tripadvisor_ratings — TripAdvisor hotel ratings (2001-2009)

Description

Bipartite networks of the reviewers of TripAdvisor and the hotels they reviewed, from the hotel reviews crawled from TripAdvisor (www.tripadvisor.com) by Hongning Wang, Yue Lu and ChengXiang Zhai for their research on latent aspect rating analysis (2010, 2011). Each node is a reviewer or a hotel, and an edge connects a reviewer to a hotel they reviewed, with the review's overall rating, its ratings of individual aspects of the hotel (value, rooms, location, cleanliness, check in / front desk, service, business service; each from 0 to 5 stars, or -1 if not given), and its date. A reviewer who reviewed the same hotel several times is connected to it by several edges.

The authors published two crawls, which are two networks:

texts: the parsed reviews of 1,759 hotels (the "Text" data set, 240,060 reviews from 2001 to 2009, 64,295 of them anonymous), with the number of readers and of helpful votes of each review.
json: the authors' "most comprehensive collection" of hotel reviews, crawled at different times and published in JSON format in 2014, with the TripAdvisor id of each review and the location given by its reviewer. The format varies slightly across hotels (e.g. some hotels have no name, address or price, and some reviews have no reviewer location).

The reviewers are identified by their TripAdvisor screen names. The reviews whose author is not given by name, i.e. those of "A TripAdvisor Member" (a reviewer without a public profile), "lass=" (apparently a parsing error of the crawl, for which the author is missing), an empty name, or "Posted by a(n) ... traveler" (reviews imported from partner sites), are anonymous: each of them gets a node of its own (marked by anonymous), since they cannot be attributed to a single reviewer. In texts, the names of reviewers with non-Latin characters were stored in mangled form (UTF-8 decoded as GBK); they are restored where this is possible without loss, and left as they are otherwise. The hotels' names and locations are taken from their TripAdvisor URLs (the names in json where they are given); 36 hotels of texts have no URL, and so no name or location. The texts of the reviews (and their titles) are left out.

The networks have the following properties:

(Node) name: the screen name of the reviewer (as given in the data for anonymous reviews), or the name of the hotel.
(Node) is_hotel: whether the node is a hotel (otherwise a reviewer).
(Node) anonymous: whether the node is an anonymous review (see above).
(Node) hotel_id: the TripAdvisor id of the hotel (-1 for reviewers).
(Node) geo_id: the TripAdvisor id of the hotel's location (-1 if unknown, and for reviewers).
(Node) location: the location of the hotel (from its URL, e.g. Seattle Washington; empty if unknown, and for reviewers).
(Node) hotel_rating: the average rating of the hotel shown by TripAdvisor (texts only; -1 for reviewers).
(Node) price: the (lowest) price of the hotel in US dollars, as shown by TripAdvisor (-1 if unknown, and for reviewers).
(Node) address: the address of the hotel (json only; empty if unknown, and for reviewers).
(Edge) rating: the overall rating of the review, from 0 to 5 (-1 if not given).
(Edge) rating_value, rating_rooms, rating_location, rating_cleanliness, rating_check_in, rating_service, rating_business_service: the ratings of these aspects, from 0 to 5 (-1 if not given); json also has rating_sleep_quality.
(Edge) time: the date of the review, in Unix time (-1 if unknown).
(Edge) readers, helpful: the number of readers of the review, and of readers who found it helpful (texts only; -1 if not given).
(Edge) review_id: the TripAdvisor id of the review (json only).
(Edge) author_location: the location given by the reviewer (json only; empty if not given).

Tags

Economic Online Preferences Bipartite Weighted Multigraph Timestamps Metadata

Added

2026-10-27

Citation

H. Wang, Y. Lu, C. Zhai, "Latent aspect rating analysis on review text data: a rating regression approach." Proceedings of the 16th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 783-792 (2010). 10.1145/1835804.1835903 Sci-Hub
H. Wang, Y. Lu, C. Zhai, "Latent aspect rating analysis without aspect keyword supervision." Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 618-626 (2011). 10.1145/2020408.2020505 Sci-Hub

Upstream URL OK

https://www.cs.virginia.edu/~hw5x/dataset.html
https://web.archive.org/web/20140914234636/http://times.cs.uiuc.edu/~wang296/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	$\left<k\right>$	$\sigma_k$	$\lambda_h$	$\tau$	$r$	$c$	$\oslash$	$S$	Kind	Mode	NPs	EPs	gt	GraphML	GML	csv
texts	211,370	240,060	2.27	19.82	11.55	592.02	-0.23	0.00	14	1.00	Undirected	Bipartite	name is_hotel anonymous hotel_id geo_id location hotel_rating price	rating rating_value rating_rooms rating_location rating_cleanliness rating_check_in rating_service rating_business_service time readers helpful	5.1 MiB	8.5 MiB	8.1 MiB	7.2 MiB
json	892,431	1,621,956	3.63	33.98	71.21	293.73	-0.23	0.00	15	1.00	Undirected	Bipartite	name is_hotel anonymous hotel_id geo_id location price address	rating rating_value rating_rooms rating_location rating_cleanliness rating_check_in rating_service rating_business_service rating_sleep_quality time review_id author_location	31.5 MiB	50.3 MiB	48.1 MiB	46.6 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.