(Feature) Handling updated rating

[BradKML]

This is relatively problematic when people's opinions change.

[gbolmier]

One can retrain the model or use an incremental model to update it, but it doesn't seem like a big issue in the sense that it shouldn't happen often and when it happens, the new rating shouldn’t be too far from the previous one and really disturb the model.

To go a bit further, people preferences and rating scales don't really change abruptly but gradually over time. There are methods that take that time dimension into account (see Recommender Systems Handbook section 5.3.3 Time-aware factor model), but not sure it's worth putting the effort.