Information Filtering SI650 : Information Retrieva l

1. Information FilteringSI650: Information Retrieval Winter 2010 School of Information University of Michigan Many slides are from Prof. ChengXiangZhai’s lecture

2. Lecture Plan • Filtering vs. Retrieval • Content-based filtering (adaptive filtering) • Collaborative filtering (recommender systems)

3. Short vs. Long Term Info Need • Short-term information need (Ad hoc retrieval) • “Temporary need”, e.g., info about used cars • Information source is relatively static • User “pulls” information • Application example: library search, Web search • Long-term information need (Filtering) • “Stable need”, e.g., new data mining algorithms • Information source is dynamic • System “pushes” information to user • Applications: news filter, recommender systems

4. Examples of Information Filtering • News filtering • Email filtering • Movie/book recommenders • Literature recommenders • And many others …

5. Content-based Filtering vs. Collaborative Filtering • Basic filtering question: Will user U like item X? • Two different ways of answering it • Look at what U likes • Look at who likes X • Can be combined characterize X content-based filtering characterize U  collaborative filtering

6. Content-based filtering is also called 1.“Adaptive Information Filtering” in TREC2. “Selective Dissemination of Information (SDI)in Library & Information Science Collaborative filtering is also called “Recommender Systems”

7. Why Bother? Why do we need recommendation? Why should amazon provide recommendation? Why does Netflix spend millions to improve their recommendation algorithm? Why can’t we rely on a search engine for all the recommendation?

8. The Long Tail Need a way to discover items that match our interests & tastes among tens or hundreds of thousands Chris Anderson, ‘The Long Tail’, Wired, Issue 12.10 - October 2004

9. Part I: Adaptive Filtering

10. Adaptive Information Filtering • Stable & long term interest, dynamic info. source • System must make a delivery decision immediately as a document “arrives” my interest: Filtering System …

11. AIF vs. Retrieval & Categorization • Like retrieval over a dynamic stream of docs, but (global) ranking is impossible • Like online binary categorization, but with no initial training data and with limited feedback • Typically evaluated with a utility function • Each delivered doc gets a utility value • Good doc gets a positive value • Bad doc gets a negative value • E.g., Utility = 3* #good - 2 *#bad (linear utility)

12. Accumulated Docs Feedback Learning A Typical AIF System User profile text Initialization Accepted Docs Binary Classifier ... User User Interest Profile Doc Source utility func.

13. Three Basic Problems in AIF • Making filtering decision (Binary classifier) • Doc text, profile text  yes/no • Initialization • Initialize the filter based on only the profile text or very few examples • Learning from • Limited relevance judgments (only on “yes” docs) • Accumulated documents • All trying to maximize the utility

14. Major Approaches to AIF • “Extended” retrieval systems • “Reuse” retrieval techniques to score documents • Use a score threshold for filtering decision • Learn to improve scoring with traditional feedback • New approaches to threshold setting and learning • “Modified” categorization systems • Adapt to binary, unbalanced categorization • New approaches to initialization • Train with “censored” training examples

15. Difference between Ranking, Classification, and Recommendation What is the query? • Ranking: no query • Classification: labeled data (a.k.a, training data) • Recommendation: user + items What is the target? • Ranking: order of objects • Classification: labels of objects • Recommendation: Score? Label? Order?

16. A General Vector-Space Approach no doc vector Utility Evaluation Scoring Thresholding yes profile vector threshold Vector Learning Threshold Learning Feedback Information

17. Difficulties in Threshold Learning • Censored data • Little/none labeled data • Scoring bias due to vector learning • Exploration vs. Exploitation 36.5 R 33.4 N 32.1 R 29.9 ? 27.3 ? … ... =30.0

18. Threshold Setting in Extended Retrieval Systems • Utility-independentapproaches (generally not working well, not covered in this lecture) • Indirect(linear) utility optimization • Logistic regression (score  prob. of relevance) • Direct utility optimization • Empirical utility optimization • Expected utility optimization given score distributions • All try to learn the optimal threshold

19. Logistic Regression (Robertson & Walker. 00) • General idea: convert score of D to p(R|D) • Fit the model using feedback data • Linear utility is optimized with a fixed prob. cutoff • But, • Possibly incorrect parametric assumptions • No positive examples initially • Censored data and limited positive feedback • Doesn’t address the issue of exploration

20. Direct Utility Optimization • Given • A utility function U(CR+ ,CR- ,CN+ ,CN-) • Training data D={<si, {R,N,?}>} • Formulate utility as a function of the threshold and training data: U=F(,D) • Choose the threshold by optimizing F(,D), i.e.,

21. Empirical Utility Optimization • Basic idea • Compute the utility on the training data for each candidate threshold (score of a training doc) • Choose the threshold that gives the maximum utility • Difficulty: Biased training sample! • We can only get an upper bound for the true optimal threshold. • Solutions: • Heuristic adjustment(lowering) of threshold • Lead to “beta-gamma threshold learning”

22. Score Distribution Approaches( Aramptzis & Hameren 01; Zhang & Callan 01) • Assume generative model of scores p(s|R), p(s|N) • Estimate the model with training data • Find the threshold by optimizing the expected utility under the estimated model • Specific methods differ in the way of defining and estimating the scoring distributions

23. Gaussian-Exponential Distributions • P(s|R) ~ N(,2) p(s-s0|N) ~ E() (From Zhang & Callan 2001)

24. Skip Score Distribution Approaches (cont.) • Pros • Principled approach • Arbitrary utility • Empirically effective • Cons • May be sensitive to the scoring function • Exploration not addressed

25. Part II: Collaborative Filtering

26. What is Collaborative Filtering (CF)? • Making filtering decisions for an individual user based on the judgments of other users • Inferring individual’s interest/preferences from that of other similar users • General idea • Given a user u, find similar users {u1, …, um} • Predict u’s preferences based on the preferences of u1, …, um

27. CF: Assumptions • Users with a common interest will have similar preferences • Users with similar preferences probably share the same interest • Examples • “interest is IR” “favor SIGIR papers” • “favor SIGIR papers” “interest is IR” • Sufficiently large number of user preferences are available

28. CF: Intuitions • User similarity (Paul Resnickvs. Rahul Sami) • If Paul liked the paper, Rahul will like the paper • ? If Paul liked the movie, Rahul will like the movie • Suppose Paul and Rahul viewed similar movies in the past six months … • Item similarity • Since 90% of those who liked Star Wars also liked Independence Day, and, you liked Star Wars • You may also like Independence Day The content of items “didn’t matter”!

29. Rating-based vs. Preference-based • Rating-based: User’s preferences are encoded using numerical ratings on items • Complete ordering • Absolute values can be meaningful • But, values must be normalized to combine • Preference-based: User’s preferences are represented by partial ordering of items (Learning to Rank!) • Partial ordering • Easier to exploit implicit preferences

30. A Formal Framework for Rating Objects: O o1 o2 … oj … on 3 1.5 …. … 2 2 1 3 Users: U Xij=f(ui,oj)=? u1 u2 … ui ... um The task • Assume known f values for some (u,o)’s • Predict f values for other (u,o)’s • Essentially function approximation, like other learning problems ? Unknown function f: U x O R

31. Where are the intuitions? • Similar users have similar preferences • If u  u’, then for all o’s, f(u,o)  f(u’,o) • Similar objects have similar user preferences • If o  o’, then for all u’s, f(u,o)  f(u,o’) • In general, f is “locally constant” • If u  u’ and o  o’, then f(u,o)  f(u’,o’) • “Local smoothness” makes it possible to predict unknown values by interpolation or extrapolation • What does “local” mean?

32. Two Groups of Approaches • Memory-based approaches • f(u, o) = g(u)(o)  g(u’)(o) if u  u’ (g =preference function) • Find “neighbors” of u and combine g(u’)(o)’s • Model-based approaches (not covered) • Assume structures/model: object cluster, user cluster, f’ defined on clusters • f(u, o) = f’(cu, co) • Estimation & Probabilistic inference

33. Memory-based Approaches (Resnicket al. 94) • General ideas: • xij: rating of object j by user i • ni: average rating of all objects by user i • Normalized ratings: • Memory-based prediction • Specific approaches differ in w(a, i) -- the distance/similarity between user a and i

34. User Similarity Measures • Pearson correlation coefficient (sum over commonly rated items) • Cosine measure • Many other possibilities!

35. Improving User Similarity Measures (Breese et al. 98) • Dealing with missing values: default ratings • Inverse User Frequency (IUF): similar to IDF • Case Amplification: use w(a, i)p, e.g., p=2.5

36. A Network Interpretation If your neighbors love it, you are likely to love it Closer neighbors make a larger impact Measure the closeness by the items you and her liked/disliked

37. Network based approaches No clear notion of “distance” now, but the “scores” are estimated based on some propagation process. Score propagation based on random walk Two approaches: • Starting from the user, how likely/how long can I reach the object? • Starting from the object, how likely/how long can I hit the user?

38. Random Walk and Hitting Time P = 0.3 0.3 k A i 0.7 P = 0.7 j Hitting Time • TA: the first time that the random walk is at a vertex in A Mean Hitting Time • hiA: expectation of TA given that the walk starts from vertex i

39. Computing Hitting Time hiA = 0.7 hjA + 0.3 hkA + 1 h = 0 • TA: the first time that the random walk is at a vertex in A 0.7 k A i • hiA: expectation of TA given that the walk starting from vertex i 0.7 Apparently, hiA = 0 for those j Iterative Computation

40. Bipartite Graph and Hitting Time • Bipartite Graph: • Edges between V1 and V2 • No edge inside V1 or V2 • Edges are weighted • e.g., V1 = query; V2 = Url 5 5 5 A A A 4 4 4 V1 V1 V1 0.4 0.4 0.4 V2 V2 V2 k 0.7 0.7 0.7 7 7 7 1 1 1 i i i w(i, j) = 3 j j j Expected proximity of query i to the query A : hitting time of i  A, hiA • convert to a directed graph, even collapse one group

41. Example: Query Suggestion Welcome to the hotel california

42. Generating Query Suggestion using Click Graph (Mei et al. 08) freq(q, url) • Construct a (kNN) subgraph from the query log data (of a predefined number of queries/urls) • Compute transition probabilities p(i  j) • Compute hitting time hiA • Rank candidate queries using hiA Query Url 300 A www.aa.com aa 15 www.theaa.com/travelwatch/planner_main.jsp mexiana american airline en.wikipedia.org/wiki/Mexicana

43. Result: Query Suggestion Query = friends

44. Summary • Filtering is “easy” • The user’s expectation is low • Any recommendation is better than none • Making it practically useful • Filtering is “hard” • Must make a binary decision, though ranking is also possible • Data sparseness • “Cold start”

45. Example: NetFlix

46. References (Adaptive Filtering) • General papers on TREC filtering evaluation • D. Hull, The TREC-7 Filtering Track: Description and Analysis, TREC-7 Proceedings. • D. Hull and S. Robertson, The TREC-8 Filtering Track Final Report, TREC-8 Proceedings. • S. Robertson and D. Hull, The TREC-9 Filtering Track Final Report, TREC-9 Proceedings. • S. Robertson and I. Soboroff, The TREC 2001 Filtering Track Final Report, TREC-10 Proceedings • Papers on specific adaptive filtering methods • Stephen Robertson and Stephen Walker (2000), Threshold Setting in Adaptive Filtering. Journal of Documentation, 56:312-331, 2000 • Chengxiang Zhai, Peter Jansen, and David A. Evans,Exploration of a heuristic approach to threshold learning in adaptive filtering, 2000 ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR'00), 2000. Poster presentation. • Avi Arampatzis and Andre van Hameren The Score-Distributional Threshold Optimization for Adaptive Binary Classification Tasks , SIGIR'2001. • Yi Zhang and Jamie Callan, 2001, Maximum Likelihood Estimation for Filtering Threshold, SIGIR 2001. • T. Ault and Y. Yang, 2002, kNN, Rocchio and Metrics for Information Filtering at TREC-10, TREC-10 Proceedings.

47. References (Collaborative Filtering) • Resnick, P., Iacovou, N., Suchak, M., Bergstrom, P., and Riedl, J., "GroupLens: An Open Architecture for Collaborative Filtering of Netnews," CSCW. 175-186. • P Resnick, HR Varian, Recommender Systems, CACM 1997 • Cohen, W.W., Schapire, R.E., and Singer, Y. (1999) "Learning to Order Things", Journal of AI Research, Volume 10, pages 243-270. • Freund, Y., Iyer, R.,Schapire, R.E., & Singer, Y. (1999). An efficient boosting algorithm for combining preferences. Machine Learning Journal. 1999. • Breese, J. S., Heckerman, D., and Kadie, C. (1998). Empirical Analysis of Predictive Algorithms for Collaborative Filtering. In Proceedings of the 14th Conference on Uncertainty in Articial Intelligence, pp. 43-52. • AlexandrinPopescul and Lyle H. Ungar, Probabilistic Models for Unified Collaborative and Content-Based Recommendation in Sparse-Data Environments, UAI 2001. • N. Good, J.B. Schafer, J. Konstan, A. Borchers, B. Sarwar, J. Herlocker, and J. Riedl. "Combining Collaborative Filtering with Personal Agents for Better Recommendations." Proceedings AAAI-99. pp 439-446. 1999. • QiaozhuMei, Dengyong Zhou, Kenneth Church. Query Suggestion Using Hitting Time, CIKM'08, pages 469-478