maypink / kino-service

"Development of film recommendation system" project for tinkoff fintech 2023

kino-service

Course project on the topic "Development of film recommendation system" made for Tinkoff fintech program in 2023.

Project has microservice architecture and RabbitMQ is used for communication.

Consequently, queues are build in that way. Not all of it are specified on the picture, but the main scheme.

Recommendations are being automatically refreshed every minute, for development purpose.

Also, TMDb API was used to get information about films.

kino-service's Issues

Impement microservices communication async/sync (Facade + Films services)

Implement Recommendations service

Если такой рекомендации нет -- сохраняем.
Если есть -- обновляем в ней дату.

Microservices communication #2 (async + sync)

Fix bug with first_air_date in films

Configure maintenance of actual recommendations

Set CRON or smth else to automatically delete not relevant recommendations. Also limit the number of such up to 20 for each user.

Formalise README

Make integration with autoML framework

Hello. I found you project interesting and useful. But what you think about utilizing some advanced autoML features for recommendation?

Implement user ratings in Films service

Implement Films service

Find and upload dataset related to kino service

Implement Users service

Add table with links to IMDB to Films service + add films info parsing from it

Fix exceptions processing with RabbitMQ

When data is requested from KinoFacade Service and request is not successful then RabbitMq retries it a lot of times.

1. The number of retries must be set to 3
2. Exceptions must be processed correctly
3. Also, add exception processing for incorrect tmdb_id and title+year requests

Implement Cache for recommendations

Lets say recommendations will be refreshed every 5 ratings or every day.
Also, Cache can be implemented for Film info parsed from IMDb