How many columns should be annotated per input file?

#1

Hi,
I am looking at the task description at https://www.aicrowd.com/challenges/iswc-2019-column-type-annotation-cta-challenge where it says “Each column should be annotated by one class”.

However the output as I understand is suppose to be one line per row like this:
“9206866_1_8114610355671172497”,”0”,”dbo:Country”
This seems a bit confusing to me.

So my question is, how many columns we should annotate in a row? For example in 21362676_0_6854186738074119688.csv:

Fans’ Rank, Title, Year, Director(s), Overall, Rank
442, The Game, 1997, David Fincher,1491

For this row , are you expecting we annotate both column Title and Director(s)? In that case:
“21362676_0_6854186738074119688”,“1”,“dbo:Film”
“21362676_0_6854186738074119688”,“3”,“dbo:Artist”

Or are expecting just one annotation for the Title column? In that case:
“21362676_0_6854186738074119688”,“1”,“dbo:Film”

PS: replaced http://dbpedia.org/ontology/ by dbo: prefix for this post as this forum doesn’t allow me to post more than 2 links
Regards,
Rakeb

#2

Hi, Rakeb

In the first round, we by default hope the participant to annotate all the potential columns that are composed of entity mentions. In your example, both “21362676_0_6854186738074119688”,“1” and “21362676_0_6854186738074119688”,“3” should be annotated.

However, in the second round, we will specify the target columns to annotate.

Hope my response is helpful to you.

Regards,
Jiaoyan