Search

CN-116701780-B - Commodity recommendation method and system based on collaborative filtering model

CN116701780BCN 116701780 BCN116701780 BCN 116701780BCN-116701780-B

Abstract

The invention discloses a commodity recommendation method and system based on a collaborative filtering model, which comprises a data collection module, a data preprocessing module, a collaborative model training module, a model verification module and a depth formulation module, wherein the data collection module is used for collecting data of users and commodities in an e-commerce platform scene, the data preprocessing module is used for preprocessing the collected data, the collaborative model training module is used for training recommendation results based on a user behavior collaborative filtering model, the model verification module is used for verifying whether the recommendation results accord with behavior trends of the users, the depth formulation module is used for carrying out secondary training and formulating recommendation of depth personalized commodity specifications, and data is analyzed by utilizing data mining and machine learning technologies so as to extract user characteristics, commodity characteristics and scene characteristics, establish a user exclusive interest preference model, and finally find commodities interested by the users from mass commodities to meet the personalized demands of the users.

Inventors

  • SUN XIAOHAN

Assignees

  • 北京全速在线科技有限公司

Dates

Publication Date
20260512
Application Date
20230607

Claims (7)

  1. 1. A commodity recommendation method based on a collaborative filtering model is characterized by comprising the following steps: Step S1, collecting user history data; S2, preprocessing the data; s3, training a collaborative filtering model; s4, performing model verification; S5, performing secondary training, and making recommendation of the depth personalized commodity specification; After outputting the result of training recommended commodity, secondarily acquiring a user behavior list, the category to which the commodity belongs, basic commodity attributes and basic user attributes, making a commodity specification training model, and analyzing the specific specification of the recommended commodity; The analysis method comprises the steps of obtaining commodities purchased by a user and different specification options under the same commodity through a user behavior list, establishing comment nodes by a single commodity, providing basic attributes of different users who obtain different purchased commodities and comments thereof in the comment nodes in the nodes, recording specifications of the commodities purchased by different users, establishing an associated attribute tag chain by taking the commodity as a center to obtain an associated attribute tag chain under the specifications of the commodities purchased by the users, repeating the analysis steps to obtain associated attribute tag chains under the specifications of all the commodities purchased by the users, establishing associated attribute tag chains for different specifications of the recommended commodities according to the current recommended commodity result, matching the associated attribute tag chains under the specifications of all the commodities purchased by the users, and selecting the commodity specification corresponding to the tag chain with the highest coincidence degree.
  2. 2. The collaborative filtering model-based commodity recommendation method according to claim 1 is characterized in that the method for collecting user history data in step S1 is specifically characterized by establishing a user behavior list, collecting historical shopping behaviors of a user on an electronic commerce platform, including long-time stay, clicking, additional purchase and purchased commodity of a user commodity page, collecting category information, price and sales trend of the last half year of the commodity, constructing user interest preference, and collecting user basic attributes, commodity basic attributes and user scene information.
  3. 3. The collaborative filtering model-based commodity recommendation method according to claim 1, wherein the method for preprocessing data in step S2 specifically comprises obtaining user history data, reducing the data range, deleting abnormal values, eliminating noise points, filling the deleted values and the deleted abnormal values, and finally normalizing the data to be in the digital signal processing category, and simultaneously converting the data into a matrix form to facilitate model training.
  4. 4. The collaborative filtering model based commodity recommendation method according to claim 1, wherein in the step S3, the collaborative filtering model training method is as follows: And training a recommended result by adopting a collaborative filtering model in combination with a specific service scene, wherein a specific calculation formula in the collaborative filtering model is as follows: ; Where i, j represents an article, u, v represents a user, The method comprises the steps of representing weight factors of commodities on i and j, inputting processed data into a model for training based on a collaborative filtering model calculation formula, outputting similarity between the commodities by the model after training, weighting similarity results between the commodities by combining specific commodity sales in order to reduce influence of head commodities, namely training based on all user behaviors to obtain similarity between the commodities.
  5. 5. The collaborative filtering model based commodity recommendation method according to claim 4, wherein the weighting factor values are defined by sales in the last half year.
  6. 6. The collaborative filtering model-based commodity recommendation method according to claim 1, wherein the model verification method is characterized in that model verification is performed by adopting an index evaluation method to ensure the rationality and accuracy of model training results, and the index evaluation method evaluates the recommendation results by adopting recall rate and popular commodity duty ratio.
  7. 7. A system for applying the collaborative filtering model-based commodity recommendation method according to any one of claims 1-6, comprising: the data collection module is used for collecting data of users and commodities in an e-commerce platform scene; The data preprocessing module is used for preprocessing the collected data; the collaborative model training module is used for training recommendation results based on the collaborative filtering model of the user behaviors; the model verification module is used for verifying whether the recommendation result accords with the behavior trend of the user; and the depth formulating module is used for performing secondary training and formulating recommendation of the depth personalized commodity specification.

Description

Commodity recommendation method and system based on collaborative filtering model Technical Field The invention relates to the field of electronic commerce, in particular to a commodity recommendation method and system based on a collaborative filtering model. Background At present, the online mall brings great convenience for the life of people, and people can purchase goods required by themselves without going out. However, with the increase of living standard, the demands of people are gradually going to the directions of diversification, enjoyment and individualization. In the face of numerous commodities, it is difficult for a user to find a commodity suitable for himself when the user's demand is not well defined. At this time, the problem can be well solved by the appearance of personalized electronic commerce recommendation, and the user can be actively provided with goods possibly liked by the user like individual shopping guide, so that the shopping efficiency of the user is improved, the personalized requirements of the user are met, and the user experience is improved. However, the conventional commodity recommendation system has low individuation degree aiming at user recommendation, is difficult to intuitively leave purchase will for users, and has low conversion rate. Therefore, a commodity recommendation method and system based on collaborative filtering model with high design practicability and high individuality are necessary. Disclosure of Invention The invention aims to provide a commodity recommendation method and system based on a collaborative filtering model, so as to solve the problems in the background technology. In order to solve the technical problems, the invention provides a commodity recommendation method based on a collaborative filtering model, which specifically comprises the following steps: Step S1, collecting user history data; S2, preprocessing the data; s3, training a collaborative filtering model; s4, performing model verification; And S5, performing secondary training, and making recommendation of the depth personalized commodity specification. According to the technical scheme, the method for collecting the user history data in the step S1 comprises the steps of establishing a user behavior list, collecting the history shopping behaviors of a user on an e-commerce platform, including long-time stay, clicking, additional purchase and purchased goods of a user commodity page, collecting category information, price and sales trend of the last half year of the goods, constructing user interest preference, and collecting user basic attributes, commodity basic attributes and user scene information. According to the technical scheme, the method for preprocessing the data in the step S2 specifically comprises the steps of obtaining user history data, reducing the data range, deleting abnormal values, eliminating noise points, filling the deleted values and the deleted abnormal values, normalizing the data to be in the digital signal processing category, and further converting the data into a matrix form so as to facilitate model training. According to the above technical solution, in step S3, the collaborative filtering model training method includes: And training a recommended result by adopting a collaborative filtering model in combination with a specific service scene, wherein a specific calculation formula in the collaborative filtering model is as follows: the method comprises the steps of i, j representing commodities, u, v representing users, W ij representing weight factors of the commodities on the i, j, inputting processed data into a model based on a collaborative filtering model calculation formula for training, outputting similarity between the commodities by the model after training, weighting similarity results between the commodities by combining specific commodity sales in order to reduce influence of head commodities, and training based on all user behaviors to obtain similarity between the commodities. According to the technical scheme, the weight factor value is defined by the influence of sales in the last half year. According to the technical scheme, the model verification method comprises the steps of performing model verification by adopting an index evaluation method so as to ensure the rationality and accuracy of a model training result, wherein the index evaluation method is mainly used for evaluating a recommended result by adopting a recall rate and a popular commodity duty ratio. According to the technical scheme, the step S5 further comprises the steps of secondarily acquiring a user behavior list, categories to which the commodities belong, commodity basic attributes and user basic attributes after a training recommendation commodity result is output, making a commodity specification training model, and analyzing specific specifications of the recommended commodities of the user; The analysis method comprises the steps of obtaining commodit