Prediction Modeling for Personalization Recommender Systems Bamshad Mobasher
Prediction Modeling for Personalization & Recommender Systems Bamshad Mobasher De. Paul University
What Is Prediction? i Prediction is similar to classification 4 First, construct a model 4 Second, use model to predict unknown value i Prediction is different from classification 4 Classification refers to predicting categorical class label (e. g. , “yes”, “no”) 4 Prediction models are used to predict values of a numeric target attribute h They can be thought of as continuous-valued functions i Major method for prediction is regression 4 Linear and multiple regression 4 Non-linear regression 4 K-Nearest-Neighbor i Most common application domains: 4 Personalization & recommender systems, credit scoring, predict customer loyalty, etc. 2
Personalization i The Problem 4 Dynamically serve customized content (books, movies, pages, products, tags, etc. ) to users based on their profiles, preferences, or expected interests i Why we need it? 4 Information spaces are becoming much more complex for user to navigate (huge online repositories, social networks, mobile applications, blogs, …. ) 4 For businesses: need to grow customer loyalty / increase sales 4 Industry Research: successful online retailers are generating as much as 35% of their business from recommendations i Recommender Systems the most common type of personalization systems 3
Recommender Systems: Common Approaches i Collaborative Filtering 4 Give recommendations to a user based on preferences of “similar” users 4 Preferences on items may be explicit or implicit 4 Includes recommendation based on social / collaborative content i Content-Based Filtering 4 Give recommendations to a user based on items with “similar” content in the user’s profile i Hybrid Approaches 4
The Recommendation Task i Basic formulation as a prediction problem Given a profile Pu for a user u, and a target item it, predict the preference score of user u on item it i Typically, the profile Pu contains preference scores by u on some other items, {i 1, …, ik} different from it 4 preference scores on i 1, …, ik may have been obtained explicitly (e. g. , movie ratings) or implicitly (e. g. , time spent on a product page or a news article) 5
Example: Recommender Systems i. Content-based recommenders 4 Predictions for unseen (target) items are computed based on their similarity (in terms of content) to items in the user profile. 4 E. g. , user profile Pu contains recommend highly: and recommend “mildly”: 6
Content-Based Recommender Systems 7
Content-Based Recommenders : : more examples i Music recommendations i Play list generation Example: Pandora 8
Content representation & item similarities i Represent items as vectors over features 4 Features may be items attributes, keywords, tags, etc. 4 Often items are represented a keyword vectors based on textual descriptions with TFx. IDF or other weighting approaches h. Has the advantage of being applicable to any type of item (images, products, news stories, tweets) as long as a textual description is available or can be constructed 4 Items (and users) can then be compared using standard vector space similarity measures
Content-based recommendation i
Collaborative Recommender Systems i Collaborative filtering recommenders 4 Predictions for unseen (target) items are computed based the other users’ with similar interest scores on items in user u’s profile hi. e. users with similar tastes (aka “nearest neighbors”) hrequires computing correlations between user u and other users according to interest scores or ratings hk-nearest-neighbor (knn) strategy Can we predict Karen’s rating on the unseen item Independence Day? 11
Collaborative Recommender Systems Many examples in real world applications Don’t need a representation for items, but compare user profiles instead 12
Collaborative Filtering: Measuring Similarities i Pearson Correlation 4 weight by degree of correlation between user U and user J Average rating of user J on all items. 4 1 means very similar, 0 means no correlation, -1 means dissimilar 4 Works well in case of user ratings (where there is at least a range of 1 -5) 4 Not always possible (in some situations we may only have implicit binary values, e. g. , whether a user did or did not select a document) 4 Alternatively, a variety of distance or similarity measures can be used 13
Collaborative Filtering: Making Predictions i In practice a more sophisticated approach is used to generate the predictions based on the nearest neighbors i To generate predictions for a target user a on an item i: 4 = mean rating for user a 4 u 1, …, uk are the k-nearest-neighbors to a 4 ru, i = rating of user u on item I 4 sim(a, u) = Pearson correlation between a and u i This is a weighted average of deviations from the neighbors’ mean ratings (and closer neighbors count more) 14
Example: User-Based Collaborative Filtering prediction Predictions for Karen on Indep. Day based on the K nearest neighbors Correlation to Karen 15
Possible Interesting Project Ideas i Build a content-based recommender for 4 News stories (requires basic text processing and indexing of documents) 4 Blog posts, tweets 4 Music (based on features such as genre, artist, etc. ) i Build a collaborative or social recommender 4 Movies (using movie ratings), e. g. , movielens. org 4 Music, e. g. , pandora. com, last. fm h. Recommend songs or albums based on collaborative ratings, tags, etc. hrecommend whole playlists based on playlists from other users 4 Recommend users (other raters, friends, followers, etc. ), based similar interests 16
Prediction Modeling for Personalization & Recommender Systems Bamshad Mobasher De. Paul University
- Slides: 17