[10:07:01] We are working on a query auto-correction system. In order to train that, we would want to use real query data. Mike Pham suggested we could ask here on about possibilities an limitations using log data from the wikidata sparql endpoint. Who would be the best person to contact? [15:32:14] hey MichaelCochez -- yes, Mike reached out to us as well. I believe it's being discussed. If you have additional thoughts etc., probably use this mailing list (https://lists.wikimedia.org/postorius/lists/research-wmf.lists.wikimedia.org/) which is internal to the research team at Wikimedia Foundation [16:22:55] Research Showcase starting in ~8 minutes [16:35:40] Dashboard for the new search engine data: https://wiki-search-referrals.wmcloud.org [16:35:46] Some documentation about the dataset: https://wikitech.wikimedia.org/wiki/Analytics/Data_Lake/Traffic/referrer_daily [16:36:55] More to come via wiki-research-l in coming weeks [16:37:04] cool - thanks for sharing :) [16:55:30] I think this referes more to price than value. [16:58:46] isaacj: I have a question for Tiziano: this only studied English Wikipedia, does he have thoughts on the impact of the other language editions? [16:59:15] got it Nettrom! thanks [17:06:01] really interesting and valuable observation about how usage differs by where the reader is located, and thanks for forwarding the question isaacj! [17:07:08] yeah, it's always a good question. I'll say for Tiziano that in his past work on citation clicking behavior, they did study English, French, and Italian. i can't remember how much variation there was but that paper is here: https://arxiv.org/abs/1902.11116 [17:07:33] Or Wikipedia is ranked highest and people by accident click it and then find the official link on there ... kind of a deffered navigational query [17:09:17] yeah, flo_meier it's an interesting thought. hard to know to what degree it's "accident" vs. a learned behavior [17:09:59] hmmm as a stepping stone, I feel WP can be for disambiguation... that is, a search engine result might be give a lot of results, some related to what you want and some not, but you know that the WP article will be a more formal definition of the thing you're interested in, and also have an expectation of what'll be in the infobox [17:10:15] true ... could simply be a matter of habit too [17:13:01] dsaez: yeah, that's the challenging aspect of this economic value work. how do you answer questions about "value" without reducing it to just price / really narrow framings. because the easiest way is generally to ask "how much would this cost if provided elsewhere" or "how much would you pay for X?" but it's super hard to put numbers to "general value of this knowledge being available" or "value to AI" or some of those other more [17:13:01] nebulous forms of value [17:13:55] On this, I agree with Marx: https://en.wikipedia.org/wiki/Exchange_value [17:14:22] but for sure, it is an interesting result. [17:17:44] yeah, AndyRussG, i like the "formal definition of the thing you're interested in" piece. people are probably good at evaluating very quickly if a wikipedia article is relevant to what they were looking for. and I think to add to that, the curation of Wikipedia gives some sense of security [17:18:56] dsaez: indeed there are different and important ways to measure the value of of Wikipedia. As isaacj mentioned in the framing of the session, there is a real need to have more accurate numbers on the economic value of Wikipedia which can be closer to "price" in the model you linked to (as you said earlier). the questions of value and utility are very important ones, too. [17:22:51] isaacj: yeah good point about curaiton [17:26:20] Nick's repo: https://github.com/nickmvincent/UGCValueRoundup [17:27:36] dsaez: very much agreed, the economic aspect of external links on WP can be seen in a bunch of different ways... for example, can you ascribe a monetary value of the free knowledge provided by WP (including links to sites they're interested in) in the lives of users? [17:28:40] Question for Nick: the finding that wikipedia links are a lot less prevalent in medical query SERPs, especially for Google, is really interesting. What do you think is going on there? Does this have implications for the content of medically-relevant articles on English Wikipedia e.g. are they less likely to be reliable or quality articles, are they less relevant for users search needs, is Wikipedia perceived as untrustw [17:28:40] y on medical topics, etc.? [17:29:11] what about the equalizing effect of providing free knowledge vs. people having to pay for it? I don't doubt that greater knowledge equity has an overall positive impact on economies [17:29:57] AndyRussG, I guess this depends on philosophical/economic point of view. There are some schools that assumes that you can put prices to (almost) everything. [17:29:57] +1 to suriname0's question: there's been lots of research papers on the quality of medical articles [17:30:13] and work in WPMED on making sure reliable sources are used, etc [17:30:52] (and the fun trivia point about WPMED's definition of importance: any biography is low importance, because WPMED is about diseases and such) [17:30:53] got it suriname0 [17:31:54] But if water can be trade on the stock market , maybe those scools are the ones that are the more accepted https://futurehuman.medium.com/water-is-being-traded-on-the-stock-market-for-the-first-time-55c02bb616ed [17:32:05] *schools [17:32:19] dsaez: yeah... I feel the evaluation of value of the links from the hypothetical of what would companies have to pay for them obviates the systemic impact on many levels, but I'm not sure which economic schools of thought would take that into account [17:33:04] ^ heh wow thanks for the link! [17:55:19] thanks to the presenters for great work and presentations, and to moderators for moderating, great to be able to attend today! :) [17:55:54] thanks to all and cu around :) [17:56:14] thanks all for the engagement! [17:56:43] thank you all for the discussion. [17:57:00] Here my bad translation of Michel Serres explanation: [17:57:01] "If you have a piece of bread and I have a euro, and I go and buy the bread with my euro, I will have a piece of bread and you will have a euro, and you will see an equilibrium in that exchange, that is, A has a euro and B has bread, conversely, B has the bread and A the euro, so this is a perfect balance. [17:57:01] But if you have a Verlaine sonnet, or Pythagoras' theorem, and I have nothing, and you teach me, at the end of that exchange I will have the sonnet and the theorem, but you will have kept them. [17:57:01] In the first case, there is equilibrium. That is a merchandise (good/). In the second, there is growth. That is culture. " [18:57:11] dsaez hmmm thx! [19:02:15] yeah one thing that seems unfortunate is when the second case, that is, knowledge or culture exchange, is understood as only if it were an exchange of material goods, and then that view is used as a basis for deciding that someone owes something to someone else... which is still not to say that there isn't economic impact... aaaargh I need to learn more about economic theories.............