steam — Steam video game reviews, ownership and bundles (2010-2018)

Description

Networks of the users of Steam, the video game distribution platform of Valve, and the games (and other products, such as soundtracks and add-ons) they reviewed or own, collected by J. McAuley and colleagues (University of California, San Diego) from the public pages of the Steam store and community.

There are four networks:

reviews: the reviews of games on Steam (version 2 of the data, collected for Kang and McAuley 2018), 7.8 million reviews by 2.6 million users of 15 thousand games, from 2010 to January 2018. A directed edge from a user to a game is a review. Users are identified by their Steam user id where it is given with the review (for about 40% of the reviews), otherwise by their user name; since user names need not be unique, and a user may appear with and without id, user nodes do not correspond exactly to users.
ownership: the games owned by 88,310 users of Australia (version 1 of the data, collected for Wan and McAuley 2018 and Pathak et al. 2017), with the time played. A directed edge from a user to a game means that the user owns the game. the 673 users listed twice are included once (with their first listing), which leaves 5.09 of the 5.15 million ownership records.
australian_reviews: the reviews written by 25,799 of these Australian users (59,305 reviews), with their recommendation and the ratings by other users.
bundles: the 615 bundles of games sold together on Steam (as listed by SteamDB), and the 3,525 items in them. A directed edge from a bundle to a game means that the game is part of the bundle.

All networks are bipartite. The games have the metadata of the Steam store (as collected in early 2018) where available; the texts of the reviews are not included.

The networks have the following properties:

(Node) is_user (reviews, ownership, australian_reviews): whether the node is a user (otherwise it is a game).
(Node) is_bundle (bundles): whether the node is a bundle (otherwise it is a game).
(Node) id: the Steam user id or user name of a user, the Steam application id of a game, or the id of a bundle.
(Node) name (games and bundles): the name of the game or bundle.
(Node) products (users of reviews): the number of products of the user, as shown with the review (the largest value over the user's reviews).
(Node) genres, tags, specs (games): the genres, user-defined tags and features of the game in the store, comma-separated.
(Node) developer, publisher, release_date, sentiment (games): as given in the store (e.g. Mostly Positive).
(Node) price (games): the price of the game in US dollars (NaN if not given or not a number, e.g. free to play).
(Node) metascore (games): the Metacritic score of the game (-1 if not given).
(Node) early_access (games): whether the game was in early access.
(Node) bundle_price, bundle_final_price (bundles): the price of the bundle without and with its discount, in US dollars.
(Node) genre (games of bundles): the genres of the game, as given with the bundle.
(Edge) time (reviews): the date of the review, as Unix time.
(Edge) hours (reviews): the hours played by the user at the time of the review (NaN if not given).
(Edge) recommended (reviews): whether the user recommends the game.
(Edge) early_access (reviews): whether the review was written while the game was in early access.
(Edge) found_funny (reviews): the number of users who found the review funny (0 if not given).
(Edge) compensation (reviews): the compensation of the reviewer, e.g. Product received for free (empty if none).
(Edge) playtime_forever, playtime_2weeks (ownership): the time played by the user, in total and in the last two weeks, in minutes.
(Edge) posted, last_edited (australian_reviews): the dates of the review, as shown on Steam (e.g. Posted November 5, 2011.; without the year for reviews of the year of collection).
(Edge) recommend (australian_reviews): whether the user recommends the game.
(Edge) funny, helpful (australian_reviews): the ratings of the review by other users, as shown on Steam (e.g. 15 of 20 people (75%) found this review helpful).
(Edge) discounted_price (bundles): the price of the game in the bundle, in US dollars.

Tags

Social Online Preferences Bipartite Multigraph Timestamps Metadata

Added

2026-10-28

Citation

W.-C. Kang, J. McAuley, "Self-Attentive Sequential Recommendation." 2018 IEEE International Conference on Data Mining (ICDM), 197-206 (2018) 10.1109/ICDM.2018.00035 Sci-Hub
M. Wan, J. McAuley, "Item recommendation on monotonic behavior chains." Proceedings of the 12th ACM Conference on Recommender Systems (RecSys), 86-94 (2018) 10.1145/3240323.3240369 Sci-Hub
A. Pathak, K. Gupta, J. McAuley, "Generating and Personalizing Bundle Recommendations on Steam." Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval, 1073-1076 (2017) 10.1145/3077136.3080724 Sci-Hub

Upstream URL OK

https://cseweb.ucsd.edu/~jmcauley/datasets.html#steam_data

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	$\left<k\right>$	$\sigma_k$	$\lambda_h$	$\tau$	$r$	$\oslash$	$S$	Kind	Mode	NPs	EPs	gt	GraphML	GML	csv
reviews	2,842,673	7,793,069	2.74	221.27	291.57	26.00	-0.13	12	1.00	Directed	Bipartite	is_user id products name genres tags specs developer publisher release_date sentiment price metascore early_access	time hours recommended early_access found_funny compensation	57.9 MiB	110.8 MiB	108.6 MiB	97.5 MiB
ownership	98,604	5,094,082	51.66	622.68	949.58	2.06	-0.23	6	0.83	Directed	Bipartite	is_user id name genres tags specs developer publisher release_date sentiment price metascore early_access	playtime_forever playtime_2weeks	12.4 MiB	29.8 MiB	28.9 MiB	23.2 MiB
australian_reviews	29,167	59,305	2.03	38.61	37.67	24.76	-0.15	14	0.98	Directed	Bipartite	is_user id name genres tags specs developer publisher release_date sentiment price metascore early_access	posted last_edited recommend funny helpful	1.1 MiB	1.5 MiB	1.5 MiB	1.5 MiB
bundles	3,413	3,525	1.03	3.85	5.92	24.45	-0.25	4	0.03	Directed	Bipartite	is_bundle id bundle_price bundle_final_price genre name genres tags specs developer publisher release_date sentiment price metascore early_access	discounted_price	165 KiB	212 KiB	208 KiB	210 KiB