How do referenceMentions offsets work?

Question

How do referenceMentions offsets work?

Closed this issue 2 months ago · 2 comments

In the paper contents JSON, under metadata.referenceMentions, there are objects with startOffset and endOffset fields. After some exploration, I was unable to determine what they are offsets of and what they mean. Are those the boundaries of the content in the context field? Where? Are they character offsets, or somehow word offsets?

Answer 1 · 2024-10-22T23:55:33.000Z

Here is an example

In this example, The reference appear at context[startOffset: endOffset], which is "(Hornik, 1991)".

Answer 2 · 2024-10-23T18:16:06.000Z

Oh, thanks. I thought this would be something completely different. I had my mind on locating where the context appeared in the paper to see the surrounding paragraph, but I solved that in other ways.

Thanks for the reply!