search

eXascaleInfolab / TRank

Ranking Entity Types using the Web of Data
Apache License 2.0
30 stars 2 forks source link

Benchmark link broken #4

Open flackbash opened 2 weeks ago

flackbash commented 2 weeks ago

Dear authors,

We are trying to solve a similar (although not the same) problem and would like to see if we can use your benchmarks. However, the link to the dataset from the paper is broken: https://exascale.info/TRank

Is is possible to repair the link or access the dataset elsewhere?

Many thanks in advance, Natalie

gianluca81 commented 1 week ago

Dear Natalie,

thanks for your message and the interest in our work. Unfortunately, I was not able to find a copy of the data hosted on that website, but looking at my own backups I could find the ground truth data that we collected by means of crowdsourcing. If this is the data you were looking for, you can download it here https://filesender.aarnet.edu.au/?s=download&token=9c3be2dd-802d-4728-97b9-773cf1b6f471 (link expires in 40 days). The files contain the document ID (if appropriate), the entity URI, the type URI, and how many crowd workers voted for each type being relevant for that entity (in the context where it appears, if appropriate). Any question, please, let us know.

Best wishes, Gianluca

flackbash commented 1 week ago

Awesome, thanks a lot for looking into this!

In the meantime, Philippe has sent me some data which I believe is the dataset that was compiled from the raw crowdsourcing results and is working on repairing the link. But this additional data is very interesting!

Best regards, Natalie