Search

CN-122022945-A - User preference analysis method and system

CN122022945ACN 122022945 ACN122022945 ACN 122022945ACN-122022945-A

Abstract

The invention belongs to the technical field of data processing, and particularly relates to a user preference analysis method and system, wherein the method comprises the steps of obtaining user interaction data of shops in a business district, including interaction time length, specific interaction behaviors such as collection, comments and the like; the method comprises the steps of encoding interaction behaviors, comparing the interaction behaviors with standard interaction behaviors to calculate interestingness, weighting original interaction duration by taking the interestingness as weight, effectively distinguishing high-intention effective interaction from low-intention ineffective interaction, and calculating the comprehensive preference degree of a user by integrating the weighted interaction duration, interaction times and the proportion of high-quality interaction. The invention improves the accuracy and reliability of user preference analysis.

Inventors

  • OU JIARONG
  • ZHU ZHI

Assignees

  • 广东赢商网数据服务股份有限公司

Dates

Publication Date
20260512
Application Date
20260107

Claims (10)

  1. 1. A method of user preference analysis, comprising: Acquiring a shop sequence of a target shop and interaction data of each shop, wherein the interaction data comprises a plurality of users with interaction records, corresponding interaction time lengths and existing interaction behaviors; The method comprises the steps of encoding interaction behaviors of interaction records, establishing interaction feature vectors of each interaction record, establishing standard interaction feature vectors, calculating the interestingness of any user to any interaction record of any store according to the similarity of the interaction feature vectors of the interaction records of the store to the standard interaction feature vectors of the user, weighting the interaction duration of the user to the interaction record of the store by taking the interestingness of the user to the interaction record of the store as a weight coefficient, and obtaining the weighted interaction duration of the user to any interaction record of any store; The method comprises the steps of obtaining the interest degree of any user on any shop by combining the interaction times and the weighted interaction time of the user on the shop, and obtaining the comprehensive preference degree of the user on the shop by combining the interest degree of the user on the shop and the proportion of the interaction records of the user on the shop with interaction behaviors; based on the comprehensive preference degree of each user for each shop, the shops of the target business district are ordered, the thermal sequence of the shops is obtained, and the user preference in the target business district is analyzed.
  2. 2. The method of claim 1, wherein the encoding the interaction behavior of the interaction records and creating the interaction feature vector of each interaction record comprises: The method comprises the steps of coding interaction behaviors of a user, coding the a-th interaction behavior to be 1 if the a-th interaction behavior exists, coding the a-th interaction behavior to be 0 if the a-th interaction behavior does not exist, coding all interaction behaviors of each interaction record of each shop for each user through 0 and 1, and forming interaction feature vectors of corresponding interaction records according to the sequence of the interaction behaviors.
  3. 3. The method of claim 1, wherein the establishing a standard interaction feature vector comprises: setting vectors with the same order as the interaction feature vector and the feature values of 1, and marking the vectors as standard interaction feature vectors.
  4. 4. The method for analyzing user preference according to claim 1, wherein the interest level of any user in any interaction record of any shop satisfies the expression: ; In the formula, Representing the interest degree of the ith user in the kth interaction record of the (d) th shop; an interaction characteristic vector representing the kth interaction record of the ith user to the (d) th shop; representing a standard interaction feature vector; representing a vector dot product operation symbol; Representing a normalization function; a minute value is represented for avoiding the denominator of the expression to be 0.
  5. 5. The method for analyzing user preference according to claim 1, wherein the weighted interaction duration of any interaction record of any user to any shop satisfies the expression: ; In the formula, The weighted interaction duration of the ith user to the kth interaction record of the (d) th shop is represented; The interaction time length of the kth interaction record of the ith user to the (d) th shop is represented; Representing the interest degree of the ith user in the kth interaction record of the (d) th shop; indicating the number of interactions of the ith user with the (d) th shop.
  6. 6. The user preference analysis method according to claim 1, wherein the interest level of the arbitrary user in the arbitrary shop satisfies the expression: ; In the formula, Representing the interest degree of the ith user to the (d) th shop; Representing the interaction times of the ith user to the (d) th shop; representing the interaction times set of the ith user for all shops; Representing the average weighted interaction duration of the ith user on all interaction records of the (d) th shop; Representing the average weighted interaction duration of the ith user on all interaction records of all shops; representing the maximum function.
  7. 7. The method for analyzing user preference according to claim 1, wherein the step of obtaining the comprehensive preference degree of the user for the shop comprises the steps of: The personal preference degree of the ith user to the (d) th shop is calculated, and a personal preference degree set of all users to all shops is obtained; and (3) recording the ratio of the personal preference degree of the ith user to the (d) th shop to the maximum value of the personal preference degree set as the comprehensive preference degree of the ith user to the (d) th shop.
  8. 8. The method of claim 7, wherein calculating the personal preference of the ith user for the (d) th store comprises: Obtaining high-quality interaction times of each user on each shop; The method comprises the steps of obtaining the ratio of the high-quality interaction times of an ith user to a d-th shop to the interaction times of the ith user to the d-th shop, adding the ratio to the interest degree of the ith user to the d-th shop, and carrying out positive correlation normalization to obtain the comprehensive preference degree of the ith user to the d-th shop.
  9. 9. The method for analyzing user preference according to claim 8, wherein the step of obtaining the number of high-quality interactions of each user with each shop comprises: and obtaining the number of interaction records of which the interaction characteristic vector is not 0 vector in the interaction records of each user to each shop, and recording the number as high-quality interaction times of each user to the corresponding shop.
  10. 10. A user preference analysis system comprising a processor and a memory storing computer program instructions which, when executed by the processor, implement a user preference analysis method according to any of claims 1-9.

Description

User preference analysis method and system Technical Field The invention relates to the technical field of data processing. More particularly, the present invention relates to a user preference analysis method and system. Background In the scenes of business operation, accurate marketing, city planning and the like, the analysis of user preferences in a specific business district is a vital link. Through analysis of interactive data of the user on-line application platforms such as life service type and map type application on the shops, user images can be accurately depicted, the heat of the shops can be evaluated, and decision basis is provided for business optimization and quotation of the shops. In the existing user preference analysis technology, a relatively direct index is generally adopted to measure the interest degree of a user, for example, the interaction time length or interaction times of the user are used as core indexes for measuring the interest degree of the user, the longer the user stays on a shop page or the more the access times are, the thicker the interest of the user in the shop is represented, the higher the preference degree is, and the heat ranking can be performed on the shops in the shop through statistics and aggregation of a large number of related indexes of the user, so that the overall preference trend of the user is judged. However, the accuracy of the analysis result which only depends on the interaction time as a judgment basis is insufficient, for example, a user may touch a shop page by mistake due to advertisement pushing or does not actually browse due to processing other matters after opening the page, but the system still records a long invalid interaction time, the pseudo high-heat data can interfere with the analysis result, and the quality of the interaction cannot be distinguished by considering only the interaction times, for example, the preference degree and decision intention of the user performing deep interaction such as collection, comment and the like are generally higher than those of the user performing passive browsing only, so that the analysis of the real preference of the user is not accurate enough. Disclosure of Invention In order to solve the above-mentioned technical problem of insufficient accuracy of preference analysis of users, the present invention provides solutions in various aspects as follows. In a first aspect, the present invention provides a user preference analysis method, including: The method comprises the steps of obtaining a shop sequence of a target shop and interaction data of each shop, wherein the interaction data comprise a plurality of users with interaction records, corresponding interaction time lengths and existing interaction behaviors, encoding the interaction behaviors of the interaction records, establishing interaction feature vectors of the interaction records, establishing standard interaction feature vectors, calculating the interest degree of any user on any interaction record of any shop according to the similarity of the interaction feature vectors of the interaction records of the users on the shops and the standard interaction feature vectors, weighting the interaction time lengths of the interaction records of the users on the shops as weight coefficients, obtaining the weighted interaction time lengths of any interaction record of any user on any shop, combining the interaction times and the weighted interaction time lengths of the users on any shop, obtaining the interest degree of any user on any shop, combining the interest degree of the users on the shops and the proportion of the interaction records of the users on the shops, obtaining the comprehensive preference degree of the shops, sequencing the shops on the basis of the comprehensive preference degree of the users on the shops, carrying out heating power analysis on the shops in the target shop sequence, and completing the analysis of the user preference in the target shop sequence. According to the method and the device, the specific interaction behaviors such as collection and comment of the user are encoded, the similarity of the specific interaction behaviors and the standard interaction behaviors is calculated to calculate the interestingness of each interaction, and then the interestingness is used as the weight to weight the interaction duration, so that high-intention effective interaction and low-intention ineffective interaction are effectively distinguished. The invention integrates the weighted interaction time length, the interaction times and the proportion of effective interaction records, builds a comprehensive preference degree model, and improves the accuracy and reliability of business district user preference analysis. Preferably, the encoding the interaction behavior of the interaction records and establishing interaction feature vectors of each interaction record includes: The method comprises the steps of coding interac