Exciting to see the responses. Based on replies/feedback, a few clarifications:
1. Anyone is welcome to participate even if you are not a current solr contributor or even a developer of any sort! Other viewpoints such as from end-users, data science, or even enthusiasts are all valuable. 2. We will broadcast publicly to the solr dev mailing list (will not be creating a new mailing list). I think we will test a monthly meetup format starting sometime in late June/early July 3. Better utilizing Jira and tagging DVS issues would be great. We can take a stab at a first iteration of this at first meetup -Kevin From: dev@solr.apache.org At: 06/03/25 05:44:31 UTC-4:00To: dev@solr.apache.org Subject: Re: Solr Dense Vector Dev Group Thanks, Kevin, and great to see interest in this! +1 on using Jira, components and why not, maybe even epic to have a better view on an "AI Roadmap". Updating Confluence will also be a desirable outcome (although I didn't even know about that roadmap, so even if decently recent maybe its utility has faded). We'll keep the dev list updated! Cheers -------------------------- *Alessandro Benedetti* Director @ Sease Ltd. *Apache Lucene/Solr Committer* *Apache Solr PMC Member* e-mail: a.benede...@sease.io *Sease* - Information Retrieval Applied Consulting | Training | Open Source Website: Sease.io <http://sease.io/> LinkedIn <https://linkedin.com/company/sease-ltd> | Twitter <https://twitter.com/seaseltd> | Youtube <https://www.youtube.com/channel/UCDx86ZKLYNpI3gzMercM7BQ> | Github <https://github.com/seaseltd> On Tue, 3 Jun 2025 at 04:44, Rahul Goswami <rahul196...@gmail.com> wrote: > Very interested in this space. A personal area of interest with Vector > search in Solr for me is to be able to utilize a local language model > instead of having to make remote API calls for text-to-vector conversion. > Have tinkered with creating a custom UpdateRequestProcessor for this using > ONXX runtime and a couple other approaches, but I am sure with > collaboration from the community, there is a lot of room for improvement. > > +1 to Ishan's suggestion of keeping it simple and keeping the > communications on the dev list itself. Just like we have a Solr meetup > every 3rd Wednesday of the month, to begin with, maybe we could fix a day > of the month for dev discussions specific to vector search? And send out a > reminder a week before the meetup. That way people who would have > "subscribed" to a focus group would anyway attend, and the rest won't miss > out on an opportunity to participate or be an audience if they so wish. I > am happy to volunteer for sending out this communication if it helps. > > - Rahul > > On Mon, Jun 2, 2025 at 6:31 PM Ishan Chattopadhyaya < > ichattopadhy...@gmail.com> wrote: > > > I am very much interested in the area and we have a plan towards making > > Solr suitable for vector search applications. > > > > > If you > > > would be interested in this, please reply to this email via the email > > > address you would like for communications going forward. > > > > However, I think we should keep the communications on this list itself, > > given how we have communicated in the past. > > > > > > On Tue, 3 Jun, 2025, 12:59 am David Smiley, <dsmi...@apache.org> wrote: > > > > > I love seeing this initiative/collaboration come from new community > > > members! > > > > > > Note that we have a Roadmap and it doesn't even have the word "vector" > on > > > it https://cwiki.apache.org/confluence/display/SOLR/Roadmap > > > IMO Updating this roadmap seems to me a useful outcome of this > > > collaboration. > > > > > > Another possible task is to create JIRA issues and use consistent JIRA > > > metadata for "vector" stuff. I think the JIRA "component" makes sense > -- > > > > > > > > > https://issues.apache.org/jira/projects/SOLR?selectedItem=com.atlassian.jira.jir a-projects-plugin:components-page > > > -- there is a bit of grooming to be done in our components by our > > > committers (we have permission). > > > > > > Should these meetings be organized just like the monthly meetups? > There > > > doesn't have to be a Confluence page, however IMO the other aspects of > > the > > > ceremony are useful -- publicize when the meeting happens and post a > > > summary somewhere (e.g. dev list) for those that didn't attend. If > this > > > isn't done, a bunch of people may feel left out just simply wanting to > > know > > > at a high level what's happening... but don't necessarily want to > attend > > > the meeting. Some months, that applies to me! > > > > > > ~ David > > > > > > On Mon, Jun 2, 2025 at 12:57 PM Kevin Liang (BLOOMBERG/ 919 3RD A) < > > > klian...@bloomberg.net> wrote: > > > > > > > Hello all, > > > > > > > > Alessandro and I were recently chatting about the current state of > > dense > > > > vector search in Solr, as well as potential directions going forward. > > One > > > > could argue whether there is a gap in the DVS feature space between > > Solr > > > > and other vector DB peers, however we both agree that one of the > > factors > > > > possibly limiting development in this area for Solr is lack of > > > visibility / > > > > coordination amongst contributors. > > > > > > > > To help encourage new improvements in this area, we would like to > gauge > > > > interest in a recurring group meeting to discuss dense vector and > > > semantic > > > > search. This would be an open invite to anyone who is currently > working > > > on > > > > DVS, using DVS, or simply just interested in DVS/AI powered search. > If > > > you > > > > would be interested in this, please reply to this email via the email > > > > address you would like for communications going forward. > > > > > > > > Since several of us in the Solr community will be attending Berlin > > > > Buzzwords in 2 weeks time, we aim to plan our first informal meeting > at > > > the > > > > conference in-person. After that communication/gathering can be > monthly > > > or > > > > ad-hoc depending on interest. > > > > > > > > Thanks all, > > > > > > > > Kevin > > > > > >
