Grants:IdeaLab/Wikimetrics for local users
What is the problem you're trying to solve?
As a Wikimedia Italia volunteer I take part or help to organize local events. One of the question I often answer is if in the area where the event take place there are wikimedians who are not members of the chapter but maybe would like to be involved. It's not just about the "numerical success": these users can provide useful information about the area. I can find them very easily from their contribution pattern, and I've realized that it is not so difficult to target them.
The role of these users can go beyond local events and local volunteers for edit-a-thons: filling specific image gaps or looking for wikivoyage or OSM editors are all different examples of the importance of user-related wikimetrics.
I would like to give my expertise in finding them to everyone in a standardized way with a plug-and-play tool, so that everyone can reach local users easily with a clic.
As a secondary effect, this will also help to reduce the ecological impact of the WMF movement, using whenever possible volunteers who live close to the event the local chapter or user group is organizing.
What is your solution?
Geographical interest is actually more easy to pinpoint than thematic interest. Creating a tool to perform this type of wikimetrics, i.e. providing a list of users statistically close to a coordinate location, should not be difficult. Here are some examples that were performed "manually" over the years.
An Italian wikivoyage sysop (I also did, much less) contacted last year dozens of itWiki users with strong local interest to get help on wikivoyage about their areas, assuming that users very interested in a specific territory on one platform, would enjoy addressing the topic on another platform. It was succesful.
On itWikipedia I've performed with other users the first targeted mass message wikimetrics for the new content translator (aimed at frequent translators) and for the 2016 Wikipedia birthday Florence meeting (aimed at users form Tuscany). if you have a good knowledge of the structure of local communities, preparing a list of interested users takes no more than one hour starting from zero. Image what a well-established tool could do.
As a spin-off activity of Grants:IEG/Wiki needs pictures I've started to inform users with a very local "interest" on itWikipedia about wikidata tools for image uploading. Three was also a gap in the knowledge of wikidata to be filled for some of them. These geographic items were easy to explain to local users, so I started to contact them one by one. Local users can fix the gap very easily in their area. This is a very common "problem", we have similar initiatives such as this one scattered on different platforms.
In June 2016 I found some users for this request raleted to the area of Edinburgh. As you can see finding a user in the area for an edit-a-thon reduce the costs of the initiative. Manually, it took me approx. 20-30 minutes.
Than I discovered this proposal on itwiki w:it:Progetto:Amministrazioni/Comuni_italiani/OSM about targeting of OSM users. I think it could have been better planned.
In August 2016 in less than eight days we arranged a successful meeting of users from Tuscany with the new WMI regional coordinator User:Kaspo. We met new users, new faces. We were only six but at last other six were simply on holiday. If had been september, we would have gathered 12 people in one week (we sent ca 50 messages in 2-3 days, including inactive users), which is a good number for a Italian meeting. More importantly, we created new connections.
I've already contacted another user to perform the first wikimetrics about image upload locations on commons, but of course wikipedia articles or wikidata items have coordinates as well.
Tools such as dewkin already create maps. it may be slow, but in the end the important thing is to get the list of "local" users, it does not matter if it is created after 1 or 10 minutes. Some shortcuts are possible, for example removing inactive users or focusing only on autopatrolled users, and they have already been tested with success to produce useful outputs in short time.
Also, no specific knowledge of connection information is needed. Contributions are public and a lot of tools already process them to provide statistics. What we can do is just extract from those edits one or more "centers of interest" (COI), i.e. areas where some registered users show a bigger activity. This do not include IPs of course, and not all the users: some of them don't have enough edits, whilst others don't show such a clear focus (if you mainly patrol you edit almost everywhere). Still it is a considerable number of users and a good density on the overall map especially in areas connected with medium-to-large wikipedias. The "COI" of a user can be his/her residence or simply a place where (s)he lived, worked or shows a strong cultural interest. Which is still good: I don't live where my parents live anymore, but i can visit them if i can attend a local event, that's in any case useful information for me. Also if I like to write abour Scotland, I would love knowing that a local museum is looking for a WIR, because i can ask this user some sources or images I need for my editing. In any case, in my experience 65%-75% of time users live in the area where they focus their edits.
In my experience the number of users who answer targeted mass messages vary from 10% to 33%. It may seem a limited percentage but on a big set of wikipedians, it does make a huge difference. Project talks and village pumps target only a specific fraction of users, if you want a tool or an event to be known deeply you need to spread it in as many places as possible. This way a completely different "density pattern" is established and the information penetrates deeply in the community
Also selecting and contacting users one by one manually is a long process. Even preparing a list for a mass message takes some time and often some technical skills ("querry"). Having a tool that produce an elegant output is much simpler. You got the list of user active on a platform around strongly related to certain area, than you submit a request for a mass message according to the local community rules. Very easy.
I hope we can have at least a brainstorming on the topic. is anyone else working on something similar?
- Prepare a tool that can produce a list of users interested or focused on specific area just inserting the coordinates or the related wikidata item (es d:Q2044)
- prepare a tool that given a list of selected users (es: autopatrolled or recently active on one platform) produce a list associating to their username one possible "centre of interest" (coordinates and standard deviation) of their activity if enough localized. Probably even a second centre if the analysis is robust and there are a lot of edits.
About the idea creator
I am an active cross-wiki user. I am already involved in a grant about mapping of requested images on wikimedia platforms and this idea is a sort of spin-off based on my previous experience. I'm also active in local events for Wikimedia Italia. I speak 5 or 6 languages at a good or fluent level.
I have been in contact with Alex for this project. It's interesting a little bit complicated to understand but I think that it can be done and it can generate an interesting tool. I asked him to apply it to Switzerland where there is a very complicated community with a lot of differences and for us it would be an booster application for the activities with the community where the linguistic differences cannot help us because our impact is traversal in different linguistic projects. Ilario (talk)
Expand your idea
Would a grant from the Wikimedia Foundation help make your idea happen? You can expand this idea into a grant proposal.