Business Association Mining Method, Device, Electronic Device and Storage Medium

Abstract

The present invention provides a method, apparatus, electronic device and storage medium for business association mining, which can identify multiple services handled by different customers in customer data, obtain the interaction distance between two services by constructing a directed business graph, and then screen at least one pair of associated services of the customer according to the interaction distance. Based on the present invention, a service combination with a relatively high degree of association can be obtained, so as to optimize the function in terms of services and improve the operation efficiency.

Description

Technical Field

[0001] The present invention relates to the field of software technology, and more specifically, to a method, apparatus, electronic device, and storage medium for business association mining. Background Art

[0002] At present, when a call center handles incoming calls for consultation, it only provides services for the business that the customer consults about. This easily leads to users making multiple calls to consult related businesses.

[0003] Therefore, how to mine the associations between businesses is an urgent problem to be solved for improving business efficiency. Summary of the Invention

[0004] In view of this, to solve the above problems, the present invention provides a method, apparatus, electronic device, and storage medium for business association mining, and the technical solutions are as follows:

[0005] A method for business association mining, the method includes:

[0006] Obtain first customer service data;

[0007] Identify multiple groups of businesses in the first customer data, where one group of businesses corresponds to one customer, and one group of businesses includes multiple businesses;

[0008] For a group of businesses corresponding to a target customer, construct a directed business graph for this group of businesses. In the directed business graph, a node represents a business, and an edge connecting two nodes represents the interaction distance between the two businesses represented by the two nodes, and the interaction distance is used to represent the degree of difference;

[0009] Screen at least one pair of associated businesses of the target customer according to the interaction distance.

[0010] Preferably, the identifying multiple groups of businesses in the first customer data includes:

[0011] Obtain the text data in the first customer data;

[0012] Extract the feature vectors and position vectors of multiple phrases in the text data;

[0013] Invoke a business extraction model, and input the feature vectors and position vectors of the multiple phrases into the business extraction model. The business extraction model is pre-trained with second customer data as a sample and with the goal that the business prediction result of the sample approaches the business calibration result of the sample;

[0014] Obtain the multiple businesses output by the business extraction model, and divide the multiple businesses according to the customers they belong to to obtain multiple groups of businesses.

[0015] Preferably, the directed service graph for constructing the group of services includes:

[0016] For the first service and the second service represented by any two nodes, obtain the time difference, the round interval of the session to which they belong, and the position interval between the first service and the second service;

[0017] Determine the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval, and the interaction distance is inversely proportional to the time difference, the round interval, and the position interval respectively.

[0018] Preferably, the determining the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval includes:

[0019] Judge whether the time difference is greater than a preset threshold;

[0020] If so, determine the interaction distance between the first service and the second service as zero;

[0021] If not, calculate the interaction distance according to the round interval and the position interval.

[0022] Preferably, the method further includes:

[0023] Output multiple pairs of associated services with the smallest interaction distance.

[0024] A service association mining device, the device includes:

[0025] A data acquisition module, configured to acquire first customer service data;

[0026] A service identification module, configured to identify multiple groups of services in the first customer data, one group of services corresponding to one customer, and one group of services including multiple services;

[0027] An association mining module, configured to construct a directed service graph for a group of services corresponding to a target customer, where a node in the directed service graph represents a service, and an edge connecting two nodes represents the interaction distance between the two services represented by the two nodes, and the interaction distance is used to represent the degree of difference; screen at least one pair of associated services of the target customer according to the interaction distance.

[0028] Preferably, the service identification module is specifically configured to:

[0029] Obtain the text data in the first customer data; extract the feature vectors and position vectors of multiple phrases in the text data; retrieve the business extraction model, and input the feature vectors and position vectors of the multiple phrases into the business extraction model, where the business extraction model is pre-trained with the second customer data as a sample and with the goal that the business prediction result of the sample approaches the business calibration result of the sample; obtain multiple services output by the business extraction model, and divide the multiple services according to the affiliated customers to obtain multiple groups of services.

[0030] Preferably, the association mining module is specifically configured to:

[0031] For the first service and the second service represented by any two nodes, obtain the time difference, the round interval of the affiliated session, and the position interval between the first service and the second service; determine the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval, and the interaction distance is inversely proportional to the time difference, the round interval, and the position interval respectively.

[0032] An electronic device includes: at least one memory and at least one processor; the memory stores a program, and the processor calls the program stored in the memory, and the program is used to implement the business association mining method described in any one of the above.

[0033] A storage medium stores computer-executable instructions, and the computer-executable instructions are used to execute the business association mining method described in any one of the above.

[0034] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0035] The present invention provides a business association mining method, device, electronic device and storage medium, which can identify multiple services handled by different customers in customer data, and obtain the interaction distance between two services by constructing a directed business graph, and then screen at least a pair of associated services of the customer according to the interaction distance. Based on the present invention, a service combination with a relatively high degree of association can be obtained, so as to optimize the function by the service part and improve the operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0037] Figure 1It is the method flowchart of the business association mining method provided by the embodiment of the present invention;

[0038] Figure 2 It is a partial method flowchart of the business association mining method provided by the embodiment of the present invention;

[0039] Figure 3 It is another partial method flowchart of the business association mining method provided by the embodiment of the present invention;

[0040] Figure 4 It is yet another partial method flowchart of the business association mining method provided by the embodiment of the present invention;

[0041] Figure 5 It is the structural schematic diagram of the business association mining device provided by the embodiment of the present invention. Detailed implementation manners

[0042] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the drawings and specific implementation manners.

[0044] The embodiment of the present invention provides a business association mining method, and the method flow of this method is as Figure 1 shown, including the following steps:

[0045] S10, Obtain the first customer service data.

[0046] In the embodiment of the present invention, the call center is the contact center between the bank and customers, and a large number of customer corpora have been accumulated through methods such as telephone banking and text robots. Therefore, useful business value can be mined from these customer corpora to improve the bank's operation efficiency. The voice data in the customer corpora is converted into text through speech recognition technology, and together with the text interaction corpora of the text robots, customer data can be formed.

[0047] S20, Identify multiple groups of services in the first customer data, one group of services corresponds to one customer, and one group of services contains multiple services.

[0048] In the embodiment of the present invention, the service list involved can be extracted from the obtained first customer data through a service extraction model, and this service extraction model can be trained based on a bidirectional transform network.

[0049] Moreover, since customer data essentially belongs to the session data generated by the interaction between the customer and the call center, it is very likely that a customer has multiple session records, and these multiple session records may also involve multiple services. Therefore, when extracting services by the service extraction model in the present invention, taking the customer as the dimension, the services belonging to the same customer are grouped together.

[0050] In the specific implementation process, step S20, "identifying multiple groups of services in the first customer data", can adopt the following steps, and the method flow chart is as Figure 2 shown:

[0051] S201, obtaining the text data in the first customer data.

[0052] S202, extracting the feature vectors and position vectors of multiple phrases in the text data.

[0053] In the embodiments of the present application, the text data is segmented into corresponding multiple phrases by word segmentation, and then the multiple phrases are filtered through preprocessing such as removing stop words. For each phrase obtained through filtering, the text feature conversion can be performed through a word vector model in the specified customer domain to obtain a vectorized text representation, that is, the feature vector of the phrase.

[0054] Furthermore, for each phrase, its relative position relationship with other phrases in the sentence can be determined according to its position in the sentence to which it belongs, and then the position vector of the phrase is obtained through vectorization.

[0055] S203, retrieving the service extraction model, and inputting the feature vectors and position vectors of multiple phrases into the service extraction model. The service extraction model is pre-trained with the second customer data as a sample and with the goal that the service prediction result of the sample approaches the service calibration result of the sample.

[0056] In the embodiments of the present invention, for the customer data used as the training sample, it also comes from the customer corpus of the call center. On the one hand, the phrases that can represent services are manually calibrated; on the other hand, the sample is subjected to text processing, including word segmentation and preprocessing, and then the text feature conversion is performed through a word vector model in the specified customer domain to obtain the feature vectors of the corresponding phrases. Similarly, based on the position of the phrase in the sentence to which it belongs, the position vector of the corresponding phrase is obtained through vectorization.

[0057] Construct a bidirectional transform network, and input the feature vectors and position vectors of the phrases in the sample into the bidirectional transform network to achieve: the bidirectional transform network performs business prediction based on the feature vectors and position vectors of the phrases in the sample, and adjusts the weights of each layer of the network inside it with the goal of continuously approaching the business represented by the labeled phrases in the sample (i.e., the business calibration result). Through multiple iterations, the bidirectional transform network after the training is completed is used as the business extraction model.

[0058] S204. Obtain multiple services output by the business extraction model, and divide the multiple services according to the affiliated customers to obtain multiple groups of services.

[0059] S30. For a group of services corresponding to the target customer, construct a directed business graph of this group of services. In the directed business graph, a node represents a service, and the edge connecting two nodes represents the interaction distance between the two services represented by the two nodes. The interaction distance is used to represent the degree of difference.

[0060] In the embodiment of the present invention, for multiple services corresponding to the same customer, first obtain the phrases representing each service, and then sort the multiple services according to the time sequence of the appearance of the phrases. Create the nodes of the directed business graph in the sorting result, where one node corresponds to one service, and then determine the edges of the directed business graph considering the degree of difference between any two services.

[0061] In the specific implementation process, step S30 "Construct the directed business graph of this group of services" can adopt the following steps. The method flow chart is as Figure 3 shown:

[0062] S301. For the first service and the second service represented by any two nodes, obtain the time difference, the round interval of the affiliated session, and the position interval between the first service and the second service.

[0063] In the embodiment of the present invention, the degree of difference between two services can be considered from three aspects, including the time difference, the round interval of the affiliated session, and the position interval. Of course, in the embodiment of the present invention, the interaction distance is calculated based on the three. It can be understood that using any one or two of them to calculate the interaction distance is also within the protection scope of the embodiment of the present invention.

[0064] It should be noted that for the first service and the second service, first obtain the phrase representing the first service (hereinafter referred to as the first phrase) and the phrase representing the second service (hereinafter referred to as the second phrase). For the time difference between the first service and the second service, it can be calculated according to the time when the first phrase and the second phrase appear in the first customer data, and the difference between the two times is the time difference. For the round interval between the first service and the second service, it can be calculated according to the rounds of the conversations in which the first phrase and the second phrase are located in the first customer data, and the difference between the two rounds is the round interval. For the position interval between the first service and the second service, the number of characters between the first phrase and the second phrase in the first customer data can be used as the position interval.

[0065] S302. Determine the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval. The interaction distance is inversely proportional to the time difference, the round interval, and the position interval respectively.

[0066] In the embodiments of the present invention, the distance calculation methods corresponding to the time difference, the round interval, and the position interval can be set respectively. The present invention does not limit the distance calculation methods. Taking the time difference as an example, the distance corresponding to the time difference is inversely proportional to the value of the time difference, that is, the greater the time difference, the greater the distance corresponding to the time.

[0067] Finally, calculate the interaction distance by comprehensively considering the distances corresponding to the time difference, the round interval, and the position interval respectively. For example, the weights of the three in calculating the interaction distance can be considered, and the sum of the products of the weights of the three and the distances can be used as the interaction distance between the first service and the second service.

[0068] In the specific implementation process, step S302 "Determine the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval" can adopt the following steps. The method flow chart is as Figure 4 shown:

[0069] S3021. Determine whether the time difference is greater than a preset threshold; if so, execute step S3022; if not, execute step S3023.

[0070] In the embodiments of the present invention, if the time between the first service and the second service is relatively large, the possibility of their service association will decrease. Therefore, for a group of services with a time difference greater than the corresponding threshold, it can be considered that they are not associated.

[0071] S3022. Determine the interaction distance between the first service and the second service as zero.

[0072] S3023. Calculate the interaction distance according to the round interval and the position interval.

[0073] In the embodiments of the present invention, the interaction distance can be calculated according to the distances corresponding to the round interval and the position interval respectively. Specifically, weights can be set for the round interval and the position interval respectively, and the sum of the product of the round interval and its weight and the product of the position interval and its weight is used as the interaction distance.

[0074] S40. Screen at least one pair of associated services of the target customer according to the interaction distance.

[0075] In the embodiments of the present invention, services have a chronological order in the directed service graph, but their interaction distance is a relative quantity and has no direction. When outputting associated services, at least one pair of services whose interaction distance meets the corresponding conditions is used as the associated services. For example, a certain number of pairs of services with the smallest interaction distance are used as the associated services. Here, one pair means two, one pair of services means two services, and one pair of associated services means two associated services.

[0076] In addition, the multiple pairs of associated services with the smallest interaction distance can be output to the business department to prompt the business department to handle transaction associations, so that the call center recommends to the customer the service combination for handling associations.

[0077] The service association mining method provided by the embodiments of the present invention can identify multiple services handled by different customers in customer data, obtain the interaction distance between two services by constructing a directed service graph, and then screen at least one pair of associated services of the customer according to the interaction distance. Based on the present invention, a service combination with a relatively high degree of association can be obtained, so that the business part can optimize the functions and improve the operation efficiency.

[0078] Based on the service association mining method provided in the above embodiments, the embodiments of the present invention provide an apparatus for executing the above service association mining method. The structural schematic diagram of the apparatus is as Figure 5 shown and includes:

[0079] A data acquisition module 10, configured to acquire first customer service data;

[0080] A service identification module 20, configured to identify multiple groups of services in the first customer data, one group of services corresponding to one customer, and one group of services including multiple services;

[0081] An association mining module 30, configured to construct a directed service graph of a group of services corresponding to a target customer. In the directed service graph, one node represents one service, and the edge connecting two nodes represents the interaction distance between the two services represented by the two nodes. The interaction distance is used to represent the degree of difference; screen at least one pair of associated services of the target customer according to the interaction distance.

[0082] Optionally, the service identification module 20 is specifically configured to:

[0083] Obtain the text data in the first customer data; extract the feature vectors and position vectors of multiple phrases in the text data; retrieve the business extraction model, and input the feature vectors and position vectors of multiple phrases into the business extraction model. The business extraction model is pre-trained with the second customer data as a sample and with the goal that the business prediction result of the sample approaches the business calibration result of the sample; obtain multiple services output by the business extraction model, and divide the multiple services according to the affiliated customers to obtain multiple groups of services.

[0084] Optionally, the association mining module 30 is specifically configured to:

[0085] For the first service and the second service represented by any two nodes, obtain the time difference, the round interval of the affiliated session, and the position interval between the first service and the second service; determine the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval. The interaction distance is inversely proportional to the time difference, the round interval, and the position interval respectively.

[0086] Optionally, the association mining module 30 for determining the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval is specifically configured to:

[0087] Judge whether the time difference is greater than a preset threshold; if so, determine the interaction distance between the first service and the second service to be zero; if not, calculate the interaction distance according to the round interval and the position interval.

[0088] Optionally, the association mining module 30 is further configured to:

[0089] Output multiple pairs of associated services with the smallest interaction distance.

[0090] The service association mining device provided by the embodiments of the present invention can identify multiple services handled by different customers in the customer data, obtain the interaction distance between two services by constructing a directed service graph, and then screen at least one pair of associated services of the customer according to the interaction distance. Based on the present invention, a service combination with a relatively high degree of association can be obtained, so as to optimize the function in the service part and improve the operation efficiency.

[0091] The embodiments of the present invention further provide an electronic device, including: at least one memory and at least one processor; the memory stores a program, and the processor calls the program stored in the memory, and the program is used to implement the service association mining method of any one of the above.

[0092] The embodiments of the present invention further provide a storage medium, in which computer-executable instructions are stored, and the computer-executable instructions are used to execute the service association mining method of any one of the above.

[0093] The above has introduced in detail a method, apparatus, electronic device, and storage medium for business association mining provided by the present invention. Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation on the present invention.

[0094] It should be noted that the various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the various embodiments can be referred to each other. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0095] It should also be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising", "including", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device that includes a series of elements includes not only those elements inherent to the process, method, article, or device, but also other elements inherent to these process, method, article, or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article, or device that includes the said element.

[0096] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for mining business associations, characterized in that, the method includes: Obtain the first customer service data; Identify multiple groups of services in the first customer data, where one group of services corresponds to one customer, and one group of services contains multiple services; For a group of services corresponding to the target customer, construct a directed service graph for this group of services. In the directed service graph, a node represents a service, and an edge connecting two nodes represents the interaction distance between the two services represented by the two nodes. The interaction distance is used to represent the degree of difference; Filter at least one pair of associated services of the target customer according to the interaction distance; Among them, constructing the directed service graph for this group of services includes: For the first service and the second service represented by any two nodes, obtain the time difference, the round interval of the session to which they belong, and the position interval between the first service and the second service; Determine the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval. The interaction distance is inversely proportional to the time difference, the round interval, and the position interval respectively; Among them, for the first service and the second service, obtain the first phrase representing the first service and the second phrase representing the second service. For the time difference between the first service and the second service, calculate it according to the time when the first phrase and the second phrase appear in the first customer data. The difference between the two times is the time difference. For the round interval between the first service and the second service, calculate it according to the rounds of the sessions where the first phrase and the second phrase are located in the first customer data. The difference between the two rounds is the round interval. For the position interval between the first service and the second service, use the number of characters between the first phrase and the second phrase in the first customer data as the position interval.

2. The method according to claim 1, characterized in that, the identifying multiple groups of services in the first customer data includes: Obtain the text data in the first customer data; Extract the feature vectors and position vectors of multiple phrases in the text data; Invoke the service extraction model, and input the feature vectors and position vectors of the multiple phrases into the service extraction model. The service extraction model is pre-trained with the second customer data as a sample and with the goal that the service prediction result of the sample approaches the service calibration result of the sample; Obtain multiple services output by the service extraction model, and divide the multiple services according to the customers to which they belong to obtain multiple groups of services.

3. The method according to claim 1, characterized in that, the determining the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval includes: Judge whether the time difference is greater than a preset threshold; If so, determine the interaction distance between the first service and the second service to be zero; If not, calculate the interaction distance according to the round interval and the position interval.

4. The method according to claim 1, characterized in that, the method further includes: Output multiple pairs of associated services with the smallest interaction distance.

5. A device for mining business associations, characterized in that, The device includes: a data acquisition module for acquiring first customer service data; a service identification module for identifying multiple groups of services in the first customer data, where one group of services corresponds to one customer, and one group of services contains multiple services; an association mining module for constructing a directed service graph for a group of services corresponding to a target customer. In the directed service graph, a node represents a service, and an edge connecting two nodes represents the interaction distance between the two services represented by the two nodes. The interaction distance is used to represent the degree of difference; screening at least one pair of associated services of the target customer according to the interaction distance; wherein, the association mining module is specifically used for: for a first service and a second service represented by any two nodes, obtaining the time difference, the round interval of the session to which they belong, and the position interval between the first service and the second service; determining the interaction distance between the first service and the second service according to the time difference, the round interval, and the position interval, and the interaction distance is inversely proportional to the time difference, the round interval, and the position interval respectively; wherein, for a first service and a second service, a first phrase representing the first service and a second phrase representing the second service are obtained. For the time difference between the first service and the second service, it is calculated according to the times when the first phrase and the second phrase appear in the first customer data, and the difference between the two times is the time difference. For the round interval between the first service and the second service, it is calculated according to the rounds of the sessions in which the first phrase and the second phrase are located in the first customer data, and the difference between the two rounds is the round interval. For the position interval between the first service and the second service, the number of characters between the first phrase and the second phrase in the first customer data is used as the position interval.

6. The device according to claim 5, wherein, the service identification module is specifically used for: obtaining the text data in the first customer data; extracting the feature vectors and position vectors of multiple phrases in the text data; invoking a service extraction model and inputting the feature vectors and position vectors of the multiple phrases into the service extraction model. The service extraction model is pre-trained with second customer data as a sample and with the goal that the service prediction result of the sample approaches the service calibration result of the sample; obtaining multiple services output by the service extraction model and dividing the multiple services according to the customers to which they belong to obtain multiple groups of services.

7. An electronic device, wherein, it includes: at least one memory and at least one processor; the memory stores a program, and the processor invokes the program stored in the memory, and the program is used to implement the service association mining method according to any one of claims 1-4.

8. A storage medium, wherein, computer-executable instructions are stored in the storage medium, and the computer-executable instructions are used to execute the service association mining method according to any one of claims 1-4.

Images