Query biased document summary in PyTerrier

[adambaker]

I'm trying to extract text snippets for a search results page. I've found the documentation page for running terrier in a JSP, which describes some configuration for snippet extraction.

I'm still not sure how to use this in pyterrier. It seems like BatchRetrieve's properties or control parameters might be necessary to pass the required configuration, but then it's not clear to me how to get the query biased summary from the result set. Is there a way to do this with PyTerrier?

Thanks!

[cmacdonald]

Hi adam,

The query-biased summariser in Terrier is NOT that great. What about just breaking the text of the retrieved documents into sentences, and passing them through another text scoring ranking pipeline?

Craig

[cmacdonald]

Adam, there is an draft PR for PyTerrier query biased summarisation at #223. We're not planning to merge for this week's PyT release, but it might be of interest.

[cmacdonald]

The corresponding PR has been merged - documentation at https://pyterrier.readthedocs.io/en/latest/text.html#query-biased-summarisation-snippets.