Search

CN-122022970-A - Service recommendation method, device, equipment and storage medium

CN122022970ACN 122022970 ACN122022970 ACN 122022970ACN-122022970-A

Abstract

The application provides a service recommendation method, a device, equipment and a storage medium, which can be applied to the technical field of artificial intelligence. The service recommendation method comprises the steps of obtaining interaction data of a user aiming at each service type in a plurality of service types in a preset history period, inputting the interaction data of a corresponding target subinterval in a target subinterval in the plurality of history subintervals into a feature extraction module, outputting a first feature vector representing the behavior feature of the user on each service type in the target subinterval, inputting the interaction data corresponding to the target subinterval in a plurality of continuous history subintervals into a long-short-period memory network to conduct time-sequence dependency feature extraction, outputting a second feature vector, fusing and transforming the first feature vector and the second feature vector to obtain the interaction success probability of each service type in the target subinterval, and recommending the service corresponding to the target service type to the user based on the interaction success probability of each service type.

Inventors

  • WANG YUE

Assignees

  • 中国工商银行股份有限公司

Dates

Publication Date
20260512
Application Date
20260123

Claims (10)

  1. 1. A business recommendation method, the method comprising: acquiring interaction data of a user for each service type in a plurality of service types in a preset history period, wherein the preset history period is divided into a plurality of history subperiods, and each history subperiod is divided into a plurality of history subintervals; Inputting interaction data of a corresponding target subinterval in a target subinterval in the plurality of history subintervals into a feature extraction module, and outputting a first feature vector representing the behavior feature of the user on each service type in the target subinterval, wherein the target subinterval is a history subinterval closest to the current moment in the plurality of history subintervals; Inputting the interactive data corresponding to the target subinterval in the continuous multiple historical subintervals into a long-short-period memory network to perform time sequence dependency feature extraction, and outputting a second feature vector; fusing and transforming the first characteristic vector and the second characteristic vector to obtain the interaction success probability of each service type in the target subinterval; and determining a target service type from the plurality of service types based on the interaction success probability of each service type, so as to recommend the service corresponding to the target service type to the user in a future period corresponding to the target subinterval.
  2. 2. The method of claim 1, wherein the fusing and transforming the first feature vector and the second feature vector to obtain the probability of success of the interaction within the target subinterval for each of the traffic types comprises: For each service type, respectively extracting a first sub-vector representing the behavior characteristic of the user on the service type in the target subinterval from the first feature vector, and extracting a second sub-vector representing the time sequence dependent characteristic of the service type from the second feature vector; performing dimension splicing on the first sub-vector and the second sub-vector to obtain a fusion sub-vector corresponding to the service type; according to a preset service type sequence, splicing the fusion sub-vectors of the service types in sequence to form a fusion feature vector; and inputting the fusion feature vector into a probability prediction network to perform nonlinear transformation, and outputting the interaction success probability of each service type in the target subinterval.
  3. 3. The method according to claim 1, wherein the method further comprises: Determining a service weight matrix based on a historical execution sequence among the service types, wherein weight values in the service weight matrix represent the association strength of executing a first service type and subsequently executing a second service type in historical interaction; And based on the service weight matrix, carrying out weighted adjustment on the interaction success probabilities to obtain the target success probability of each service type, so as to determine the target service type from the service types according to the target success probability of each service type.
  4. 4. A method according to claim 1 or 3, characterized in that the method further comprises: For the user, sequentially taking a plurality of recommended subintervals divided in the future period as the target subintervals, and respectively determining the target service type and the corresponding interaction success probability of each recommended subinterval; According to the time arrangement sequence of the plurality of recommendation subintervals in the future period, correlating and sequencing the target service types of the recommendation subintervals and the corresponding interaction success probabilities, and generating a service recommendation list corresponding to the user; acquiring respective service recommendation lists of a plurality of users in a preset user pool, and performing global sequencing and cross-user integration on the target service types of the plurality of users based on the target service types in the service recommendation lists and the corresponding interaction success probability to generate a target recommendation list; And recommending the service of the corresponding target service type to the corresponding user respectively in each recommendation subinterval in the future period based on the target recommendation list.
  5. 5. The method of claim 4, wherein the generating the target recommendation list by globally sorting and cross-user integration of the target service types of the plurality of users based on the interaction success probabilities of the target service types in each of the service recommendation lists comprises: Comparing the interaction success probability of each user in the recommendation subinterval aiming at each recommendation subinterval, and selecting the user with the highest interaction success probability and the corresponding target service type as a recommendation item of the recommendation subinterval; and integrating the recommended items of the recommended subintervals according to the time arrangement sequence to generate the target recommended list.
  6. 6. The method according to claim 4, wherein the method further comprises: In response to recommending a service to users based on the target recommendation list, monitoring interaction response states of the users in real time; and deleting recommended items associated with the target user from the target recommendation list in response to the detection of abnormal interaction response states of the target user, and obtaining an updated recommendation list so as to execute service recommendation based on the updated recommendation list.
  7. 7. The method of claim 1, wherein the interaction data comprises a plurality of user interaction metrics, the method further comprising: According to the numerical characteristics and the distribution characteristics of the user interaction indexes, carrying out standardized processing on the interaction data to eliminate dimension differences among different user interaction indexes, and obtaining standardized feature data for generating the first feature vector and the second feature vector; The plurality of user interaction indexes at least comprise indexes representing historical behavior characteristics of a user, indexes representing time preference of the user and indexes representing user interaction risk levels.
  8. 8. A service recommendation device, the device comprising: The data acquisition module is used for acquiring interactive data of each service type in a plurality of service types in a preset history period of time of a user, wherein the preset history period of time is divided into a plurality of history subperiods, and each history subperiod of time is divided into a plurality of history subintervals; the behavior vector generation module is used for inputting interaction data of a corresponding target subinterval in a target subinterval in the plurality of history subintervals into the feature extraction module, and outputting a first feature vector representing the behavior feature of the user on each service type in the target subinterval, wherein the target subinterval is a history subinterval closest to the current moment in the plurality of history subintervals; The time sequence vector generation module is used for inputting the interactive data corresponding to the target subinterval in a plurality of continuous historical subintervals into a long-short-period memory network to extract time sequence dependency characteristics and outputting a second characteristic vector; the probability determining module is used for fusing and transforming the first characteristic vector and the second characteristic vector to obtain the interaction success probability of each service type in the target subinterval; And the service recommending module is used for determining a target service type from the plurality of service types based on the interaction success probability of each service type so as to recommend the service corresponding to the target service type to the user in a future period corresponding to the target subinterval.
  9. 9. An electronic device, comprising: one or more processors; A memory for storing one or more computer programs, Characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1-7.
  10. 10. A computer-readable storage medium, on which a computer program or instructions is stored, which, when executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.

Description

Service recommendation method, device, equipment and storage medium Technical Field The application relates to the field of artificial intelligence, in particular to a service recommendation method, device, equipment and storage medium. Background With the increasing complexity of business scenarios, personalized business recommendation has become a key technology for improving service efficiency and user experience. In the related art, service recommendation depends on a predefined static mapping rule, and a recommendation result is generated through condition comparison or attribute association. In the process of realizing the inventive concept, the related technology is found to have at least the following problems that the service recommendation is performed by depending on the static mapping relation, the recommendation accuracy is low, and the user requirements are difficult to meet. Disclosure of Invention In view of the above, the present application provides a service recommendation method, apparatus, device, medium, and program product. According to a first aspect of the present application, there is provided a service recommendation method, including obtaining interaction data of a user for each of a plurality of service types in a preset history period, wherein the preset history period is divided into a plurality of history subperiods, each of the history subperiods is divided into a plurality of history subperiods, an interaction data input feature extraction module for inputting interaction data representing a behavior feature of the user for each of the service types in a target subperiod of the plurality of history subperiods, wherein the target subperiod is a history subperiod closest to a current time in the plurality of history subperiods, performing time-dependent feature extraction for interaction data corresponding to the target subperiod in a continuous plurality of history subperiods, outputting a second feature vector, merging and transforming the first feature vector and the second feature vector to obtain an interaction success probability for each of the service types in the target subperiod, and determining a success probability for each of the service types in the target subperiod from among the plurality of target subperiods based on the interaction probability, and determining a success probability for each of the service types in the target subperiod. The second aspect of the application provides a service recommendation device, which comprises a data acquisition module, a behavior vector generation module, a probability determination module and a probability determination module, wherein the data acquisition module is used for acquiring interaction data of a user in a preset history period and aiming at each service type in a plurality of service types, the preset history period is divided into a plurality of history subperiods, each history subperiod is divided into a plurality of history subperiods, the behavior vector generation module is used for inputting interaction data of a corresponding target subperiod in a target subperiod in the plurality of history subperiods into a feature extraction module, the first feature vector is used for outputting a first feature vector representing the behavior feature of the user in the target subperiod on each service type, the target subperiod is the history subperiod closest to the current moment in the plurality of history subperiods, the time sequence vector generation module is used for extracting time sequence dependent features of the interaction data corresponding to the target subperiod in the plurality of history subperiods, the probability determination module is used for merging and transforming the first feature vector and the second feature vector to obtain a first feature vector of the corresponding target subperiod in the plurality of history subperiods, the probability determination module is used for determining success probability of the service type in the target type and the target type is used for recommending the service type from the target subperiod. A third aspect of the application provides an electronic device comprising one or more processors and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method. A fourth aspect of the application also provides a computer readable storage medium having stored thereon a computer program or instructions which when executed by a processor performs the steps of the above method. The fifth aspect of the application also provides a computer program product comprising a computer program or instructions which, when executed by a processor, carries out the steps of the method described above. According to the embodiment of the application, the interactive data in the target subinterval is input into the feature extraction module to generate the first feature vector