movie-recommender | a movie recommender | Recommender System library

I think one could think about 2 scenarios:

1) if you need very low latency, you can fill up the matrix indeed, i.e. compute all recos for all users, and store it in a key/value backend queried by your app. You can definitely predict multiple users at a time, using the one-hot encoded technique above.

2) predict on-demand by invoking the endpoint directly from the app. This is quite simpler, at the cost of a little latency.

Hope this helps.

First, your Cypher can be simplified for more efficient planning by only matching what we need, and handling the rest in the WHERE (so that filtering can possibly be done while matching)

I'm the author of this post :)

FM will "simply" fill the missing values in the recommendation matrix. What you could do is batch-predict all movies for all users, sort the results by descending score and store the top 10 results for each user in a cache, why not. That would make it easy to retrieve results in real-time from any kind of app. I suppose you would also retrain periodically to account for new user recos.

Hope this helps.

You are trying to subtract an integer from a string. The error message should include the actual line in which the error occurs. If the string is indeed an integer you can convert it using int("2")

Always try to include the full error message in a question. This makes it easiert to point to the error directly.

Edit: since you provided the full output this is the failing line:

As per Dominic's comment, upgrading kafka python to 1.3.3 fixed the issue for me.