Bipartite or recommendation system datasets?

107 views
Skip to first unread message

Deklan Webster

unread,
Nov 26, 2020, 12:01:22 PM11/26/20
to Open Graph Benchmark

Hi all,

Just noticed that there are no link prediction tasks with bipartite networks. Most recommendation systems can be cast this way, so it feels like an important omission.

I saw the call for datasets here https://docs.google.com/document/d/1bSVtHGtSMwu6B3setIrEAnNxI4qBYRoFIClAkk7aOAg/edit but I don't have any particular datasets in mind. The RecSys community has plenty, surely?

Even disregarding recommendation systems, there could be another MAG citation network but with papers and authors as nodes instead of just papers.

Thoughts?

Thanks

Weihua Hu

unread,
Nov 26, 2020, 12:12:43 PM11/26/20
to Deklan Webster, Open Graph Benchmark
Hi,

Great point! A bipartite graph dataset is what we are currently working on, so stay tuned!

Best,
Weihua

2020年11月26日(木) 9:01 Deklan Webster <dek...@gmail.com>:
--
You received this message because you are subscribed to the Google Groups "Open Graph Benchmark" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-graph-bench...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-graph-benchmark/3efb5062-af5f-41b9-aa67-4e9fed90de8cn%40googlegroups.com.

Deklan Webster

unread,
Nov 26, 2020, 12:40:57 PM11/26/20
to Open Graph Benchmark
Great news! Eager to see the results. Thanks!
Reply all
Reply to author
Forward
0 new messages