最近在学习Coursera上的Recommender System这门课,用这篇博客来记录。该课程大约持续3~4个月。
-第一周
Introduction to Recommender Systems
Basic Model:
- Users
- Items
- Ratings
- (Community) - users+items
Non-personalized Recommender (No need of models):
- External Community Data (Best-seller; most popular; trending hot)
- Summary of Community Ratings (Best-liked)
比如: Zagat restaurant ratings, Billboard music rankings, TripAdvisor hotel ratings.
Content-Based Filtering
User ratings
∗
<script type="math/tex" id="MathJax-Element-157">*</script> Item attributes => Model
Aternative: Knowledge-based
- Item attributed from model of item space (users navigate/browse that space)
Collaborative Filtering Techniques
- User-user (user similarity; or pick people user trusts)
- Item-item (item similarity)
- Dimensionality reduction (compression -> low-dimensional matrix)
Note on Evaluation
- Accuracy of predictions
- Usefulness of recommendations
- correctness
- non-obviousness
- diversity
- Computational performance
Other approaches
Interactive recommenders:
- Critique-based
- dialog-based