Jester Datasets for Recommender Systems and Collaborative Filtering Research
6.5 million anonymous ratings of jokes by users of
the Jester Joke Recommender System
(Ken Goldberg,
AUTOLab, UC Berkeley)
Archived Older Version of this page (pre-2020)
Freely available for research use when acknowledged with the following reference:
Eigentaste: A Constant Time Collaborative Filtering Algorithm. Ken Goldberg, Theresa Roeder, Dhruv Gupta, and Chris Perkins. Information Retrieval, 4(2), 133-151. July 2001.
As a courtesy if you use the data, please send me an email:
Ken Goldberg: goldberg@berkeley.edu.
Dataset 1: 4.1 million ratings
Values from (-10.00 to +10.00) of 100 jokes from 73,421 users: collected between April 1999 - May 2003
The text for each of the 100 Dataset 1 jokes can be downloaded here: jester_dataset_1_joke_texts.zip (92KB)
Format:
- 100 files
- Each file has name init1.html,...,init100.html
- The numbers refer to the ID's of the jokes in the Excel files
The ratings data:
- jester_dataset_1_1.zip: (3.9MB) Data from 24,983 users who have rated 36 or more jokes, a matrix with dimensions 24983 X 101.
- jester_dataset_1_2.zip: (3.6MB) Data from 23,500 users who have rated 36 or more jokes, a matrix with dimensions 23500 X 101.
- jester_dataset_1_3.zip: (2.1MB) Data from 24,938 users who have rated between 15 and 35 jokes, a matrix with dimensions 24,938 X 101.
Format:
- Data files are in .zip format, when unzipped, they are in Excel (.xls) format
- Ratings are real values ranging from -10.00 to +10.00 (the value "99" corresponds to "null" = "not rated").
- One row per user
- The first column gives the number of jokes rated by that user. The next 100 columns give the ratings for jokes 01 - 100.
- The sub-matrix including only columns {5, 7, 8, 13, 15, 16, 17, 18, 19, 20} is dense. Almost all users have rated those jokes (see discussion of "universal queries" in the above paper).
Dataset 3: 2.3 million ratings
Values from (-10.00 to +10.00) of 100 jokes from 73,421 users: collected between April 1999 - May 2003
Includes 150 jokes( 50 not in Dataset 1) and an updated version of our old Dataset 2 with over 115,000 new
ratings from 82,366 total users: data collected from November 2006 - Mar 2015
The text of the 150 Dataset 3 jokes: jester_dataset_2/3_joke_texts.zip (29KB)
Format:
- An excel spreadsheet with 150 rows
- The row number corresponds to the joke ID referred to in the Excel files above
- The first 100 jokes and their ID's are consistent with the 100 jokes in Dataset 1
The Ratings Data, Save to disk, then unzip: jester_dataset_3.zip (6MB)
Format:
- The data is formatted as an excel file representing a 54,905 by 151 with rows
as users and columns as jokes. The left-most column contains the total amount of jokes
rated by the user. There are a total of 54905 users
and 150 jokes in this dataset.
- 22 of these jokes have few ratings as they
were removed as of May 2009 deemed to be out of date (eg, Bill Clinton jokes;)
their ids are:
{1, 2, 3, 4, 5, 6, 9, 10, 11, 12, 14, 20, 27, 31, 43, 51,
52, 61, 73, 80, 100, 116}.
- Each rating is from (-10.00 to +10.00)
and 99 corresponds to a null rating (user did not rate that joke).
Note that the ratings are real values ranging from -10.00 to +10.00. As of May 2009, the jokes
{7, 8, 13, 15, 16, 17, 18, 19} are the "gauge set" (as discussed in the Eigentaste paper) and the jokes
{1, 2, 3, 4, 5, 6, 9, 10, 11, 12, 14, 20, 27, 31, 43, 51, 52, 61, 73, 80, 100, 116} were removed
(i.e. they are never displayed or rated).
Dataset 4
Over 100,000 new ratings from 7,699 total users: data collected from April 2015 - Nov 2019
The text of the jokes: jester_dataset_4_joke_texts.zip (30KB)
Format:
- Includes 8 new jokes 151-158.
- An excel sheet with 158 rows.
- The row number corresponds to the joke ID referred to in the Excel files below
- The first 150 jokes and their ID's are consistent with the jokes from earlier datasets
The Ratings Data: Save to disk, then unzip: jester_dataset_4.zip (1.4MB)
Format:
- The data is formatted as an excel file representing a 7699 by 159 matrix with rows as
users and columns as jokes. The left-most column represents the amount of jokes rated by each user.
There are a total of 7699 users and 158 jokes in this dataset.
- 22 of the jokes don't have ratings, their ids are:
{1, 2, 3, 4, 5, 6, 9, 10, 11, 12, 14, 20, 27, 31, 43, 51,
52, 61, 73, 80, 100, 116}.
- 8 jokes were added to
this version {151-158}
- Each rating is from (-10.00 to +10.00) and 99 corresponds to a
null rating (user did not rate that joke).
Note that the ratings are real values ranging from -10.00 to +10.00. The jokes {1, 2, 3, 4, 5, 6, 9, 10, 11,
12, 14, 20, 27, 31, 43, 51, 52, 61, 73, 80, 100, 116} have been removed (i.e. they are never displayed or rated).
As of April 2015, 8 jokes were added.
Other Collaborative Filtering Datasets:
- The MovieLens Dataset: 1,000,000 integer ratings (from 1-5) of 3500 films from 6,040 users.
- The EachMovie Dataset: 2,811,983 integer ratings (from 1-5) of 1628 films from 72,916 users.
- The BookCrossing Dataset: 1,149,780 integer ratings (from 0-10) of 271,379 books from 278,858 users.
1700 papers using these Jester Datasets
For further information please contact:
Ken Goldberg
goldberg at berkeley dot edu
Prof of IEOR and EECS
UC Berkeley
(510) 643-9565 (phone)