Super fun and engaging session. Fellow panelists were awesome as were the audience questions. You did a great job moderating it, @milda_nor! Thanks for the invitation.
@QuareWorkshop#QUARE23#RecSys2023
Robin Burke: "If my livelihood depends on what a recsys does with my creative material, I need more transparency around why the algorithm is reaching some people, why is not reaching others, etc." #QUARE23#RecSys2023@ACMRecSys
A fast-food mentality to deploy ready-made #ML approaches may be what's hampering to move beyond post-hoc rationale generation for #RecSys@UpolEhsan – great discussion in the session session of #QUARE23@ACMRecSys#RecSys2023
Work by @UpolEhsan et al. encourages to think not only about different stakeholder needs but also how their background impacts perception of explanations & how surprising some of the perceptions can be. Thank you for joining & sharing! #QUARE23@QuareWorkshop#RecSys2023
The second session of #QUARE23 has started! Robin Burke's first challenge for the #explanations community is "let's build explanations which are not a lie!" #RecSys2023, @ACMRecSys
@UpolEhsan is the second panellist, talking now about his human-centred #xAI approach, looking not only at users of the systems, but also at those who set the goals of explanations for #recommendations.
How can an explanations boost people’s critical thinking? We should be able the system as needed, not just encourage to trust systems more – @navatintarev #QUARE23, @QuareWorkshop, #RecSys2023, @ACMRecSys
Assessing whether users comprehend explanations is inherently hard, that's why pbjective measurement for explanations are so hard to achieve, says @chunhuatsai one of the panellists of the first session of #QUARE23, @QuareWorkshop, #RecSys2023, @ACMRecSys
If you're looking for a #PhD in human-centric or participatory #artificialintelligence, get in touch. I have a scholarship for candidates who qualify for home/UK fees. You'd be joining a vibrant, diverse department with lots of interesting projects @kclinformatics@KingsNMES
I went to 3 workshops and all of them had nice discussions and interaction. The 1st Workshop on Measuring the Quality of Explanations in Recommender Systems was especially great, with short talks and a moderated, multi-disciplinary discussion session https://t.co/VUgsDMZI86
After lots of interesting presentations, QUARE '22 plenary discussion is about to start – looking forward to plenty of engaging contributions from our attendees. #xAI#quality#RecSys@SIGIRConf
According to Chen et al user studies are now most common method to evaluate the user’s response to #XAI of recommender sys-that's really good to hear. I guess the next thing we should work towards is making them more standardized and detailed #SIGIR2022@QuareWorkshop@SIGIRConf