Television Product Recommendations and User
Portraits Based on Viewing History
Abstract: In the integration of three networks age, on the one hand, more and
more families actually have higher demand for set-top TVs and hope to quickly obtain
target resources; on the other hand, TV service providers also hope to effectively tap
user needs and senses. Resources and information of interest, realizing big data
analysis, and forming personalized product marketing and paid services.
This paper is based on user viewing history information, product information and
user package data. It uses the idea of collaborative filtering and classification to make
product recommendations for users first, and then makes recommendations for
categorized users. Also, we make suggestions for package settings.
First, the data is preprocessed. The data is cleaned and converted, and we use the
crawler captures network data to supplement the data.
For the first question, the last two sheets in the first annex are mainly used, and
user-based collaborative filtering is combined with collaborative filtering based on
television products to recommend television products for users.
For the second question, our method is based on the results of the first question
and the three attachments were combined to build a user tag system and a product
labeling system. The users and products were classified according to the tag system,
and the classified user-product category matrix was used again. We again use
collaborative filtering to achieve classification recommendations.
Analyze the existing package setup structure and split the sales item field of the
package. Using the result of the splitting of the package and the result of the user's
portrait, the association rule is found, and the rules between the portraits of different
users and the contents of the package are found, and suggestions are given for the
package recommendation.
At the end of the article, a summary of this experiment is given, which is about
harvesting and looking forward to the future.
Key words: Excel; Collaborative filtering; Web crawler; Python