Building Your Own Knowledge Base?

Flobster asks: "I just joined a company that has left the state of start-up and now wants to deliver professional end user support by the web as well as means to share product information internally. I know programming and databases and I have this fantastic MS Knowledge Base in the back of my mind - but just how did they organise that? All we have until now is plenty of word and visio based docs and release notes along with different FAQs, but the info the help desk is giving to customers is more or less never tracked or gathered somewhere. So where could i get a system or help to build a system that organises these texts, diagramms and future information so that i could search for them on a website? Anything from complete source code to books i should read would help!"

Several companies doing this...

[Anonymous Coward]

Not sure about the non-text side of things, but there are quite a few companies/products in the Customer Relationship Management space. A few OSS projects, and a bunch of commercial players like RightNow [rightnow.com] have pretty cool products

Re:KB

[tzanger]

Color me curious, but if you're short on time why rewrite the app in perl? For readability? :)

Because after coding in PHP for a while I have discovered that

• I don't need two scripting languages
• I like Perl's OO MUCH better
• I feel that Perl is more powerful

Good going for getting all that working. The only thing i'm not sure about is the inability to edit entries? if i have that correct.

Yes that's correct; if you need to make a correction that is one thing (spelling, /usr/local/bin instead of /usr/bin, etc.) but the idea of not allowing edits is to build up a base of "what not to do."

• i.e. if you say in an article that "rm -rf / is the best way to ensure security" and then want to change it to read "Actually a good security model is to run a kernel with a nonexec stack, don't install programs you don't know and be vigilant" then the system should instead add another article, marking the first one as supersceeded and (in another table) link up the superceeded article to the newer one. That way when I come along and i search for "linux security" It will pull up the new entry first but have a link to the "prior knowledge" at the bottom. That way I can't take a step back but I know that someone else did and that it didn't work.

It'll come along. I would like to have the system automatically determine if an edit should be a new article or simply a revision. Something along the lines of a character-granularity diff; if more than x characters in a row over y lines have changed, it's a new entry, otherwise it's just an edit.

I will have to check out amaya though; I had not heard of it (or rather I had but didn't take a look).

KB

[tzanger]

I've created my own KB already: http://www.mixdown.org/kb [mixdown.org]. This was done quite a while ago in PHP as a learning lesson and I've been itching for quite some time to finish it up. Three kids, a full time job and a couple of contract jobs make time scarce though. :-)

What the vision is:

• Full text entry (including accepting HTML and code)
• Adding graphics to the entries
• auto-keywording (basically eliminate all common words and phrases, the rest is keywords)
• manual keywording
• auto-rating (searches turn up entries, if the user clicks on the entry the link between the keywords used in the search and the entry retrieved is nudged up so that entry will rate higher next time)
• admin and peer rating - the traditional "was this useful to you?" type of rating
• supersceeding of entries - you can't eliminate an entry. If you want to make a change you can change it but the original entry is kept and marked as "supersceeded". Searches turn up the new entry but link to the old one as well. That way you build up not only knowledge but also "how not to do it" knowledge which is very important.
• Entry linking - Kind of like E2 [everything2.com] but with a datastore a little more connected. :-)

I've got the text entry and bits and pieces of the keywording and rating working. I'll be redoing it in Perl and finishing it up When I Get Time<tm> -- the current source is available to anyone if they want. As you can see if you go and look, I use it as a place to store things I have learned the hard way or found interesting and know I won't be able to find again.

This isn't directly related to business KBs but I figure this type of system would work most excellently in a customer-service and engineering environment.

Have a look at TWiki

[Empty Sands]

TWiki [twiki.org] is based on the WikiWiki concept, but adds version management (including of file attachment), access control, plugins and change notification. It has an active development communtity at TWiki.Codev [twiki.org].

Its used sucessfully by quite a few large corporates for their knowledge bases and support tools. See TWikiSuccessStories [twiki.org] for some specifics.

TWiki is based on the WikiWiki concept, but adds version management (including of file attachment), access control, plugins and change notification. It has an active development communtity at TWiki.Codev [twiki.org].

Indiana University

[macdaddy]

I did some research into this a few years back when we in the helpdesk at Kansas State University [ksu.edu] decided we should put up some sort of knowledge base or searchable FAQ. Or generic html-based FAQ was greatly showing it's age and relied heavily on the time of senior consultants to update it. We were understaffed and overworked. Anything to help was needed. I found this one site purely on accident and was greatly impressed by it. Indiana University [indiana.edu] built their own knowledge base [indiana.edu] with in-house knowledge and resources. The outcome [indiana.edu] is very impressive.

Another method of doing this is to use the FAQ-O-MATIC [sourceforge.net], written primarily by Jon Howell I believe. Jon's approach differs because the FOM is meant to be user driven. It can easily be closed up for in-house maintenance only but it's original intent was to be a user-driven and user-support tool to aide other users. I've used it and have been impressed with it as well. It gets better and better with each new release.

Any knowledge base type of tool will have one very important thing in common. They require time, and possibly lots of it. They require time to build, time to administer, time to update, and time to maintain. It's not always an easy task. If you can delegate some of the responsibility for bits and pieces of it down to others better suited to those bits and pieces, all the better. Making sure they keep up on their end of the deal though will need some superior oversight. Indiana Unv says they spend 300 hours a week on their KB. While it may not be exactly that much time and of that time they may not be working hard on the KB, they do spend a lot of time on it (I imagine they took their total number of student consultant * a percentage of their weekly working hours to get 300 hours--it's still a lot of hours). I never built K-State a knowledge base. I started more than once but I always ended up running out of time. If this is something they want you to do, make sure they know that it can use up a lot of your time, especially in the beginning. Make sure they acknowledge this and don't expect you to do this job and another fulltime job on top of it. Good luck!

--

Web Based Knowledge Base

[selectspec]

Regardless of how you want to do this, you'll have to convert and binary documents to at least HTML or TEXT. The cheapest way is probably to serve up all of your documents on the web and then set up a deal with a search company to index and catalog your site. Whatever fee they charge you will be cheaper than devoting a server to being a database and dataminer. http://www.freefind.com/plans.html currently offers this service for free if you have less than 32Megs of data. Anyway, the $100 you spend a year will be cheaper than the servers, setup, etc. However, if you want to do it the hard way go get aspseek from http://www.aspseek.org/. Its a GPL dataminer, search engine. Has everything you'll need.

Use Zope...

[supton]

It is a flexible open-source system that can be used for roll-your-own content systems; it uses a built-in object databse and cataloging system to allow you to create full-text and fielded searches. There are some Zope 'products' that already start to tackle the knowledge managment space (Knowledge bases, FAQ bases, Bug tracking, etc), and I'm sure that you could improve on them; rolling your own KB system, in this case, would be easy.

www.zope.org [zope.org]

Re:KB

[Your_Mom]

the current source is available to anyone if they want

Paint me stupid, but where?

IBM

[gavcam]

Look at the IBM Content Management software suite at http://www-4.ibm.com/software/data/cm/ [ibm.com].

EIP (Enterprise Information Portal) is a definite possibility if you need to search over multiple content stores.

Downside: price...

Simplifying support

[the_mom]

I don't have better advice about KB than the other replies here, but this site is worth taking a look- http://www.viewlet.com . We are going to use their professional edition to create very user-friendly help files for our software and deliver it over the web straight to the end user.

It cuts out some of the direct helpdesk interaction, which is more cost effective. Plus, endusers can be difficult sometimes, and the viewlet lets you "hold their hand" and show them the exact mouse moves and clicks they need to do!

Hire someone

[kruczkowski]

Why don't you hire someone to do the work? A great place to get someone is a highschool, someone that could come in, get the data, organize it and sort it in folders. Once you have that and the basic layout, you can easly teach them how to keep the system going.

I know if someone back in high school offered me a job like this, I would take it. Beside it looks good on a resume, Mantained Knowlage Base for XYZ Corp.

All Your Knowledge Base Are belong to Us!

[spongebob]

Ok it's cheese but I couldn't resist.

Re:KB

[shibboleth]

Color me curious, but if you're short on time why rewrite the app in perl? For readability? :)

Good going for getting all that working. The only thing i'm not sure about is the inability to edit entries? if i have that correct.

I'm pretty happy using Amaya for my personal database--I just maintain an html page of subjects, and click to read/edit a subject page: Amaya is a combo editor/browser, with easy, simultaneous access to the html source. Available from www.w3.org/amaya.

To jot quick notes, i have a "mergeme" subject to save me the time of opening the correct subject page. i periodically batch insert its contents into the correct subject pages after sorting it with a simple perl script.

To index the pages for searching, i use ht://Dig, which gives me the relevant filenames. But i need to do more work with her. The ht://Dig people may have improved her lately, i have to check, but, eg, their docs are flaky in places, a main developer is a testy, and support can be hard to get. But hey, it's a free product.

Here is one solution...

[Ayende Rahien]

Since you are impressed with MS' Knowledge Base(Correctly, I would say). Why not explore the offers from MS itself?
http://www.microsoft.com/business/km/default.asp
Is a place to start.