## Eigentaste: A Constant Time Collaborative Filtering Algorithm (2000)

### Cached

### Download Links

Citations: | 273 - 4 self |

### BibTeX

@MISC{Goldberg00eigentaste:a,

author = {Ken Goldberg and Theresa Roeder and Dhruv Gupta and Chris Perkins},

title = {Eigentaste: A Constant Time Collaborative Filtering Algorithm},

year = {2000}

}

### Years of Citing Articles

### OpenURL

### Abstract

Eigentaste is a collaborative filtering algorithm that uses universal queries to elicit real-valued user ratings on a common set of items and applies principal component analysis (PCA) to the resulting dense subset of the ratings matrix. PCA facilitates dimensionality reduction for offline clustering of users and rapid computation of recommendations. For a database of n users, standard nearest-neighbor techniques require O(n) processing time to compute recommendations, whereas Eigentaste requires O(1) (constant) time. We compare Eigentaste to alternative algorithms using data from Jester, an online joke recommending system. Jester has collected approximately 2,500,000 ratings from 57,000 users. We use the Normalized Mean Absolute Error (NMAE) measure to compare performance of different algorithms. In the Appendix we use Uniform and Normal distribution models to derive analytic estimates of NMAE when predictions are random. On the Jester dataset, Eigentaste computes recommendations two ...

### Citations

2712 | Indexing by latent semantic analysis
- Deerwester, Dumais, et al.
- 1990
(Show Context)
Citation Context ...ix and identifies latent factors in the data. An application of SVD in the context of document retrieval has been patented and is widely known as Latent Semantic Indexing (LSI) (Landauer et al. 1994, =-=Deerwester et al. 1990-=-, Ding 1999, Hofmann 1999). In LSI, SVD is applied to factor the non-square termdocument frequency matrix into othogonal factor matrices with corresponding singular values. The largest singular values... |

1111 | Grouplens: An Open Architecture for Collaborative Filtering of Netnews
- Resnick, Iacovou, et al.
- 1994
(Show Context)
Citation Context ...r the new user are based on these predictions (Pennock and Horvitz 1999b). Other terms that have been proposed are “social information filtering” (Shardanand and Maes 1995), and “recommender system” (=-=Resnick et al. 1994-=-). In each case, users collaborate in the sense that each rating improves the performance of the overall system. The fundamental assumption is that if users A and B rate k items similarly, they share ... |

1014 | Empirical Analysis of Predictive Algorithms for Collaborative Filtering - Breese, Heckerman, et al. - 1998 |

879 | Social Information Filtering: Algorithms for Automating "Word of Mouth
- Shardanand, Maes
- 1995
(Show Context)
Citation Context ...unknown preferences of a new user; recommendations for the new user are based on these predictions (Pennock and Horvitz 1999b). Other terms that have been proposed are “social information filtering” (=-=Shardanand and Maes 1995-=-), and “recommender system” (Resnick et al. 1994). In each case, users collaborate in the sense that each rating improves the performance of the overall system. The fundamental assumption is that if u... |

783 | Probabilistic latent semantic indexing
- Hofmann
- 1999
(Show Context)
Citation Context ... the data. An application of SVD in the context of document retrieval has been patented and is widely known as Latent Semantic Indexing (LSI) (Landauer et al. 1994, Deerwester et al. 1990, Ding 1999, =-=Hofmann 1999-=-). In LSI, SVD is applied to factor the non-square termdocument frequency matrix into othogonal factor matrices with corresponding singular values. The largest singular values correspond to the most s... |

724 | Social Choice and Individual Values - Arrow - 1963 |

691 | Using Collaborative Filtering to Weave an Information Tapestry
- Goldberg, Nichol, et al.
- 1992
(Show Context)
Citation Context ...89 to Chislenko et al. (2000) in 2000. In 1992, D. Goldberg et al. coined the term “collaborative filtering” in the context of a system for filtering emailsEIGENTASTE 135 using binary category flags (=-=Goldberg et al. 1992-=-). Excellent surveys of research can be found in Varian and Resnick (1997), Herlocker et al. (1999a), and Delgado (2000). Shardanand and Maes (1995) designed a collaborative filtering system for music... |

591 | Grouplens: applying collaborative filtering to usenet news
- Konstan, Miller, et al.
- 1997
(Show Context)
Citation Context ... their neighborhood-based algorithms require time linear in the number of users. GroupLens is a pioneering and ongoing effort in collaborative filtering (Resnick et al. 1994, Konstan and Bharat 1996, =-=Konstan et al. 1997-=-, Herlocker et al. 1999). The GroupLens team initially implemented a neighborhood-based CF system for rating Usenet articles. They used a 1–5 integer rating scale and computed distance using Pearson c... |

524 |
An algorithmic framework for performing collaborative filtering
- Herlocker, Konstan, et al.
- 1999
(Show Context)
Citation Context ...ased algorithms require time linear in the number of users. GroupLens is a pioneering and ongoing effort in collaborative filtering (Resnick et al. 1994, Konstan and Bharat 1996, Konstan et al. 1997, =-=Herlocker et al. 1999-=-). The GroupLens team initially implemented a neighborhood-based CF system for rating Usenet articles. They used a 1–5 integer rating scale and computed distance using Pearson correlations. One of the... |

521 |
Analysis of a complex of statistical variables into principal components
- Hotelling
- 1933
(Show Context)
Citation Context ...nd positive definite. 3.4. Principal component analysis Principal Component Analysis was first introduced in 1901 by Karl Pearson (Pearson 1901). Hotelling generalized it to random variables in 1933 (=-=Hotelling 1933-=-). We apply eigenanalysis to solve for matrices E and � such that and C = E T �E, ECE T = �. Let B = AE T be a linear transform of A such that the transformed points are uncorrelated (its correlation ... |

368 | On lines and planes of closest fit to systems of points in - Pearson - 1901 |

267 | Explaining collaborative filtering recommendations - Herlocker, Konstan, et al. - 2000 |

249 | Learning collaborative information filters - Billsus, Pazzani - 1998 |

234 | User Modelling via Stereotypes - Rich - 1979 |

171 | Collaborative filtering by personality diagnosis: A hybrid memory-and model-based approach - Pennock, Horvitz, et al. - 2000 |

115 | Pointing the way: Active collaborative filtering - Maltz, Ehrlich - 1995 |

53 | A Similarity-based Probability Model for Latent Semantic Indexing - Ding - 1999 |

11 | The effects of singular value decomposition on collaborative filtering - Pryor - 1998 |

8 | E.: Analysis of the axiomatic foundations of collaborative filtering - Pennock, Horvitz |

8 | Agent-based information filtering and recommender systems on the Internet - Delgado - 2000 |

5 | Jester 2.0: A linear time collaborative filtering algorithm applied to jokes - Gupta, Goldberg - 1999 |

4 | Classification, pattern recognition, and reduction of dimensionality. Handbook of statistics - Kanal - 1982 |

4 | Personality and sense of humor - Ziv - 1984 |

2 | Integrated personal and community recommendations in collaborative filtering - Konstan, Bharat - 1996 |

1 |
Continuous vs discrete semantic differential ratings scales
- Albaum, Best, et al.
- 1981
(Show Context)
Citation Context ...granularity of HTML image maps), we can distinguish approximately 200 levels of ratings in the scale. Continuous ratings avoid discretization effects in matrix computations and may offer measurement (=-=Albaum et al. 1981-=-) and user-interface advantages as discussed in the conclusion. Eigentaste splits computations into offline and online phases. Offline, Eigentaste uses principal component analysis for optimal dimensi... |

1 | NN Pattern Classification Techniques - BV - 1991 |

1 | Agent-Based Information Filtering and Recommender Systems on the Internet - JA - 2000 |

1 | A similarity-based probability model for latent semantic indexing - CHQ - 1999 |

1 | Us patent 4870579: System and method of predicting subjective reactions - Hey - 1989 |

1 | A User Guide to Principal Components: A Problem Solving Approach - JE - 1991 |

1 | Pointing the way: Active collaborative filtering - DA, Ehlrich - 1995 |

1 | Konstan J and Riedl JT (2000) Application of dimensionality reduction in recommender systems—a case study - BM, Karypis |

1 | Special issue on cf and recommender systems - Varian, Resnick - 1997 |

1 | NN Pattern Classification Techniques - Dasarathy - 1991 |

1 | et al. Us patent 6092049: Method and apparatus for efficiently recommending items using automated collaborative filtering and feature-guided automated collaborative filtering - Chislenko - 2000 |

1 | A User Guide to Principal Components: A Problem Solving Approach - Jackson - 1991 |

1 | Bell Communications Research (Bellcore). Us patent 5301109: Computerized cross-language document retrieval using latent semantic indexing - Landauer, Littman - 1994 |