User talk:Charles Matthews/Facto Post/Issue 24 - 17 May 2019

@Charles Matthews: Thanks, this is really interesting. I have been using the CM software and did a few dedicated searches related to the work of the EU NanoSafety Cluster. But with a lot of text mining results, crowdsourcing to double check the extracted things is very useful to remove false positives. Is there a way for http://sciencesource-review.wmflabs.org/ to show just a subset of articles around a particular topic? So that it can target a specific research community? It seems to focus now on random articles, but I'm certainly not an expert on all those topics, making many questions the system asks ("is this sentence about that") often non-trivial. This is, for example, for chemical names, where a claim about an "X derivative" is not about "X". Second, do you accept the output from a custom ContentMine workflow into this system, like of https://github.com/egonw/cmnanotox ? Oh, and a last thing... can we talk with you about open pull requests for the software too? --Egon Willighagen (talk) 06:14, 18 May 2019 (UTC)

Oh, a quick follow up idea! Another possible subset I can think of is the set of articles for some author!! Who else can better evaluate the text mining results than the author of the article. And since we increasingly have Twitter handles for authors, I was thinking about tweeting authors on improving there article annotation ("main subject") anyway, via Scholia, particularly for chemicals. But having them review their own articles with http://sciencesource-review.wmflabs.org/, that would be awesome! (apologies for the excitement ;) --Egon Willighagen (talk) 06:18, 18 May 2019 (UTC)