CHAPTER 1: INTRODUCTION - Subbarao Kambhampati

CHAPTER 1: INTRODUCTION - Subbarao Kambhampati

Filtering and Recommender Systems Content-based and Collaborative ed s a b ides l s e es f th d i o l e S Som ooneys On M

Personalization Recommenders are instances of personalization software. Personalization concerns adapting to the individual needs, interests, and preferences of each user. Includes: Recommending Filtering Predicting (e.g. form or calendar appt. completion) From a business perspective, it is viewed as part of Customer Relationship Management (CRM). Feedback & Prediction/Recommendation Traditional IR has a single userprobably working in single-shot modes Relevance feedback WEB search engines have: Working continually User profiling know this one Profile is a modelYou of the user

(and also Relevance feedback) Many users Collaborative filtering Propagate user preferences to other users Recommender Systems in Use Systems for recommending items (e.g. books, movies, CDs, web pages, newsgroup messages) to users based on examples of their preferences. Many on-line stores provide recommendations (e.g. Amazon, CDNow). Recommenders have been shown to substantially increase sales at on-line stores. Feedback Detection Non-Intrusive Click certain pages in certain order while ignore most pages. Read some clicked pages longer than some other clicked pages. Save/print certain clicked

pages. Follow some links in clicked pages to reach more pages. Buy items/Put them in wish-lists/Shopping Carts Intrusive Explicitly ask users to rate items/pages Justifying Recommendation.. Recommendation systems must justify their recommendations Even if the justification is bogus.. For search engines, the justifications are the page synopses Some recommendation algorithms are better at providing human-understandable justifications than others Content-based ones can justify in terms of classifier features.. Collaborative ones are harder-pressed other than saying

people like you seem to like this stuff In general, giving good justifications is important.. Content/Profile-based Content-based vs. Collaborative Recommendation Red Mars Found ation Jurassic Park Lost World 2001 Machine Learning User

Profile Neuromancer 2010 Difference Engine CollaborativeFiltering User Database A9 B3 C : : Z5 A B C9 : : Z10 A5

B3 C : : Z7 A B C8 : : Z Correlation Match Active User A9 B3 C . . Z5 A6 B4

C : : Z A9 B3 C : : Z5 A10 B4 C8 . . Z1 A10 B4 C8 . . Z1 Extract Recommendations

C Collaborative Filtering User Database A B C : Z 9 3 : 5 sis y l a an he n t

o i o t t rela imilar ers r o C s st s u i l c e Her ciation Asso ysis! l A na Active User A

B C 9 : : Z 10 A B C : Z 5 3 A B C 8 : : Z : 7 Correlation Match

A 9 B 3 C . . Z 5 A 6 B 4 C : : Z A B C : Z 9 3 : 5 A 10 B 4

C 8 . . Z 1 A 10 B 4 C 8 . . Z 1 Extract Recommendations C Item-User Matrix The input to the collaborative filtering algorithm is an mxn matrix where rows are items and columns are users Sort of like term-document matrix (items are terms and documents are users) Can do vector similarity between users And find who are most similar users..

Can do scalar clusters over items etc.. And find what are most correlated items Think users docs Items keywords Can think of users as vectors in the space of items (or vice versa) A Collaborative Filtering Method (think kNN regression) Weight all users with respect to similarity with the active user. How to measure similarity? Could use cosine similarity; normally pearson coefficient is used Select a subset of the users (neighbors) to use as predictors. Normalize ratings and compute a prediction from a weighted combination of the selected neighbors ratings.

Present items with highest predicted ratings as recommendations. Today Complete Filtering Discuss Das/Datar paper 3/27 Homework 2 Solns posted Midterm on Thursday in class Covers everything covered by the first two homeworks Qns?? Finding User Similarity with Person Correlation Coefficient Typically use Pearson correlation coefficient between ratings for active user, a, and another user, u. covar(r , r ) ca ,u a

u ra ru ra and ru are the ratings vectors for the m items rated by both a and u m (ra ,i ra )(ru ,i ru ) ri,j is user is rating for item j covar(ra , ru ) i 1 m m m 2 ( r r ) x ,i x r

x ,i rx i 1 m rx i 1 m Neighbor Selection For a given active user, a, select correlated users to serve as source of predictions. Standard approach is to use the most similar k users, u, based on similarity weights, wa,u Alternate approach is to include all users whose similarity weight is above a given threshold. Rating Prediction

Predict a rating, pa,i, for each item i, for active user, a, by using the k selected neighbor users, u {1,2,k}. To account for users different ratings levels, base predictions on differences from a users average rating. covar(r , r ) c Weight users ratings contribution by their n similarity to the active user. wa ,u (ru ,i ru ) u 1 p r a ,i a n ri,j is user is rating for item j | wa,u |

a a ,u ra u 1 ru u Similarity Weighting=User Similarity Typically use Pearson correlation coefficient between ratings for active user, a, and another user, u. covar(r , r ) ca ,u a u

ra ru ra and ru are the ratings vectors for the m items rated by both a and u m (ra ,i ra )(ru ,i ru ) ri,j is user is rating for item j covar(ra , ru ) i 1 m m m 2 ( r r ) x ,i x r x ,i

rx i 1 m rx i 1 m Significance Weighting Important not to trust correlations based on very few co-rated items. Include significance weights, sa,u, based on number of co-rated items, m. wa ,u sa ,u ca ,u s a ,u 1 if m 50 m if m 50

50 ca ,u covar(ra , ru ) ra ru Problems with Collaborative Filtering Cold Start: There needs to be enough other users already in the system to find a match. Sparsity: If there are many items to be recommended, even if there are many users, the user/ratings matrix is sparse, and it is hard to find users that have rated the same items. First Rater: Cannot recommend an item that has not been previously rated. New items Esoteric items Popularity Bias: Cannot recommend items to someone with unique tastes. Tends to recommend popular items. WHAT DO YOU MEAN YOU DONT CARE FOR BRITNEY SPEARS YOU DUNDERHEAD? #$%$%$&^ Content-Based Recommending

Recommendations are based on information on the content of items rather than on other users opinions. Uses machine learning algorithms to induce a profile of the users preferences from examples based on a featural description of content. Lots of systems Adapting Nave Bayes idea for Book Recommendation Vector of Bags model E.g. Books have several different fields that are all text Authors, description, A word appearing in one field is different from the same word appearing in another Can give a profile of a user in terms of words that are most predictive of what they like

Want to keep each bag differentvector of m Bags; Conditional probabilities for each word w.r.t each class and bag P(cj ) S |dm| P(cj | Book ) P(ami | cj , sm) P( Book ) m 1 i 1 Odds Ratio P(rel|example)/P(~rel| example) An example is positive if the odds ratio is > 1 Strengh of a keyword Log[P(w|rel)/P(w|~rel)] We can summarize a users profile in terms of the words that have strength above some

threshold. Advantages of Content-Based Approach No need for data on other users. No cold-start or sparsity problems. Able to recommend to users with unique tastes. Able to recommend new and unpopular items No first-rater problem. Can provide explanations of recommended items by listing content-features that caused an item to be recommended. Well-known technology The entire field of Classification Learning is at (y)our disposal! Disadvantages of Content-Based Method Requires content that can be encoded as meaningful features. Users tastes must be represented as a learnable function of these content features. Unable to exploit quality judgments of other

users. Unless these are somehow included in the content features. Content-Boosted CF - I User-ratings Vector Training Examples Content-Based Predictor Pseudo User-ratings Vector User-rated Items Unrated Items Items with Predicted Ratings Content-Boosted CF - II User Ratings Matrix Content-Based Predictor Pseudo User

Ratings Matrix Compute pseudo user ratings matrix Full matrix approximates actual full user ratings matrix Perform CF Using Pearson corr. between pseudo user-rating vectors This works better than either! Why cant the pseudo ratings be used to help content-based filtering? How about using the pseudo ratings to improve a content-based filter itself? (or how access to unlabelled examples improves accuracy) Learn a NBC classifier C0 using the few items for which we have user ratings Use C0 to predict the ratings for the rest of the items Loop Learn a new classifier C1 using all the ratings (real and predicted) Use C1 to (re)-predict the ratings for all the unknown items Until no change in ratings

With a small change, this actually works in finding a better classifier! Change: Keep the class posterior prediction (rather than just the max class) This means that each (unlabelled) entity could belong to multiple classeswith fractional membership in each We weight the counts by the membership fractions E.g. P(A=v|c) = Sum of class weights of all examples in c that have A=v divided by Sum of class weights of all examples in c This is called expectation maximization Very useful on web where you have tons of data, but very little of it is labelled Reminds you of K-means, doesnt it? (boosted) content filtering You train meI train you Co-training Small labeled data needed Suppose each instance has two parts:

x = [x1, x2] x1, x2 conditionally independent given f(x) Suppose each half can be used to classify instance f1, f2 such that f1(x1) = f2(x2) = f(x) Suppose f1, f2 are learnable f1 H1, f2 H2, [x1, x2] learning algorithms A1, A2 A1 Unlabeled Instances <[x1, x2], f1(x1)> Labeled Instances A2 ~

f2 Hypothesis Observations Can apply A1 to generate as much training data as one wants If x1 is conditionally independent of x2 / f(x), then the error in the labels produced by A1 will look like random noise to A2 !!! Thus no limit to quality of the hypothesis A2 can make Discussion of the Google News Collaborative Filtering Paper

Recently Viewed Presentations

  • Milwaukee Mathematics Partnership Spring Principals Breakfast Sharonda Harris

    Milwaukee Mathematics Partnership Spring Principals Breakfast Sharonda Harris

    Formative Assessment: A vision from the MMP Use assessments formatively throughout the teaching and learning process to: Diagnose student needs Help plan the next steps in teaching Provide the learner with feedback to improve their thinking process Help the learner...


    Systematic review by Simmons-Mackie, Raymer, Armstrong, Holland and Cherney defined communication partner training as "an intervention directed at people other than the person with aphasia with the intent of improving the language, communication, participation, and/or wellbeing of the person with...
  • Away from Supervision -

    Away from Supervision -

    What is Away From Supervision/Runaway Monthly Report? Residential Facilities must fill out a monthly report about children who are away from supervision. This report will include the following information: the number of runaway episodes, the number of episodes per gender,...
  • Section 5.23

    Section 5.23

    Three Aging Empires Section 5.23 Chapter 5: The Transformation of Eastern Europe, 1648-1740
  • I N D I A N A U

    I N D I A N A U

    Getting More for Less: A Software Distribution Model John V. Samuel, Craig A. Stewart, and Kevin J. Wilhite University Information Technology Services
  • Stock Market

    Stock Market

    A market in which securities are bought and sold: "the company was floated on the Stock Exchange". The initial offering of stocks and bonds to investors is by definition done in the primary market and subsequent trading is done in...
  • Texas Competitive Market Management and Improvement Process

    Texas Competitive Market Management and Improvement Process

    3. Examine the Texas competitive market structure and governance influences on 3rd Party access to customer data and in-home devices . E Kent/ J Schatz . 4. Understand data ownership and data privacy rules, laws, and security policies. Michele Gregg....

    Raising Expectations and Building Communities Community Investment Self Management Community Multidisciplinary Rehabilitation Tele-Care & Assistive Technologies Vocational and Education support Housing Ensure the ongoing viability of NENP (NENN & NA's) I Our Future Our Future: The Gateway Project - Middlehaven...