How to integrate my recommender systems into my website?

Question

I am building a recommendation system for my website where in the users would be given recommendations based on their choice of visiting products on my website. Similar products would be recommended to the users. However, I have am fairly done with my machine learning part and testing the results.

I am looking forward to deploy this machine learning model that generates a set of recommendations to the user on my website. I am seeking help for architectural solutions/technical solutions to deploy this model on the website. As the recommendations have to be real-time recommendations, therefore, the model has to continuously learn from real-time data of users.

Presently my user data is coming from a different server and the problem with that server is that because its a third party service, it takes more than 24 hours to get access to that data. This was fine while I needed data to build machine learning model, but for actual deployment, this server could not be leveraged.

Should I think of building my own server? If, in that case, what could be the technology that I can make use of? I am really looking for IT solutions for this.

Any guidance would be appreciated!

Answer 1

It isn't complicated. Pretty standard solution is:

1. WEB API, something like GET /api/recommendations/ .
2. Separate script that is being executed every time recommendations are requested that can communicate with caller via STDIN/STDOUT for example.

The first one is more traditional for web apps, while the second one works better for prototypig.