[08:16:13] yes, office hours will be later today (6pm UTC) https://www.mediawiki.org/wiki/Wikimedia_Research/Office_hours [08:16:53] stelt hopefully see you then [17:50:40] office hours in 10 mins [18:00:16] o/ hi all [18:00:45] Hey everyone, welcome to office hours for the next 1 hour. [18:00:56] If you would like to ask a question or discuss a particular topic, simply type in the chat. We will do our best to answer as quickly as possible (I will try to relay to specific members of our teams); but everyone else please feel free to chime in as you see fit. [18:02:16] One brief note before we start: If you are planning to answer a particular question, we encourage to signal by using the “/me” statement, e.g. by typing “/me is working on a response to XYZ”, which will appear as [18:02:22] /me is working on a response to XYZ [18:02:32] * mgerlach is working on a response to XYZ [18:03:10] hi all :) [18:03:23] so, let's go ... [18:05:08] hello! I am attending for the 1st time, without specific questions in mind, just to know what those Research Office Hours are [18:05:49] hi mariedestandau [18:07:11] Hello world [18:07:30] What is wikiresearch up to? [18:08:07] oh hi mariedestandau ! is this your new paper? https://twitter.com/WikiResearch/status/1265630610535710720 I thnk djellel might be interested :) [18:08:41] mariedestandau: no fixed agenda from our side. happy to discuss any questions about wiki-related research/analytics [18:08:42] I think I have some nice ideas for applications on wikimedia content and love to hear why I'm wrong, or who will help, or anything in between :-) [18:10:27] hi steltenpower: curious to hear your ideas. lets see if we can be of any help : ) [18:10:51] Here's one: https://github.com/steltenpower/KnowledgeDock [18:11:12] * mgerlach looking up the paper [18:11:19] Here's two: https://github.com/steltenpower/Train-Of-Thought [18:12:37] And the last: https://github.com/steltenpower/connecting-REFI-QDA-and-LODcloud [18:12:42] hi all [18:13:05] steltenpower: can you give some context to the links [18:13:11] miriam yes, it is a preprint, happy to have feedback [18:14:57] KnowledgeDock would be having context-specific help to embed RDFa in your primary writing flow, without having to know much at all about the technology [18:15:02] are these ongoing projects of yours? what type of feedback are you looking for? [18:16:15] maridestandau yes, djellel is the best person to give feedback I think, but also i'll take a look! Great work! [18:17:05] I have enough ideas now, so want to actually implement some now. Useful feedback would be if they make sense, or links to very similar things preferably open source. If it makes sense I would be extra cool if someone wants to help. [18:17:05] * mgerlach is taking a look at the github-links [18:17:58] Train-of-Thought is querying and creating a knowledge graph while multi-touch mindmapping [18:18:42] The last one is connecting a standard for qualitative research tools onto the LOD cloud [18:19:10] mariedestandau: maybe GoranSM from wmde can comment too (not sure he is around) [18:19:17] * steltenpower ducks [18:21:54] * GoranSM looking at the mariedestandau paper now [18:23:46] steltenpower: this is a lot of information. at least for me, this takes some time to process in order to give useful feedback. [18:24:35] I was wondering, do you know about work regarding offensive content for living persons.I am thinking about not the obvious offensive, but sometimes the biographies are a selection of mean details about the persons, each is sourced and does not qualify for deletion, but the sum of them can be really mean. [18:25:38] On Twitter and GitHub I go by the same nick so if I'm not here and you want to send me feedback ... [18:25:40] mariedestandau, I haven't heard about something like that, but would be interesting. [18:26:37] Thanks, this is something I would be interested in working on in the future (but I have to write my thesis first, just thinking for later) [18:26:53] got you [18:27:02] what is your thesis about? [18:27:26] visualisation / interaction techniques for Linked Data [18:28:08] mariedestandau, i am checking if there exist a template for the case you mentioned [18:28:27] mariedestandau: In your tool, the clusters of entities to the left activate a list of properties to the right that are missing from the entities in the respective cluster, right? [18:28:53] Hi mariedestandau, I'm also interested in those topics of interaction design with linked data. Where can I find more information about your work? [18:30:15] Are you looking at any specific domain, or linked data in general (e.g. Wikidata)? [18:30:24] steltenpower: ok. and thanks for sharing those ideas. general feedback will require a bit more time. are there any specific aspects you would highlight/discuss? [18:33:07] mariedestandau: maybe you can check articles tagged with - https://en.wikipedia.org/wiki/Template:Systemic_bias; https://en.wikipedia.org/wiki/Template:Cherry-picked; https://en.wikipedia.org/wiki/Template:Unbalanced; https://en.wikipedia.org/wiki/Template:Undue_weight [18:33:16] GoranSM on the left the map represents all entities, clustered when they miss the same attributes. Then the first column shows all properties for all entities. And the next one all properties for a selection. You can select clusters and compare the properties of a subset with the whole collection, to try to understand why they miss attributes [18:33:47] miriam thanks!! [18:34:29] Nothing specific now [18:34:34] mariedestandau: anytime, this is super interesting! [18:34:40] sergionunes https://www.lri.fr/~mdestandau/index.html I started with Linked Data in general, but for the last project the user study was with wikidata users [18:34:55] o/ sergionunes [18:37:29] mariedestandau Thanks, will look. [18:38:15] sergionunes: are you working on similar things? [18:41:47] steltenpower: I took a look at your ideas and they're definitely interesting with lots of potential. But they're fairly early on, and it's not very clear to me how the data model that would have to underlie your concepts would be different from Wikidata's data model (which is basically storing triples in a funky format). My advice is to pick one idea, get very concrete with it and figure out a 1-5 minute way to pitch it to a [18:41:47] potential user. I'm happy to help you test your pitch, if you can convince me to use your tool you'd be in a great position since I'm grumpy and I only use vim :) [18:42:21] mgerlach: Yes, I am involved in a project were we are exploring user interaction solutions (specifically for searching, browsing and adding) for domain of the national archives. [18:43:47] sergionunes great, for which country ? [18:44:21] mgerlach: We are also working on a specific ontology for this domain. Country: Portugal. [18:44:22] milimetric, thanks for that feedback. I'll carry on then. I'll be back here when I have developed some demo (code or video) [18:45:46] also wondering: are there guidelines in Wikidata about how to balance precision and `reasonable amount of data' ? and related research ? [18:46:21] (thinking of qualifiers) [18:50:47] mariedestandau: sorry I dropped out of the session for a moment. Can you clarify on the relationship between "precision" and "reasonable amount of data", please? [18:52:07] mariedestandau: I mean, I am not sure if I understand exactly what you need. [18:52:43] I mean when a contributor adds a statements, should he add all possible qualifiers, or should he try to think of which will be really useful, and if so, on which criteria should he decide? [18:53:45] mariedestandau: Got it. Let me see if we have any guidelines on that. [18:55:14] sergionunes you might know already about the PIAAF project with french national archives https://www.ica.org/en/semantizing-and-visualising-archival-metadata-the-piaaf-french-prototype-online [18:55:30] mariedestandau: Currently, this is the best that I can offer: https://www.wikidata.org/wiki/Help:Qualifiers#How_to_add_qualifiers - but give me some more time to search through the Wikidata docs [18:57:05] mariedestandau Yes, thanks nonetheless. [18:57:11] mariedestandau: Also wikidata.org/wiki/Wikidata:Verifiability#Using_qualifiers, but not strict policy on qualifiers found yet, and maybe we don't have one indeed... [18:57:21] is it a concern that people might add too many details (leading to unnecessary massive amounts of data) or not at all ? [18:57:46] First time here. Do you also address Wikipedia issues or mostly Wikidata? [18:58:08] mariedestandau: Of course it is a concern, however, not that I am aware of any strict constraints imposed in an attempt to protect Wikidata from overcrowding information [18:59:15] sergionunes: not restricted to any in particular [18:59:31] GoranSM1 do you think a tool to help people make decisions and be consistent on this could be of interest ? or a first step would be to have guidelines and maybe it would be enough ? [19:00:17] mgerlach Thanks. Logging off now. Bye [19:00:23] sergionunes: do you use wikipedia / wikiedata on those projects? [19:00:55] mariedestandau: Typically, the process starts with the community putting forward some sort of request, following public discussions (mailing list, perhaps). Then the requests are prioritized and finally a choice on what will be developed and when is made. [19:01:22] sergionunes: thanks for joining [19:01:50] GoranSM1 thanks! [19:01:58] mariedestandau: So I would say most probably the process would start with (a) rising an issue publicly, (b) discussing it, (c) having a wiki page for it, (d) building an archive of the discussion on the respective Talk page, (e) putting forward the request to develop something to resolve the issue (Phabricator, I Guess). [19:02:40] thanks everyone for joining the discussion [19:02:44] Ok I see. Thanks a lot for all answers [19:02:45] mariedestandau: the Wikidata mailing list is, I'd say, probably the best place to open a discussion: https://lists.wikimedia.org/mailman/listinfo/wikidata [19:02:52] mariedestandau: You are welcome! [19:03:04] and for sharing your projects, ideas, and thoughts [19:03:25] hi mariedestandau, paper looks interesting, I have to dig in it! I've worked on class incompleteness in Wikidata. [19:03:31] yes, thanks a lot!! [19:04:07] hope to see everyone for the next edition in june (most likely june 24) [19:04:37] djellel will try to come to the next edition to know more about your work on it [19:04:49] goodbye [19:04:57] Bye everyone [19:05:11] feel free to reach out here if you have follow-up questions or comments : ) [19:06:34] yes, mariedestandau, please feel free to reach out in between :)