Amazon ratings

This bipartite network contains product ratings from the Amazon online shopping website. The rating scale ranges from 1 to 5, where 5 denotes the most positive rating. Nodes represent users and products, and edges represent individual ratings.


Internal nameamazon-ratings
NameAmazon ratings
Data source
AvailabilityDataset is available for download
Consistency checkDataset passed all tests
Rating network
Node meaningUser, product
Edge meaningRate
Network formatBipartite, undirected
Edge typeRatings, multiple edges
Temporal data Edges are annotated with timestamps
Snapshot Is a snapshot and likely to not contain all data


Size n =3,376,972
Left size n1 =2,146,057
Right size n2 =1,230,915
Volume m =5,838,041
Unique edge count m̿ =5,743,258
Wedge count s =627,186,651
Claw count z =704,564,362,291
Cross count x =1,684,620,121,457,183
Square count q =35,849,304
4-Tour count T4 =2,807,043,116
Maximum degree dmax =12,217
Maximum left degree d1max =12,217
Maximum right degree d2max =3,146
Average degree d =3.457 56
Average left degree d1 =2.720 36
Average right degree d2 =4.742 85
Fill p =2.174 15 × 10−6
Average edge multiplicity m̃ =1.016 50
Size of LCC N =2,892,456
Diameter δ =28
50-Percentile effective diameter δ0.5 =5.924 44
90-Percentile effective diameter δ0.9 =8.044 95
Median distance δM =6
Mean distance δm =6.629 27
Gini coefficient G =0.650 684
Balanced inequality ratio P =0.248 637
Left balanced inequality ratio P1 =0.276 822
Right balanced inequality ratio P2 =0.236 379
Relative edge distribution entropy Her =0.916 513
Power law exponent γ =2.957 80
Tail power law exponent γt =2.071 00
Degree assortativity ρ =−0.035 657 9
Degree assortativity p-value pρ =0.000 00
Spectral norm α =7,942.59
Spectral separation 1[A] / λ2[A]| =3.442 97
Negativity ζ =0.377 003
Controllability C =1,690,624
Relative controllability Cr =0.500 633


Fruchterman–Reingold graph drawing

Degree distribution

Cumulative degree distribution

Lorenz curve

Spectral distribution of the adjacency matrix

Spectral distribution of the normalized adjacency matrix

Spectral distribution of the Laplacian

Spectral graph drawing based on the adjacency matrix

Spectral graph drawing based on the normalized adjacency matrix

Degree assortativity

Hop distribution

Delaunay graph drawing

Item rating evolution

Edge weight/multiplicity distribution

Temporal distribution

Diameter/density evolution

Signed temporal distribution

Rating class evolution

Inter-event distribution

Node-level inter-event distribution



[1] Jérôme Kunegis. KONECT – The Koblenz Network Collection. In Proc. Int. Conf. on World Wide Web Companion, pages 1343–1350, 2013. [ http ]
[2] Ee-Peng Lim, Viet-An Nguyen, Nitin Jindal, Bing Liu, and Hady Wirawan Lauw. Detecting product review spammers using rating behaviors. In Proc. Int. Conf. on Information and Knowl. Management, pages 939–948, 2010.
[3] Nitin Jindal and Bing Liu. Opinion spam and analysis. In Proc. Int. Conf. on Web Search and Web Data Min., pages 219–230, 2008.
[4] Arjun Mukherjee, Bing Liu, and Natalie Glance. Spotting fake reviewer groups in consumer reviews. In Proc. Int. Conf. on World Wide Web, pages 191–200, 2012.