3 ms·
At some point you presumably have to assess user-similarity in order to let similar users recommend items to each other? How's this different from e.g. user-use
by anjc 7y ago
At some point you presumably have to assess user-similarity in order to let similar users recommend items to each other? How's this different from e.g. user-user CF in terms of compiling everyone's info into a single model? I'm not criticising but rather just curious
- fjabre 7y agoI haven't found enough of a need for it at this point to make the application more useful as it stands without making great effort in training the necessary models to assess user-similarity for example. It is certainly faster to build an expert system that makes a lot of assumptions and iterate from there. There are plenty of use cases for ML and I hope to integrate it more into app some day to help automate repetitive tasks.