Open migalkin opened 2 years ago
I see, but creating datasets, for instance, as done in KGReasoning from Stanford, involves a lot of random sampling calls. This means the query datasets sampled from those original triples might very well be different (eg, sampled by you and sampled by me), which renders results in the paper hardly reproducible 😕
There are many options to store larger datasets - Dropbox / Google Drive / S3 buckets - most of them are free to use
Yes. That makes sense. Let me check if I can get my dataset and put them in a dropbox link for everyone's use. Thanks a lot for the suggestion.
Hi, the repo does not have any links to the datasets used in the paper, esp. the new ones on DRKG and DBpedia. Do you have any plans sharing those?