Development of a Software Package for Using the Goods Recommendation Model on Online Marketplaces

The work develops modules of the recommendation system, which are universal and can be integrated into various e-commerce sites. The analysis of algorithms for training cluster analysis models for implementing a recommendation model that is configured for different customer bases of different stores is carried out. The algorithm that best generalizes the models on different data is chosen. The analysis of methods of data preparation for the model is made: combining data into thematic groups. The system architecture is designed and the system requirements are described. A feasibility study is provided. The choice of technologies used for the development of the system is made and their choice is justified. The system was developed. The result of the work is the development of recommendation system modules, which are ready-made software packages and can be integrated with the store's services. The following modules have been developed: a web interface for system administration; an analytical server that analyzes data, trains the recommendation model, and validates it; a database and methods for automatically generating and updating tables; and a server for automated management of regular processes and tasks. A script has been developed for building and deploying the system in Docker containers. The work consists of 43 pages of the main text, 1 table, 19 illustrations, 1 appendice.