Search

CN-117312670-B - Recommendation generation method, device, equipment and medium based on static and dynamic data

CN117312670BCN 117312670 BCN117312670 BCN 117312670BCN-117312670-B

Abstract

The invention relates to the field of financial science and technology and discloses a recommendation generation method based on static and dynamic data, which comprises the steps of obtaining static basic information and dynamic behavior data of a target user, converting the static basic information into a static vector matrix, classifying the dynamic behavior data, calculating a frequency value and a trend change coefficient of the dynamic behavior data under each classification label, converting the dynamic behavior data into a dynamic vector matrix according to the classification labels, the frequency value and the trend change coefficient, weighting the static vector matrix and the dynamic vector matrix respectively by using a preset attention mechanism, fusing the weighted static vector matrix and the weighted dynamic vector matrix, and generating a recommendation result for the target user according to the fused vector matrix by using a pre-trained recommendation model. The invention also provides a recommendation generating device, an electric device and a medium based on the static and dynamic data. The method and the device can improve the generation accuracy of intelligent recommendation in the financial field.

Inventors

  • LIU JIAN

Assignees

  • 中国平安财产保险股份有限公司

Dates

Publication Date
20260508
Application Date
20231008

Claims (7)

  1. 1. A recommendation generation method based on static and dynamic data, the method comprising: acquiring static basic information and dynamic behavior data of a target user; Converting the static basic information into a static vector matrix; sequentially extracting data characteristics of each piece of dynamic behavior data and label characteristics of each piece of preset dynamic behavior labels, successively calculating similarity between the data characteristics of each piece of dynamic behavior data and each label characteristic, selecting a dynamic behavior label corresponding to the maximum value of the similarity as a classification label of the corresponding dynamic behavior data, and calculating a frequency value and a trend change coefficient of the dynamic behavior data under each classification label; Converting corresponding dynamic behavior data into a dynamic vector matrix according to the classification labels, the frequency values and the trend change coefficients; Weighting the static vector matrix and the dynamic vector matrix respectively by using a preset attention mechanism, and fusing the weighted static vector matrix and the weighted dynamic vector matrix to obtain a fused vector matrix; generating a result recommended to the target user according to the fusion vector matrix by utilizing a pre-trained recommendation model; The calculating the frequency value and the trend change coefficient of the dynamic behavior data under each classification label comprises the following steps: sequentially taking dynamic behavior data of the same classification mark sign as a target object to obtain a time stamp set of the target object; removing the minimum time stamp from the time stamp set, and executing random selection of two time stamps with preset times to serve as metering point time stamps; Calculating the difference of frequency values corresponding to dynamic behavior data under the same classification label when the time stamp of each two metering points is relative to the minimum time stamp; When the difference of the frequency values is greater than 0, the count is 1, when the difference of the frequency values is less than 0, the count is-1, and when the difference of the frequency values is equal to 0, the count is 0; Calculating the sum of the counts corresponding to each classification label in the preset times, setting the trend change coefficient of the dynamic behavior data under the corresponding classification label to be 1 when the sum of the counts is larger than 0, setting the trend change coefficient of the dynamic behavior data under the corresponding classification label to be 0 when the sum of the counts is equal to 0, and setting the trend change coefficient of the dynamic behavior data under the corresponding classification label to be-1 when the sum of the counts is smaller than 0.
  2. 2. The recommendation generation method based on static and dynamic data according to claim 1, wherein said converting the static basic information into a static vector matrix comprises: semantic division is carried out on the static basic information to obtain one or more static information units; extracting keywords of each static information unit, and performing fuzzy matching on the keywords and preset static information labels to obtain static information labels corresponding to each static information unit; Acquiring a value range corresponding to each static information label, and converting the corresponding static information unit into a static numerical value point by utilizing a segmentation conversion method according to the value range; Vector conversion operation is carried out on the static numerical value points corresponding to all the static information units, and the static vector matrix is obtained.
  3. 3. The recommendation generation method based on static and dynamic data according to claim 1, wherein said converting the corresponding dynamic behavior data into a dynamic vector matrix according to the classification tags, frequency values and trend change coefficients comprises: sequentially carrying out vector conversion and vector splicing on the classification label, the frequency value and the trend change coefficient corresponding to the dynamic behavior data under the same classification label to obtain a one-dimensional vector corresponding to the dynamic behavior data; And acquiring the dimension of the static vector matrix, and combining one-dimensional vectors corresponding to all dynamic behavior data according to the dimension of the static vector matrix to obtain the dynamic vector matrix.
  4. 4. A recommendation generation device based on static and dynamic data, the device comprising: The original data acquisition module is used for acquiring static basic information and dynamic behavior data of a target user; The static vector conversion module is used for converting the static basic information into a static vector matrix; the dynamic vector conversion module is used for sequentially extracting the data characteristics of each piece of dynamic behavior data and the label characteristics of each piece of preset dynamic behavior label, successively calculating the similarity between the data characteristics of each piece of dynamic behavior data and each label characteristic, selecting the dynamic behavior label corresponding to the maximum similarity as a classification label of the corresponding dynamic behavior data, calculating the frequency value and the trend change coefficient of the dynamic behavior data under each classification label, and converting the corresponding dynamic behavior data into a dynamic vector matrix according to the classification label, the frequency value and the trend change coefficient; the dynamic and static fusion module is used for respectively weighting the static vector matrix and the dynamic vector matrix by utilizing a preset attention mechanism, and fusing the weighted static vector matrix and the weighted dynamic vector matrix to obtain a fused vector matrix; The recommendation prediction module is used for generating a result recommended to the target user according to the fusion vector matrix by utilizing a pre-trained recommendation model; The calculating the frequency value and the trend change coefficient of the dynamic behavior data under each classification label comprises the following steps: sequentially taking dynamic behavior data of the same classification mark sign as a target object to obtain a time stamp set of the target object; removing the minimum time stamp from the time stamp set, and executing random selection of two time stamps with preset times to serve as metering point time stamps; Calculating the difference of frequency values corresponding to dynamic behavior data under the same classification label when the time stamp of each two metering points is relative to the minimum time stamp; When the difference of the frequency values is greater than 0, the count is 1, when the difference of the frequency values is less than 0, the count is-1, and when the difference of the frequency values is equal to 0, the count is 0; Calculating the sum of the counts corresponding to each classification label in the preset times, setting the trend change coefficient of the dynamic behavior data under the corresponding classification label to be 1 when the sum of the counts is larger than 0, setting the trend change coefficient of the dynamic behavior data under the corresponding classification label to be 0 when the sum of the counts is equal to 0, and setting the trend change coefficient of the dynamic behavior data under the corresponding classification label to be-1 when the sum of the counts is smaller than 0.
  5. 5. The recommendation-generation apparatus of claim 4 wherein the raw data-acquisition module converts the static basic information into a static vector matrix by: semantic division is carried out on the static basic information to obtain one or more static information units; extracting keywords of each static information unit, and performing fuzzy matching on the keywords and preset static information labels to obtain static information labels corresponding to each static information unit; Acquiring a value range corresponding to each static information label, and converting the corresponding static information unit into a static numerical value point by utilizing a segmentation conversion method according to the value range; Vector conversion operation is carried out on the static numerical value points corresponding to all the static information units, and the static vector matrix is obtained.
  6. 6. An electronic device, the electronic device comprising: At least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the static and dynamic data-based recommendation generation method according to any one of claims 1 to 3.
  7. 7. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the static and dynamic data based recommendation generation method according to any one of claims 1 to 3.

Description

Recommendation generation method, device, equipment and medium based on static and dynamic data Technical Field The present invention relates to the field of financial science and technology, and in particular, to a recommendation generating method, device, electronic apparatus and computer readable storage medium based on static and dynamic data. Background The intelligent recommendation system is widely applied in the financial field, such as financial product recommendation, insurance service recommendation, financial current affair information recommendation and the like of funds, bonds and the like. Conventional intelligent recommendation systems mostly employ content-based recommendation algorithms, i.e. the system will recommend content to the user that may be of interest to the user based on past behavior information of the user. In this way, since the past behavior data of the user is not further subdivided and studied, the content recommended to the user is single and solidified, the recommended content cannot be adjusted in real time following the change of the interest of the user, and the user always repeatedly contacts the content of the same type, so that the user may be bored about the recommended content, and the user experience is reduced. Disclosure of Invention The invention provides a recommendation generation method, device, electronic equipment and computer readable storage medium based on static and dynamic data, and aims to improve the generation accuracy of intelligent recommendation in the financial field. In order to achieve the above object, the present invention provides a recommendation generating method based on static and dynamic data, comprising: acquiring static basic information and dynamic behavior data of a target user; Converting the static basic information into a static vector matrix; Classifying the dynamic behavior data, determining classification labels corresponding to the dynamic behavior data, and calculating the frequency value and trend change coefficient of the dynamic behavior data under each classification label; Converting corresponding dynamic behavior data into a dynamic vector matrix according to the classification labels, the frequency values and the trend change coefficients; Weighting the static vector matrix and the dynamic vector matrix respectively by using a preset attention mechanism, and fusing the weighted static vector matrix and the weighted dynamic vector matrix to obtain a fused vector matrix; And generating a result recommended to the target user according to the fusion vector matrix by utilizing a pre-trained recommendation model. Optionally, the converting the static basic information into a static vector matrix includes: semantic division is carried out on the static basic information to obtain one or more static information units; extracting keywords of each static information unit, and performing fuzzy matching on the keywords and preset static information labels to obtain static information labels corresponding to each static information unit; Acquiring a value range corresponding to each static information label, and converting the corresponding static information unit into a static numerical value point by utilizing a segmentation conversion method according to the value range; Vector conversion operation is carried out on the static numerical value points corresponding to all the static information units, and the static vector matrix is obtained. Optionally, the classifying the dynamic behavior data to determine a classification tag corresponding to the dynamic behavior data includes: Sequentially extracting the data characteristics of each piece of dynamic behavior data and the label characteristics of each piece of preset dynamic behavior label; Successively calculating the similarity between the data characteristics of each piece of dynamic behavior data and each tag characteristic; And selecting the dynamic behavior label corresponding to the maximum similarity as a classification label of the corresponding dynamic behavior data. Optionally, the calculating the frequency value and the trend change coefficient of the dynamic behavior data under each classification label includes: sequentially taking dynamic behavior data of the same classification mark sign as a target object to obtain a time stamp set of the target object; removing the minimum time stamp from the time stamp set, and executing random selection of two time stamps with preset times to serve as metering point time stamps; Calculating the difference of frequency values corresponding to dynamic behavior data under the same classification label when the time stamp of each two metering points is relative to the minimum time stamp; When the difference of the frequency values is greater than 0, the count is 1, when the difference of the frequency values is less than 0, the count is-1, and when the difference of the frequency values is equal to 0, the count is 0; Calcu