On 6/10/06, James Mohr <mediawiki(a)jimmo.com> wrote: I would think that somewhere there's a heated discussion on this.. probably archived somewhere in meta. However if there isn't, or it's dated and disorganized then I'd be willing to participate. My angle is a huge amount of "knowledge" with only a few users (mostly just me). I hate categories, and try to avoid lists, but string things together with small lists, searchability and breadcrumb links.

On Sunday 11 June 2006 02:48, Morten Blaabjerg wrote: That's more or less what we are doing. I work as a system administrator/operator/help desk for a company that provides software for online brokers as well as manages a number of datacenters for various customers. We have about 20 different customers, each with their own environments. A few are online brokers with two dozens machines in several network segements spread across multiple physical locations, connection to a dozen external systems,and so forth. At the other end are small Webshops with a single machine and an extra cold-standby. All told, we have over 1000 pieces of hardware that we manage. The asset management is done with a help desk tool, but it isn't set up to store things like configuration and troubleshooting information at the level we need it. We have over 150 different applications/programs that we use in one way or another. These rangein complexity from things like Oracle and MySQL to rsync. However, we are all expected to know how to configure, manage and troubleshoot each of these applications (plus the hardware). Then we have all of the administrative information like how to open service calls, who to notify at the customer's site when something happens, what to pay attention to in our daily datacenter walk-throughs. One obviously important aspect is find the information again. Currently we store things in MS-Word Documents using Visual Source Safe. Independant of problems storing information in monolithic documents like MS-Word, there is the problem of checking the out of VSS, editing them and checking them back in. Not to forget is the fact that only the department head and team leaders can add documents to VSS. That makes adding new information to this pool cumbersome at best. As a result people are less willing to store information there and either have their own repository (mine was in HTML files) or they simply don't write anything down. My Wiki at home is in English and the one at work is in German. For us, there is no need to have multiple languages. However, I also run the Linux Tutorial (link below) and have been considering moving it to a wiki. I have been working on the idea of having it available in multiple languages. One important things for me would be to have those part which are not yet translated into a specific language still appear in English. Or maybe a user configuration option that switches this on/off. It would be nice if the admin could define a set of categories which are then available from a drop-down menu. Alternatively, you could have a set of links on the edit page and clicking on them uses Javascript to insert the necessary text into the form. Personally, I good enough with cut-n-paste that I don't see it as a big problem. However, currently anyone can create a new category and you typically end up with a mess. For example, you have the categories Application and Applications. With a menu the admin could ensure that the name are consistant. At this point I am not too worried about the flexibility since it is open source and I am well-enough versed in PHP to make changes myself. :-) However, I agree with you that it will improve as time goes by. Considering how useful it already is, I am anxious to see how it is in a couple of years. I would like to see is better seach capabilities. It would be nice to be able to seach by category as well as a little more intelligence. For example, on my personal wiki I have a page that described what directories are backed up and how. So, I search for "backup" and nothing is found. I need to search for "backups". (or is this a problem with my site?) The idea of a site map is not a bad idea. There is obviously a lot of manual work to do (or not?), but that offers one more method of finding the information. I have run into the argument that the information **has to** be stored hierarchically. That there is some absolute requirement somewhere. Granted humans natually put things into categories, but I see that most of the belief that things must be in hierarchies is because that is the way we are used to storing files on hard disks. Thus we store files for our web servers like that. Also, up to now, most web sites have had some kind of hierarchy due to limited search mechanisms. I am not exactly sure what you mean by "top-level categories doesn't link to the new sitemap, since they do not belong in any category." What would be a top-level category on your site where you do not yet have wiki categories. I am asking because we are working on a start page with various blocks for "top level" categories. For example, administration, technical, customer, etc. To some extent, these are too vague to put into wiki categories, so I am not sure if it really makes sense to create wiki categories. <snip> What I would like to see is a hook that allows easy access to other data sources. For example, as I mentioned we have a help desk tool which keep tracks of assets, customer organizations and people, service contracts (internal and external) and so forth. It would be nice to be able to pull things directly out of the database. For example, on a given customer's start page, there is a tag {{ServiceDeskCI:MaschineName}} that pulls specific information for the configuration item (asset) or {{ServiceDeskCustomer:CustomerName}} which pulls specific customer information. Another thing is printing of lots of articles. We have a requirement to deliver a handbook to certain customers once a month (whether things have changed or not). Since we currently use MS-Word, it is fairly straight forward to create a PDF. Although I have written a script to convert the HTML pages on the Linux Tutorial to a PDF and it would be fairly easy to create a script that pulls pages out of the wiki, parses them and passes them to the same mechanism, it would be nice if this was built in. There was a post on the MW users forum from someone who creates "site books". To me, this would be another good use of a process like this. Granted something like that is not the goal of a wiki, but more like DocBook. However, DocBook does not have nearly the same features as MediaWiki and I see it would be harder to make DocBook fit for our needs like a Wiki as opposed to making a wiki do things that DocBook does (note that the DocBook knowledge base is a wiki and does not use DocBook itself). regards, jimmo -- --------------------------------------- "Be more concerned with your character than with your reputation. Your character is what you really are while your reputation is merely what others think you are." -- John Wooden --------------------------------------- Be sure to visit the Linux Tutorial: http://www.linux-tutorial.info

James Mohr wrote: <snip> You can append some keywords after Special:Search : http://en.wikipedia.org/wiki/Special:Search/foobar_variable You can use a space instead of the underscore. So you can probably use some mod_rewrite stuff to have: http//yourwikihost/search/ redirect to special search :) -- Ashar Voultoiz - WP++++ http://en.wikipedia.org/wiki/User:Hashar http://www.livejournal.com/community/wikitech/ IM: hashar(a)jabber.org ICQ: 15325080

----- Original Message ----- From: "James Mohr" &lt;mediawiki(a)jimmo.com&gt; To: &lt;mediawiki-l(a)wikimedia.org&gt; Sent: Sunday, June 11, 2006 9:38 AM Subject: Re: [Mediawiki-l] Knowledge Management and MediaWiki At first I tried to develop a useredited sitemap/index, which clearly has the advantage of providing additional information on the pages linked to. In effect though, this is clearly often not updated effectively enough, especially on a smaller wiki. So, I decided to have a go at using categories more efficiently. I edited the text at "Special:Categories" to contain our top-level categories - as well as all categories in the wiki (which are included automatically on this special-page). Our top-level categories are "Cultural Networks" (which covers everything from rockbands and periodicals to global media institutions and broadcast networks), "Persons" (all pages on individual artists, theorists, directors etc.), "Cultural Productions" (which covers every single publication, song or film in the wiki)... etc. My idea is that all pages belonging to a subcategory will also belong to one of these top-level categories, so effectively all pages in the wiki is in a category. We also employ more subject-oriented indexing categories, which are much more diverse, spreading over several namespaces and across other categories. These are completely non-hierarchical. It is of course also possible for pages to coexist, say in both a "network" category and a "cultural works" category - in the case of a periodical for instance, which is a publication, yet also an editorial institution. The top-level categories are very vague, yet deliberately cut out to describe the different kinds of content we have and want in the wiki. They also have the purpose to simply display the mass of, as well as what kind of information is in the wiki, for users and visitors(which can otherwise be difficult to get an overview of). The only (minor) problem with using "Special:Categories" as a sitemap is, that since our top-level categories aren't themselves in any category, they won't automatically give a link back to the Special:Categories (our sitemap) page (as all pages or categories in a category do). This is a minor issue, really, one which I expect could be solved comparatively easy (if by no other way, then just by inserting this link manually). What would be vastly more interesting, is to have more customizable category pages - for instance, entries in a category being able to display data on the pages in the category at a glance (say the first two lines of wikitext on a page, latest author, page views), as well as sorting the pages in the category for display after userchosen criteria. "Show me the latest edited pages in this category" - "Show me the most popular pages in this category" - "Show me only pages from this namespace, this date interval" etc... What I guess I am talking about is combining the very efficient cross-non-hierarchical category system with the MySQL requests of what is now hidden in very "blocky" and "not-so-flexible" Specialpages (which only is used for the wiki as a whole, and for the primary namespace in particular). Ideally I want all the functionality of the specialpages to be enabled for each category, each namespace and for all files in the wiki. This could be implemented with an expanded dropdown menu system, which is already in place for several Specialpages. This kind of expanded functionality would make MediaWiki ultimately the most powerful indexing tool you could find ever. I know this kind of functionality will probably not find its way into Wikipedia at present, because it will constitute too great server loads etc. But in the long run, I am not so worried about server load issues. On smaller wikis the load times will probably be less of a problem, and you could make this kind of indexing functionality optional, to be dis/enabled in the localsettings. Best wishes, Morten :-) -- Crews Cut Production Morten Blaabjerg Danmarksgade 97 - DK-5000 Odense C Tlf. 65 90 60 88 / 51 80 91 55 http://morten.crewscut.com morten(a)crewscut.com

On Monday 12 June 2006 12:02, Morten Blaabjerg wrote: <original snipped> That does not surprise me. First people will have their own perception of what is a "good" organization. In my experience, either you make the site map be automatic or you have someone manage it or a combination of both You always need an owner who takes responsibility for things. The way we have it organized is that for each area there is a "expert" and a backup. The expert either fills the KB or ensure that it gets done for his/her area. He/she is also responsible for reviewing the material. Even on projects as big as the Wikipedia, there are people who review articles. Potentially you could have a TOC for aspecific area which are created by the owner/expert and them simply include automatically into the larger site map. Sounds responsable. Why can't these be real categories? That's very similar to the way we are setting up things. However, we are planning to "force" a little hierarchy into the structure because that is what a lot of people expect and to ensure standardization. A good example would be Database_Troubleshooting exists in both the database and troubleshooting categories. However, we have *defined* database to be at a higher level than troubleshooting. Therefore the main page is Database_Troubleshooting and if someone wants to create the page Troubleshooting_Databases it is *required* to redirect to Database_Troubleshooting. The customers are even higher, so the main page would be CustomerName_Database_Troubleshooting and so on. This ensures that on Special:Allpages things are automatically organized hierarchically. At least to some extent. That seems to be similar to the way the Wikipedia Portal works. Look into sub-categories: http://meta.wikimedia.org/wiki/Category#Subcategories That might provide a different solution. I am not saying your's is wrong, just offering up an alternative. Well, you've gone beyond my experience with MediaWiki. Naturally, you could create a hack yourself to give you the functionality. There is already the ability to create dynamic article lists: http://meta.wikimedia.org/wiki/Dynamic_Article_List One parameters is which is the "Name of category where (including its children) all articles reside". This might be what you are looking for. If you sort by hits, you get the most popular articles in the category. I think it is getting there. As more people like us add their ideas about how to acces information and potentially our own hacks and add-ons, I see it coming. The only thing I see blocking it is some people's perception that it should stay a tool to create KBs like the Wikipedia. That is more or less "random" information. In my job, the information is much more "directed." I might even be tempted to say "solution oriented". Personally, I think the load would be minimal with the right indexes. -- --------------------------------------- "Be more concerned with your character than with your reputation. Your character is what you really are while your reputation is merely what others think you are." -- John Wooden --------------------------------------- Be sure to visit the Linux Tutorial: http://www.linux-tutorial.info