Are there any benchmarks/papers looking at LLM based recsys vs embeddings/scores/all the other good science of #recsys based recommendations? At what point is a recommendation system not a search engine?

Daniel Roßner’s setup for his #demo of SPORE (Spatial #hypertext oriented #recommender system) at #acmht23. See his virtual poster at https://drossner.github.io/ht23-spore/
@HUMAN #recsys

NeMiG is a bilingual (en/de) news #dataset on the topic of #migration, which can be used, among others, to conduct controlled experiments on #polarization by news #recommendersystems. https://arxiv.org/abs/2309.00550 Joint work in the #ReNewRS project w/
@dwsunima @fizise
@kitkarlsruhe
via HeikoPaulheim@twitter
#knowledgegraph #recsys

RecRec: Algorithmic Recourse for Recommender Systems https://arxiv.org/pdf/2308.14916.pdf #preprint ML models underlying #RecSys are often opaque, making it difficult for users, content providers, and developers to understand why certain recommendations are made.

To address this problem, this paper proposes a recourse framework for recommender systems. A recourse is a set of actions that, if executed, would modify the recommendations (or ranking) of an item in the desired manner.

in #recsys models, multi-task or mixed-media models suffer from a bunch of issues like negative transfer. What is the major advantage of training larger and larger models to solve this issue and others? besides being able to infer using one model, i feel like recsys benefits from ensembles. In general also, this drive for large models is that because there are significant advantages or is this something else?

4/5
It won't be easy to fix recommender systems. Imagine transforming what evolved to be a ""casino"" into a public space, transforming ""users"" into citizens...Where to start? Panoptykon, ICCL and People vs BigTech investigated their most harmful features & call for change.

Fixing Recommender Systems. From identification of risk factors to
meaningful transparency and mitigation:
https://panoptykon.org/fixing-rec-sys-pdf

#EU #TechRegulation #RecommenderSystems #RecSys #Algorithms

Tomorrow we will be presenting our paper "Toward Job Recommendation for all" at #IJCAI in the AI for Social Good Track. It is about Job Recommender Systems.

Feel free to come and say hi and reach out if you have any questions.

Paper, Slides, Poster and Code available on my website: https://solalnathan.com/publications/

#ai #ml #ijcai23 #recsys #fairness #NewPaper

Awesome Ph.D opportunity on “#recsys for youngsters” advised by one of my favorite people, Sole Pera @ TU Delft. Please share and apply! https://www.tudelft.nl/over-tu-delft/werken-bij-tu-delft/vacatures/details/?nPostingId=4772&nPostingTargetId=13732&id=QEZFK026203F3VBQBLO6G68W9&LG=UK&languageSelect=UK&mask=external

New paper out in #humancomputerinteraction journal: A sensemaking system for grouping and suggesting stories from multiple affective viewpoints in #museums 50 days free access: https://www.tandfonline.com/eprint/AHINMIJGM93A9EM4IQFU/full?target=10.1080/07370024.2023.2242355

#hci #recommendersystems #recsys #affectivecomputing #storytelling #culturalAI

@academicchatter
@cscw @digitalhumanities

https://tandfonline.com/doi/full/10.1080/07370024.2023.2242355

New paper went up while I wasn't looking! @841io, Ben Carterette, and I wrote about the need to evaluate #recsys and IR systems by considering distributions, not just point estimates of effectiveness metrics. We outline the different sources of uncertainty in evaluation over which we can compute distributions, and provide a case study that derives insight from distributions that is obscured by standard point-estimate practice. Full code available. Share and enjoy! https://dl.acm.org/doi/10.1145/3613455

We’re still seeking applicants for a research project specialist / software engineer to help build a news #recsys platform to support user-facing research on news recommendation, personalization, etc. Happy to answer questions! Details and application info here: https://www.linkedin.com/jobs/view/3670339186

Please boost for reach!

I will attend #sigir2023 and present two papers, both related to #RecSys. The perspectives paper is on 24 July (1PM) and the TOIS paper on 25 July (9AM). My school has a few faculty position openings. If you want to know more, feel free to talk to me.

Hi all! We're hiring a research software engineer for a new NSF-funded #recsys research infrastructure project (hybrid position, on-site home possible at Minneapolis, Clemson, Philadelphia, Boulder, or Evanston). Please share! Details and application info at https://hr.myu.umn.edu/jobs/ext/356411.

It is now official, we have won KDD CUP this year.

https://discourse.aicrowd.com/t/announcing-the-winners-of-amazon-kdd-cup-2023-get-to-know-them/9001

#machinelearning #recsys

Hi everyone! Six more months passed since my last #introduction, so here is an updated one:

AKA: +mala, AiTTaLaM

Job: Doin’ trustworthy #AI @ moz://a.ai - more generally I love #teaching, no matter if to humans or machines :-)

Projects: 3564020356.org is the oldest (~22yrs 😅), #PicoGopher the most recent... Look around and find the rest! 😜

Interests: #bouldering #gopher #SelfHosting #opensource #reversing #fediverse #recsys #ML #solarpunk #CommunitiesOfExperience

Glad to announce that the paper "A sensemaking system for grouping and suggesting stories from multiple affective viewpoints in museums" (available @ https://lnkd.in/dU5S3uee ) - based on the results of the SPICE H2020 EU project https://spice-h2020.eu/ on the #deaf community of museum users - has been accepted in the special issue on #IntelligentSystems for people with diverse cognitive abilities of the Human–Computer Interaction journal (Taylor & Francis).

#AI #recsys
@chi @academicchatter

Tell me you’ve never operated a recommender system in production for any length of time without telling me.

I’ll start: you think #RecSys is primarily about models and everything else can be an afterthought.

short talk tomorrow, will perhaps be my most opinionated serious academic talk ever.

trying out some material for a short paper I'm working on this summer about how no, #recsys shouldn't diverse-SERP our societies into fascism, and we as researchers & practitioners should have an opinion on that subject.

Launch day is here! Presenting #mast, an #OpenSource #privacy first recommender and discovery tool for #Mastodon to help you find more people to connect with in the #Fediverse. Built with #Python and @sklearn (I love using @sklearn and their #documentation)
https://github.com/buckbaskin/mast
#recsys

For podcast fans — I joined Marcel for the latest episode of Recsperts to talk about fairness in search and #recsys: https://www.recsperts.com/episodes/16-fairness-in-recommender-systems-with-michael-d-ekstrand