mining based on custom association relationships can solve the technical problem of low target data acquisition rate and accuracy due to the excessive amount of data in the target network. Based on the mining of target information, a pan-target detection result is obtained through big data detection, and then the custom association relationship is determined. The custom association relationship is used to perform target traffic detection and analysis on the actual traffic of the key network to obtain high-value traffic clues. Finally, based on the high-value traffic clues and the mining target information, the target correlation is re-identified to obtain the target clue identification result. It can automatically realize the correlation retrieval of massive data in the target network, and improve the speed and accuracy of obtaining target traffic clues.

[0020] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flowchart of a target deep mining method based on custom association relationships is provided for this application;

[0022] Figure 2 A flowchart of determining a custom association relationship in a target deep mining method based on a custom association relationship is provided for this application;

[0023] Figure 3 This application provides a flowchart for obtaining high-value traffic leads using a target deep mining method based on custom association relationships;

[0024] Figure 4 A structural diagram of a target deep mining system based on custom association relationships is provided for this application.

[0025] Explanation of the accompanying symbols: mining target information acquisition module 1, general target detection module 2, custom association relationship determination module 3, key network detection module 4, target traffic detection and analysis module 5, target clue identification result acquisition module 6. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0027] like Figure 1 As shown, the present application provides a target deep mining method based on custom association relationships, which is applied to a target deep mining system and includes:

[0028] Step S100: obtaining mining target information;

[0029] Step S200: Based on the mining target information, perform pan-target detection on the massive data existing in the target network through big data to obtain pan-target detection results;

[0030] Specifically, mining target information is obtained, including target data range, target data type, and target data characteristics. Based on this mining target information, the target network range is determined. Then, using big data technology, the massive amount of data in the target network is detected and acquired, generating pan-target detection results. These pan-target detection results refer to data related to the mining target information. These pan-target detection results provide data support for the next step of data correlation analysis.

[0031] Step S300: Based on the general target detection results and target information, performing correlation indicator parameter analysis to obtain a related indicator set, and determining a custom correlation relationship based on the related indicator set;

[0032] like Figure 2 As shown, in one embodiment, step S300 of the present application further includes:

[0033] Step S310: performing indicator confidence analysis based on the relevant indicator set and the mining target information to determine the confidence of each indicator;

[0034] Step S320: Screen the indicators based on the confidence level of each indicator to determine the target indicator;

[0035] Step S330: performing indicator relationship analysis based on the target indicator to obtain the target indicator relationship;

[0036] Step S340: performing a custom relationship setting based on the target indicator relationship to obtain the custom association relationship.

[0037] Specifically, correlation indicator parameter analysis is performed on the general target detection results and the mining target information. This correlation indicator parameter analysis involves mining for correlation indicator parameters between the two, obtaining a set of correlation indicator parameters. The set of correlation indicator parameters is then used to determine the probability of the trustworthiness between the set of correlation indicator parameters and the mining target information. For example, if two products, AB, are represented by the number of times AB appears together, divided by the number of times A appears, the probability that a customer, having already purchased A, also purchases B. This probability is also called the confidence level. This formula expresses the degree of confidence and certainty that the customer's behavior of purchasing both A and B is related, thereby obtaining the confidence level of each indicator. A preset indicator confidence threshold is set, which can be customized based on the accuracy requirements of the target data search. The set of correlation indicators is then screened based on the indicator confidence threshold to determine the target indicator. The target indicator is the indicator with a confidence level exceeding the threshold. Then, according to the target indicator, a correlation analysis is performed based on the relevant indicator set and the mining target information to obtain the target indicator relationship. Then, according to the target indicator relationship, a custom relationship is set to obtain the custom association relationship. For example: when using data to mine customer satisfaction, the target indicator relationship includes indicators such as customer shopping list, customer shopping price, type and quantity of customer returned items, type and number of customer complaints, etc. When the custom relationship is set to customer satisfaction, the type and quantity of customer returned items and the type and number of customer complaints are the custom association relationship. By performing a confidence analysis on the relevant indicator set and the mining target information, screening the confidence of each indicator, and then obtaining the custom association relationship, the accuracy of obtaining target traffic leads can be improved.

[0038] In one embodiment, step S340 of the present application further includes:

[0039] Step S341: Setting a custom rule list, wherein the custom rule list includes indicator proportion and indicator correlation;

[0040] Step S342: Determine the number of indicators and the cross-correlation degree of each indicator based on the target indicator relationship;

[0041] Step S343: determining the indicator combination relationship according to the indicator quantity, target indicator relationship, and indicator proportion;

[0042] In one embodiment, step S343 of the present application further includes:

[0043] Step S3431: Based on the target indicator relationship, construct a fitness function for mining target information and related indicator sets;

[0044] Step S3432: Determine the relationship indicator requirements at each level based on the number of indicators and the proportion of indicators;

[0045] Step S3433: adding the relationship index requirements at all levels as constraints to the fitness function;

[0046] Step S3434: Based on the fitness function, a global optimization algorithm is used to perform iterative optimization to obtain the indicator combination relationship corresponding to each level of relationship.

[0047] Step S344: Calculating the combined correlation of the indicator combinations based on the cross-correlation of the indicators to determine the combined correlation of the indicators;

[0048] Step S345: Based on the indicator associations in the custom rule list, the indicator combination associations are matched to obtain the custom association relationship.

[0049] Specifically, a custom rule list is set, which includes an indicator ratio and an indicator correlation. The indicator ratio refers to the proportion of the indicator in the custom relationship. For example, if the custom rule list contains 100 indicators and 78 of them are associated with the custom relationship, the indicator ratio is 78%. The indicator correlation refers to the closeness of the relationship between the indicator and the custom relationship. Then, based on the target indicator relationship, the number of indicators and the cross-correlation of each indicator are determined. The number of indicators refers to the number of indicators included in the target indicator relationship. The cross-correlation refers to the degree of mutual correlation between each indicator. Based on the target indicator relationship, a fitness function is constructed using a genetic algorithm to mine target information and a set of related indicators. The fitness function is used to measure the fitness of individual indicators in the set of related indicators. Then, based on the number of indicators and the indicator ratio, multiple levels of indicator relationships are determined. The level of the indicator relationship is represented by the product of the number of indicators and the indicator ratio. The larger the product, the higher the level of the indicator relationship. For example, if the number of indicators is 10, the first-level indicator relationship is 2, the second-level indicator relationship is 5, and the third-level indicator relationship is 7, then the proportions are 20%-50% for the first-level indicator relationship, 50%-70% for the second-level indicator relationship, and 70%-100% for the third-level indicator relationship. The requirements for the indicator relationships at each level are then added to the fitness function as constraints for the individuals in the genetic algorithm. Finally, an iterative optimization algorithm is used to perform optimization. First, a population is selected, where the population refers to the relevant indicator set. Multiple individuals are randomly selected from the population to form a set, where each individual represents multiple indicators in the relevant indicator set. The next generation of individuals is selected based on fitness. For the next generation of individuals, the same position of two individuals is randomly selected and exchanged according to the crossover probability. Then, based on the principle of genetic mutation, a position in the individual is mutated according to the mutation probability. When the fitness of the optimal individual reaches a given threshold or the fitness of the optimal individual and the fitness of the population no longer increase, the algorithm ends, and the indicator combination relationship corresponding to the relationships at each level is obtained. Then, based on the cross-correlations of each indicator, a combined correlation calculation is performed on the indicator combination relationship. Specifically, the sum of the individual indicator cross-correlations is added to obtain the indicator combination correlation. Finally, the combination correlation is input into the indicator correlations in the custom rule list for matching. Specifically, the combination correlation is matched to the corresponding level in the custom rule list to obtain the custom correlation relationship. The custom correlation relationship includes multiple custom levels and corresponding indicator combinations. Using a genetic algorithm to obtain the custom correlation relationship, the optimal indicator combination within the custom range can be obtained, further improving the accuracy of obtaining target traffic leads.

[0050] Step S400: performing key network detection on the general target detection result to obtain actual traffic of the key network;

[0051] Step S500: Utilizing the user-defined association relationship, target traffic detection and analysis is performed on the actual traffic of the key network to obtain high-value traffic clues;

[0052] like Figure 3 As shown, in one embodiment, step S500 of the present application further includes:

[0053] Step S510: obtaining a multi-layer association relationship according to the target association relationship;

[0054] Step S520: Utilizing the multi-layer association relationship, target traffic detection and analysis are performed on the actual traffic of the key network to obtain a traffic clue set;

[0055] Step S530: constructing a multi-level clue map based on the multi-layer association relationship and traffic clue set;

[0056] Step S540: determining the tag value of each level based on the multi-level clue map, performing overlap rate analysis on each clue to determine the overlap rate, and determining the overlap tag value based on the overlap rate;

[0057] Step S550: Screening high-value traffic leads based on the multi-level tag values, the overlapping tag values, and preset weights to obtain the high-value traffic leads.

[0058] Specifically, the source analysis of the general-target detection results is performed to obtain multiple network source types, namely the key networks. Data is then acquired from the key networks to obtain the actual traffic of the key networks. Based on the customized association relationships, multi-layer association relationships are obtained, corresponding to multiple customized levels. The actual traffic of the key networks is then screened based on the multi-layer association relationships to obtain a set of traffic clues. A multi-level clue map is then constructed based on the multi-layer association relationships and the traffic clue sets corresponding to each layer of association relationships. Each level of tag values is determined based on the multi-level clue map. Each level of tag value refers to the indicator parameter corresponding to each level of the clue map. Based on the tag values, an overlap analysis is performed on each lead. This involves determining the number of overlaps in the indicator parameters of each lead, determining the overlap rate based on the number of overlaps, and then marking the overlaps based on the overlap rate. Finally, high-value traffic leads are screened based on the multi-level tag values, the overlap tag values, and a preset weight. The preset weight refers to the range within which traffic leads are obtained, which can be expressed as a number or a time range. High-value traffic leads are obtained. Obtaining these high-value traffic leads provides data support for the next step of obtaining targeted leads.

[0059] Step S600: Based on the high-value traffic clues, target information is mined to perform target relevance re-identification to obtain target clue identification results.

[0060] In one embodiment, step S600 of the present application further includes:

[0061] Step S610: Input the high-value traffic clue into the feature recognition model, perform feature recognition and extraction, and obtain a clue feature set;

[0062] Step S620: Obtaining core features based on the clue feature set and the semantic feature influence of the high-value traffic clue;

[0063] Step S630: performing influence analysis using the core features and the mining target information to obtain clue relevance;

[0064] Step S640: Based on the clue relevance, the core features are screened according to a preset clue quantity threshold to obtain target mining features; and fusion analysis is performed based on the target mining features to obtain the target clue recognition result.

[0065] Specifically, a feature recognition model is constructed based on the target mining information. The feature recognition model is a neural network model that can be iteratively optimized in machine learning and is obtained through supervised training on a training dataset. The high-value traffic leads are input into the feature recognition model for feature extraction to obtain a lead feature set. Based on the lead feature set, the influence of the semantic features of the high-value traffic leads is analyzed to obtain core features of high importance. An influence analysis is then performed on the core features and the target mining information. The influence analysis involves determining the degree of correlation between the core features and the target mining information to obtain lead relevance. A preset lead quantity threshold, which can be customized based on the amount of data, is set. The core features are filtered based on the lead quantity threshold to obtain the filtered core features, i.e., the target mining features. Finally, data mining is performed on the high-value traffic leads based on the target mining features to obtain the target lead identification results. This method can solve the technical problem of low target data acquisition speed and accuracy due to the large amount of data in the target network. It automatically performs relevance retrieval on the massive amount of data in the target network, thereby improving the speed and accuracy of target traffic lead acquisition.

[0066] In one embodiment, Figure 4 The system provides a target deep mining system based on custom association relationships, including: a mining target information acquisition module 1, a general target detection module 2, a custom association relationship determination module 3, a key network detection module 4, a target traffic detection and analysis module 5, and a target clue identification result acquisition module 6, wherein:

[0067] A mining target information obtaining module 1, wherein the mining target information obtaining module 1 is used to obtain mining target information;

[0068] A general target detection module 2 is configured to perform general target detection on the massive data in the target network based on the mined target information through big data to obtain a general target detection result;

[0069] A custom association relationship determination module 3 is used to perform an analysis of associated indicator parameters based on the pan-target detection results and target information mining to obtain a set of related indicators, and determine a custom association relationship based on the set of related indicators;

[0070] A key network detection module 4 is configured to perform key network detection on the general target detection result to obtain actual traffic of the key network;

[0071] Target traffic detection and analysis module 5, the target traffic detection and analysis module 5 is used to use the custom association relationship to perform target traffic detection and analysis on the actual traffic of the key network to obtain high-value traffic clues;

[0072] The target clue identification result obtaining module 6 is used to mine target information based on the high-value traffic clues, perform target relevance re-identification, and obtain target clue identification results.

[0073] In one embodiment, the system further comprises:

[0074] An indicator confidence determination module, the indicator confidence determination module is used to perform indicator confidence analysis based on the relevant indicator set and the mining target information to determine the confidence of each indicator;

[0075] A target indicator determination module, the target indicator determination module is used to screen indicators according to the confidence level of each indicator and determine the target indicator;

[0076] A target indicator relationship acquisition module, wherein the target indicator relationship acquisition module is used to perform indicator relationship analysis based on the target indicator to obtain the target indicator relationship;

[0077] A custom association relationship obtaining module is used to set a custom relationship based on the target indicator relationship to obtain the custom association relationship.

[0078] In one embodiment, the system further comprises:

[0079] A rule list setting module, wherein the rule list setting module is used to set a custom rule list, wherein the custom rule list includes an indicator proportion and an indicator correlation;

[0080] An indicator information determination module, the indicator information determination module is used to determine the number of indicators and the cross-correlation degree of each indicator according to the target indicator relationship;

[0081] An indicator combination relationship determination module, the indicator combination relationship determination module is used to determine the indicator combination relationship based on the indicator quantity, target indicator relationship, and the indicator proportion;

[0082] An indicator combination correlation determination module is used to calculate the combination correlation of the indicator combinations based on the cross-correlation of the indicators to determine the indicator combination correlation;

[0083] A custom association relationship obtaining module is used to match the indicator combination association degrees based on the indicator association degrees in the custom rule list to obtain the custom association relationship.

[0084] In one embodiment, the system further comprises:

[0085] A fitness function construction module, wherein the fitness function construction module is used to construct a fitness function for mining target information and a set of related indicators based on the target indicator relationship;

[0086] A relationship indicator requirement determination module, the relationship indicator requirement determination module is used to determine the relationship indicator requirements at each level according to the number of indicators and the proportion of indicators;

[0087] A constraint condition adding module, the constraint condition adding module is used to add the relationship index requirements at each level as constraint conditions to the fitness function;

[0088] The indicator combination relationship obtaining module is used to perform iterative optimization based on the fitness function using a global optimization algorithm to obtain the indicator combination relationship corresponding to each level of relationship.

[0089] In one embodiment, the system further comprises:

[0090] A multi-layer association relationship obtaining module, the multi-layer association relationship obtaining module is used to obtain a multi-layer association relationship according to the target association relationship;

[0091] A traffic clue set acquisition module, the traffic clue set acquisition module is used to use the multi-layer association relationship to perform target traffic detection and analysis on the actual traffic of the key network respectively to obtain a traffic clue set;

[0092] A multi-level clue map construction module, the multi-level clue map construction module is used to construct a multi-level clue map based on the multi-layer association relationship and the traffic clue set;

[0093] an overlapping mark value determination module, the overlapping mark value determination module being configured to determine the mark value of each level based on the multi-level clue map, perform an overlapping rate analysis on each clue, determine the overlapping rate, and determine the overlapping mark value based on the overlapping rate;

[0094] A high-value traffic clue acquisition module is used to screen high-value traffic clues according to the multi-level tag values, the overlapping tag values and the preset weights to obtain the high-value traffic clues.

[0095] In one embodiment, the system further comprises:

[0096] A clue feature set acquisition module, which is used to input the high-value traffic clue into a feature recognition model, perform feature recognition extraction, and obtain a clue feature set;

[0097] A core feature acquisition module, configured to acquire core features based on the influence of the clue feature set and the semantic features of the high-value traffic clue;

[0098] A clue relevance acquisition module. The clue relevance acquisition module is used to perform influence analysis using the core features and the mining target information to obtain clue relevance;

[0099] The target clue identification result acquisition module is used to screen all core features based on the clue relevance according to a preset clue quantity threshold to obtain target mining features; and perform fusion analysis based on the target mining features to obtain the target clue identification result.

[0100] In summary, this application provides a method and system for deep mining of targets based on custom association relationships, which have the following technical effects:

[0101] 1. It solves the technical problem of low target data acquisition speed and accuracy caused by excessive data volume in the target network. It can automatically perform correlation retrieval on massive data in the target network, and improve the speed and accuracy of obtaining target traffic clues.

[0102] 2. By conducting confidence analysis on the relevant indicator set and the mining target information, and screening the confidence of each indicator at the same time, and using the genetic algorithm to obtain the custom association relationship, the optimal indicator combination within the custom range can be obtained, further improving the accuracy of obtaining target traffic leads.

[0103] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0104] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A target deep mining method based on custom association relationships, characterized in that: The method is applied to a target deep mining system, and the method comprises: Obtain mining target information; Based on the mining target information, a pan-target detection is performed on the massive data existing in the target network through big data to obtain a pan-target detection result; Based on the pan-target detection results and target information, correlation indicator parameter analysis is performed to obtain a relevant indicator set, and a custom correlation relationship is determined based on the relevant indicator set; Performing key network detection on the pan-target detection results to obtain actual traffic of the key networks; Utilize the customized association relationship to perform target traffic detection and analysis on the actual traffic of the key network to obtain high-value traffic clues; Based on the high-value traffic clues, target information is mined to perform target relevance re-identification to obtain target clue identification results; Determining the custom association relationship according to the relevant indicator set includes: Based on the relevant indicator set, perform indicator confidence analysis with the mining target information to determine the confidence of each indicator; According to the confidence level of each indicator, perform indicator screening and determine the target indicator; Performing indicator relationship analysis based on the target indicators to obtain target indicator relationships; According to the target indicator relationship, a custom relationship setting is performed to obtain the custom association relationship; The method of using the customized association relationship to perform target traffic detection and analysis on the actual traffic of the key network to obtain high-value traffic clues includes: According to the custom association relationship, a multi-layer association relationship is obtained; Utilizing the multi-layer association relationship, target traffic detection and analysis are performed on the actual traffic of the key network to obtain a traffic clue set; Constructing a multi-level clue map based on the multi-layer association relationship and traffic clue set; Based on the multi-level clue map, determining the marking value of each level, performing an overlap rate analysis on each clue to determine the overlap rate, and determining the overlapping marking value based on the overlap rate; According to the tag values at each level, the overlapping tag values and the preset weights, high-value traffic leads are screened to obtain the high-value traffic leads.

2. The method according to claim 1, wherein According to the target indicator relationship, a custom relationship setting is performed to obtain the custom association relationship, including: Setting a custom rule list, wherein the custom rule list includes indicator proportion and indicator correlation; According to the target indicator relationship, determine the number of indicators and the cross-correlation degree of each indicator; Determine the indicator combination relationship based on the number of indicators, the target indicator relationship, and the indicator proportion; Calculate the combined correlation of the indicators according to the cross-correlation of the indicators to determine the combined correlation of the indicators; Based on the indicator associations in the custom rule list, the indicator combination associations are matched to obtain the custom association relationship.

3. The method according to claim 2, wherein Determine the indicator combination relationship based on the number of indicators, the target indicator relationship, and the indicator proportion, including: Based on the target indicator relationship, construct a fitness function for mining target information and related indicator sets; Determine the relationship indicator requirements at all levels based on the number of indicators and the proportion of the indicators; Adding the relationship index requirements at all levels as constraints to the fitness function; Based on the fitness function, a global optimization algorithm is used to perform iterative optimization to obtain the indicator combination relationships corresponding to the relationships at all levels.

4. The method according to claim 1, wherein Based on the high-value traffic clues, target information is mined and target relevance is re-identified to obtain target clue identification results, including: Inputting the high-value traffic clues into a feature recognition model to perform feature recognition and extraction to obtain a clue feature set; Obtaining core features based on the influence of the clue feature set and the semantic features of the high-value traffic clue; Performing an influence analysis using the core features and the mining target information to obtain clue relevance; Based on the clue relevance, all core features are screened according to a preset clue quantity threshold to obtain target mining features; and fusion analysis is performed based on the target mining features to obtain the target clue recognition result.

5. A target deep mining system based on custom association relationships, characterized in that: The system is used to perform the method according to any one of claims 1 to 4, and the system comprises: A mining target information obtaining module, wherein the mining target information obtaining module is used to obtain mining target information; A pan-target detection module, configured to perform pan-target detection on the massive data in the target network based on the mined target information through big data to obtain pan-target detection results; A custom association relationship determination module is used to perform an analysis of associated indicator parameters based on the pan-target detection results and target information mining to obtain a set of related indicators, and determine a custom association relationship based on the set of related indicators; A key network detection module, configured to perform key network detection on the general target detection result to obtain actual traffic of the key network; A target traffic detection and analysis module, which is used to perform target traffic detection and analysis on the actual traffic of the key network by using the custom association relationship to obtain high-value traffic clues; A target clue identification result acquisition module is used to mine target information based on the high-value traffic clues, perform target relevance re-identification, and obtain a target clue identification result; An indicator confidence determination module, the indicator confidence determination module is used to perform indicator confidence analysis based on the relevant indicator set and the mining target information to determine the confidence of each indicator; A target indicator determination module, the target indicator determination module is used to screen indicators according to the confidence level of each indicator and determine the target indicator; A target indicator relationship acquisition module, wherein the target indicator relationship acquisition module is used to perform indicator relationship analysis based on the target indicator to obtain the target indicator relationship; A custom association relationship obtaining module, the custom association relationship obtaining module is used to set a custom relationship according to the target indicator relationship to obtain the custom association relationship; A multi-layer association relationship obtaining module, wherein the multi-layer association relationship obtaining module is used to obtain a multi-layer association relationship according to a target association relationship; A traffic clue set acquisition module, the traffic clue set acquisition module is used to use the multi-layer association relationship to perform target traffic detection and analysis on the actual traffic of the key network respectively to obtain a traffic clue set; A multi-level clue map construction module, the multi-level clue map construction module is used to construct a multi-level clue map based on the multi-layer association relationship and the traffic clue set; an overlapping mark value determination module, the overlapping mark value determination module being configured to determine the mark value of each level based on the multi-level clue map, perform an overlapping rate analysis on each clue, determine the overlapping rate, and determine the overlapping mark value based on the overlapping rate; The high-value traffic clue acquisition module is used to screen high-value traffic clues according to the multi-level tag values, the overlapping tag values and the preset weights to obtain the high-value traffic clues.

Images