The A-Z of Knowledge Technology (Part 2)

David J. Skyrme

Concluding our eclectic choice of KM technologies to watch.

N: Natural Language Processing. Its a very active field at the moment with lots of bright computational linguistic graduates starting up highly promising start-ups with one of three predictable results: going under, joining forces with others in the create-classify-retrieve information chain or being so good that one of the big portal vendors takes them over. (Rarely is the fourth option - growing a successful business in its own right - achieved). Most experts reckon that for auto-classification statistical (clustering, pattern matching etc.) algorithms perform best today, though NLP promises more in the future. The problem is going from syntactics (structure of language) which they do well at to semantics Meaning). Us humans have a lot of common sense that eludes computers. Think about it might interpret the phrase "Harry flew to New York on a cheap ticket"!

O: Open Standards. Do you know your SCORM from your SOAP? As we strive to share knowledge between computers, we need common definitions, not just of information entities, but business processes, XML schema etc. The advent of web services is also creating a growing need. At its lowest level basic exchange protocols like SOAP (Simple Object Access Protocol) are needed. Other level include description layer (WSDL, ebXML), discovery layer (UDDI) and management layer (ADS). Then there are resource description standards (RDF) and metadata standards (e.g. Dublin Core). Still confused. See the Web services primer or type your acronym into Webopdeia. SCORM by the way is Shareable Content Object Reference Model - used to describe elearning objects. As for common standards? Who was it who said: "The nice thing about standards is that there are so many to choose between."

P: P2P (Peer-to-Peer). You might immediately think of the now defunct Napster which made it easy to share files between PCs without downloading from a server. But P2P is growing up. Think of all the wasted time and bandwidth when sending email attachments. What if they were on someone else's PC accessible for you to download. One application of P2P is sharing idle computer cycles, as is happening with SETI and cancer research projects. In the field of knowledge management it lets you share your files collaboratively. Groove Networks (backed to the tune of some $60million of Microsoft funding) are a marekt leader in P2P collaborative workspace software for corporates. SmithKlineGlaxo has a trial to connect its 10,000 scientists this way.

Q: Question boards. You don't know the answer to something? Then pose your question on a question board, such as that which is part of AskMe's expertise finder. Using fuzzy matching, the system will try and find the answers to some frequently asked questions. It will also list names of people who from their written material it believes are experts on the subject. This kind of question/response system is increasingly used by companies to reduce the costs and improve the quality of post sales customer support. Have you sent a support email to Dell recently?

R: Retrieval. The days are rapidly disappearing (though you sometimes wouldn't believe it) when the result of search gives you 1 million unsorted hits. Using classification software and taking advantage of a taxonomy (see T), guided retrieval is becoming more the norm. You can restrict search to selected branches or levels of a taxonomy (as in Semio) or you can have your search results clustered into those that share related concepts (to see this at work try Northern Light. Such 'smart' or 'intelligent' searching is estimated to cut in half the time that people spend searching for relevant information.

S: Semantic Web. The Web as we know it today may be totally different in five years time. After all, the Web as we know it today is less than a decade old. The inventor of the Web Tim Berners-Lee is the driving force behind the W3C project to build the semantic Web "a gigantic brain" which understands relationships between web resources (through resource descriptions and ontologies) and will create an environment that supports intelligent agents seeking knowledge and performing transactions. Will it work? Well XML is here today and that was similarly a pipe dream just five years ago.

T: Taxonomy. A hot issue in knowledge management at the moment, simply to get better access to information through classifying and organizing it in a specified way. There is specialist software to help manual construction of taxonomies (e.g. Multites); others that support with collaborative taxonomy development (e.g. Wherewithal), and many more that do automatic classification (e.g. GammaSite) and/or work interactively with the user in fine-tuning terms after some initial automatic classification (e.g. Semio). No portal vendor who claims to be comprehensive is without some form of taxonomy support software.

U: Usability. It seems amazing, that despite years of research and a high level of HCI (human-computer interface) knowledge that there are many lousy computer software applications and hard to use intranet and Internet sites. Forget the hefty sums you have to pay to usability experts like Jakob Nielsen. Read his Alertbox articles at http://www.useit.com/alertbox/ and then go and check your website against the evidence based guidelines at http://www.usability.gov (US Cancer Research Institute). If you are feeling flush with funds, then invest in some end-user testing at a usability lab - or to do it on the cheap just observe users at the screen).

V: Visualization. Whether its knowledge maps. search results, taxonomies, conceptual relationships, then visual output can help many people enormously. At a fairly basic level many KM professional use MindManager as a day-to-day brain mapping tool. Many text mining tools, such as InXight, allow you to check relationships. Other interesting devices are contour maps generated by Aurigin's Themescape or maps of social capital (knowledge connections) that are mapped by Orgnet's InFlow. Such tools are not still as prevalent as they should be, but with number crunching hardware and high performance video drivers, there's no reason that they should grow more popular - perhaps one problem is that everyone has their own preferred visual format.

W: Who. In reviewing technology vendors for this A-Z, I also came up with an A-Z for vendors. There are familiar names like Automony, Broadvision, Convera etc. But there are many small niche companies like 3Path (P2P), Akiva (idea management), Bungo (virtual team workspace) and Cipher (intelligence gathering software). No space for them all here, but if I get enough interest, I might upload the one line entries from my database of around 300 vendors. By the way, things are a bit light at the XYZ end - I could only come up with XDegrees (which Microsoft acquired on 8th September so that leaves an X-gap!), Yagi and ZyLabs. Any better offers?

X: XML Topic Maps. Topic Maps are an ISO standard (ISO 13250) for describing entities and their relationships. There are three basic classes of XML tags - for topics (types and names), instances and associations (such as "made of", "part of", "lives in"). In one sense they are an extension of taxonomies that add additional richness i.e. they are ontologies. Proponents say that topic maps make it easier to understand the relevance of information and convey meaning. Some tools to create and manipulate Topic Maps are now available (e.g. K42 from Empolis) and Ontopia. To navigate through one look at Ontopia's demonstration Topic Map and navigator at http://www.ontopia.net. See also http://www.topicmaps.org.

Y: Yellow Pages. An expertise directory. Creating these has typically been a chore in many organizations, though there is software that makes it easier. We like the approach that BP used which is now available as SigmaConnect from Adept) (http://www.sigmaconnect.com).

Z: We are now at the end, or are we? The volume of information in the world grows exponentially. A year ago I got a 100 gigabyte drive for my computer. I though the storage space was infinite compared to what I had been using - but now I'm not so sure. So we have gone through: kilobytes (1,000); Megabytes (1,000,000) and now most of us are familiar with gigabyte (10⁹). Then there's:
10¹² - Terabyte - a few years ago the world's largest data warehouse was around 10Tb
10¹⁵ - Petabyte
10¹⁸ - Exabyte.

The estimate of data in the world in 2000 was 22Eb, but it is increasing at around 4Eb a year, so my nomination for Z is Zillion - "a very large but unspecified number".

This is the bottom line - if we have problems in finding knowledge now, then how will we cope when we have zillibytes. The need for knowledge technologies is as critical as ever.

I hope you have enjoyed our romp around the technology scene, and don't forget to send me some of your favourite ABCs and XYZs.

I3 Update / Entovation International News:
Current Issue - Archive - Knowledge Digest - Events - Subscribe