epinions_massa — Epinions product ratings and trust (2003)

Description

The "downloaded Epinions dataset" of Paolo Massa, collected in a 5-week crawl of the consumer review site Epinions.com in November and December 2003, and distributed on the TrustLet wiki: the ratings that the users gave to products (items) when reviewing them, and the trust statements among the users (their "web of trust"). Users and items are identified by anonymized numeric ids, from 1 to 49,289 for the users (the documentation gives 49,290 users, but no id above 49,289 appears in the files) and from 1 to 139,738 for the items. The distrust statements (the "block list") are not included, since the site kept them private.

ratings: a bipartite network of the users and the items, with 664,824 ratings (no pair of user and item is repeated). An edge connects a user to an item they rated, on a scale of 1 to 5. All the users of the dataset are included, also the 9,126 who rated no item (but appear in the trust statements), so that the user nodes are the same as in trust.
trust: a directed network of the 487,183 trust statements among the users (the documentation gives 487,181), where an edge from user A to user B means that A trusts B. All the users are included, also those making or receiving no statement; 181 users trust themselves.

Other Epinions data in the repository: epinions is the larger dataset that the Epinions staff gave to Massa (2001-2003), with the helpfulness ratings of the reviews (not of the products), and the trust and distrust statements with their dates; epinions_trust is the trust network collected by Richardson, Agrawal and Domingos in 2003; and the network epinions of signed_social_media is the same trust and distrust network as in epinions, without the dates. The user ids differ between these datasets.

The networks have the following properties:

(Node) is_user (ratings): whether the node is a user (otherwise an item).
(Node) id (ratings): the id of the user or item.
(Edge) rating (ratings): the rating (1 to 5).
(Node) id (trust): the id of the user.

Tags

Social Online Economic Preferences Bipartite Weighted Unweighted

Added

2026-10-31

Citation

P. Massa, P. Avesani, "Trust-aware recommender systems." Proceedings of the 2007 ACM Conference on Recommender Systems (RecSys), 17-24 (2007) 10.1145/1297231.1297235 Sci-Hub

Upstream URL OK

https://web.archive.org/web/20100308104815/http://www.trustlet.org/wiki/Downloaded_Epinions_dataset

Networks

Tip: click on a column header to sort the list, or hover over it for a description.

Column legend

$\left<k\right>$: Mean degree
$\sigma_k$: Standard deviation of the degrees
$\lambda_h$: Largest eigenvalue of the Hashimoto (non-backtracking) matrix
$\tau$: Random walk mixing time
$r$: Degree assortativity
$c$: Global clustering coefficient
$\oslash$: (Pseudo-) diameter
$S$: Relative size of the largest component
NPs: Node properties
EPs: Edge properties

Name	Nodes	Edges	$\left<k\right>$	$\sigma_k$	$\lambda_h$	$\tau$	$r$	$c$	$\oslash$	$S$	Kind	Mode	NPs	EPs	gt	GraphML	GML	csv
ratings	189,027	664,824	7.03	24.81	81.94	80.85	-0.09	0.00	13	0.94	Undirected	Bipartite	is_user id	rating	4.2 MiB	6.9 MiB	6.9 MiB	5.8 MiB
trust	49,289	487,183	9.88	64.85	191.54	56.74	-0.01	0.10	14	1.00	Directed	Unipartite	id		1.5 MiB	2.8 MiB	2.7 MiB	2.3 MiB

Ridiculograms^*

^* These are automatically generated force-directed visualizations, and can be quite meaningless for networks both big and small. They should not be taken seriously as sources of scientific insight. See here for a discussion.