EPrints Technical Mailing List Archive
See the EPrints wiki for instructions on how to join this mailing list and related information.
Message: #08227
< Previous (by date) | Next (by date) > | < Previous (in thread) | Next (in thread) > | Messages - Most Recent First | Threads - Most Recent First
Re: [EP-tech] Experimental integration with MS Azure Cognitive Services
- To: <eprints-tech@ecs.soton.ac.uk>, Liam Green-Hughes <L.E.Green-Hughes@kent.ac.uk>
- Subject: Re: [EP-tech] Experimental integration with MS Azure Cognitive Services
- From: David R Newman <drn@ecs.soton.ac.uk>
- Date: Mon, 29 Jun 2020 10:05:31 +0100
One thing I forgot to mention was that this call is probably
better as an event queue task rather than a before commit
trigger. As if you are relying on a third-party application you
don't want to be waiting on this before the page can reload. If
you are only send the abstract that is probably going to be fairly
rapid. However, sometimes abstracts can be rather long and I can
imagine the service being less responsive at times. So you may
wait a while before you get a response back, so you can reload the
page. Obviously the disadvantage of having an event queue task is
that you will end up with two revisions rather than one. However,
your code at the moment suggests this service will only be called
once, as once there are keywords it cannot be run again.
Just another thought: If you were to change this so keywords
could be updated, obviously you would want to check to see the
fields that were being sent for keywords analysis had changed and
only call this service if they had. You would also need some code
to parse the current and returned keywords to merge them together.
Hi Liam,
Looks interesting. I have been working on improving search on keywords within EPrints by introducing a new Keywords MetaField type that is backwards with the Longtext MetaField that is currently the type with the keywords field. I have also introduced a Idci (short for ID Case Insentive) field that could be used for keywords when set to be a multiple field. Hopefully, I will be able to make the official release of 3.4.2 that includes these available this week.
Regards
David Newman
On 29/06/2020 09:44, Liam Green-Hughes via Eprints-tech wrote:
Hi everyone,
I've been experimenting with integrating EPrints with an off the shelf AI solution to generate keywords. The service I used was the Text Analytics service element of the Microsoft Azure Text Analytics. I've only gone as far as experimenting with the Key Phrases endpoint so far, but the Named Entities endpoint looks like it could add real value to EPrints records too.
The integration file is here: https://github.com/liamgh/eprints-ai-expt/blob/master/z_azure_keywords.pl. It isn't really production ready, but it is a starting point.
Let me know your thoughts!
ThanksLiam
*** Options: http://mailman.ecs.soton.ac.uk/mailman/listinfo/eprints-tech *** Archive: http://www.eprints.org/tech.php/ *** EPrints community wiki: http://wiki.eprints.org/
*** Options: http://mailman.ecs.soton.ac.uk/mailman/listinfo/eprints-tech *** Archive: http://www.eprints.org/tech.php/ *** EPrints community wiki: http://wiki.eprints.org/
- References:
- [EP-tech] Experimental integration with MS Azure Cognitive Services
- From: Liam Green-Hughes <L.E.Green-Hughes@kent.ac.uk>
- Re: [EP-tech] Experimental integration with MS Azure Cognitive Services
- From: David R Newman <drn@ecs.soton.ac.uk>
- [EP-tech] Experimental integration with MS Azure Cognitive Services
- Prev by Date: Re: [EP-tech] Experimental integration with MS Azure Cognitive Services
- Next by Date: Re: [EP-tech] Experimental integration with MS Azure Cognitive Services
- Previous by thread: [EP-tech] EPrints/CRIS
- Next by thread: [EP-tech] DOI handling in orcid_support_advance
- Index(es):