Most people have judged about Wikis in a negative way. Wikis are seen as the most stressful content management system ever. Some smaller Wikis are available for amateur topics, like bicycles and LInux but in most cases the amount of articles in these wiki is smaller than 500. The problem is, that apart form 3-4 users nobody is interested to write articles for such a wiki. The problem is not, that the users are unfamiliar with the Markdown syntax or a in fear from the version control system. But the entry barrier is, that they simply doesn't like content wikis at all.
A naive approach for creating a Wiki for the Artificial Intelligence community is to install a wiki on a public webserver and start to write some basic articles about neural networks, expert systems and robotics. The problem is, that creating such articles need time. Suppose after one month the admin has created 3 basic articles plus an extra article in which he is explaining what the AI wiki is. From a technical point of view, this project will work great. The admin can put the 4 articles into the wiki and it can be read by 7 billion people in the world. The problem is, that the expected traffic of such a wiki will be zero, and the propability that the admin will loose after a month his motivation is high.
So what is wrong the AI community wiki? The problem is to see to focus only on a content wiki. The much better idea is to build an AI social network with a mediawiki installation. The only thing what is different over a content wiki is the self-description. A possible introduction article for an AI Social network is given here:
Hello World, this is the AI social network. The world is invited to post links to existing AI ressources outside of the wiki. The requirement for submitting a link is:
1. Only links to new postings are allowed, which means that the general link https://trollheaven.blogspot.com/ is forbidden, while a URL in the format https://trollheaven.blogspot.com/2018/01/why-is-marketshare-of-windows-so-high.html is allowed
2. The URL must be fresh, that means it must be created within the last month
3. Additional to the link a title, and small description is desired, the overall posting should look like a Facebook post which was produced by a Autoposter
4. Only links about Artificial Intelligence are allowed
The result is, that the AI will become a playlist curator. Different users are posting URLs to existing content and they comment and judge about the URLs. If somebody likes to upload a paper which was self-written it is not allowed to do so in the AI wiki. What he can do is to upload the paper to it's own blog and post the URL into the AI wiki.
Unfurtuanatly, this AI Wiki description is not tested yet in reality. A project which comes close to the idea is Wikinews. Wikinews is working similar to a social network. The users can post a URL. But Wikinews is very authoritative about the quality of the URL. How an AI community can work in reality depends strongly on the guidelines and the admins who are fulfill the rules. The good news is, that for an AI social network there is no need to create new content. Because in the internet tons of useful ressources are already there which have to be peer reviewed. For example, the RSS feed of the latest arxiv papers is endless, if the complete stream or only a part of it would feed into a social network the reader will become a lot of fun. Because they have to read all the papers. Otherweise they can't decide to upvote the content.
No comments:
Post a Comment