Hi Neri,
A simple recommender system can use inner products and norms to measure similarity between users using cosine similarity. This approach compares users based on their rating patterns.
Cosine similarity between two users u and v is computed as:
Below is a sample code snippet explaining the above approach:
u = userVector(mask); % Ratings from target user
v = R(i,mask); % Ratings from other user
% Cosine similarity using inner product and norm
similarity(i) = dot(u, v) / (norm(u) * norm(v));
You can refer to the documentation of 'dot' function as well as 'norm' function to learn more about their syntax:
I hope it solves the query to create a simple recommender system using inner products and norms.
