« An alternative approach to Knowledge Management | Main | BPMN will kill BPEL. Will it kill BPM too? »

In the previous blog entry, I discussed the limitations of the RDF stack from W3C for knowledge management (KM) in business (if you're new to this blog, it would be worth reading that entry before continuing). This is not to say that the RDF approach is somehow “wrong” – far from it - just that it is worth exploring alternative approaches to KM. And, as we shall see, there is such an approach which not only has its own advantages but for which powerful open source tools are emerging.

The approach in question is Topic Maps. Topic Maps originate from OASIS, and have been ratified as an ISO standard, with an XML representation known as XTM (XML for Topic Maps). Roughly speaking, it is possible to represent most knowledge either using the RDF stack or using a Topic Map – techniques have even been proposed for automatic conversion from one format to the other, though there are technical difficulties inherent in doing this. However, the intentions of the two approaches are different, and it would be more sensible to view Topic Maps as complementary to the RDF stack, than as interchangeable with it.

In fact, the RDF and Topic Maps standards are of about the same vintage, dating in the first case from about 1999 and the second from about 2000. The two techniques were developed in parallel, by teams unaware of each other’s work, and each team looked at the problem of knowledge representation from quite a different angle.

RDF is essentially an implementation technique for the Semantic Web: “The RDF specifications provide a lightweight ontology system to support the exchange of knowledge on the Web”. Topic maps, on the other hand, are focused heavily on making it easier to find what you are looking for: “Topic Maps is a very explicit form of knowledge representation and it forces you to consider some of the issues that are critical when it comes to optimizing findability. It focuses a lot on how to name and how to identify subjects.”

In a nutshell, RDF is aimed at organizing information resources into predefined categories, whereas Topic Maps are aimed at exposing the internal structure of a particular body of information. If RDF is for directories, Topic Maps are for indexes.

Topic Maps are in some ways cuter than RDF. The principles are few and simple, yet enormously sophisticated structures can be created using them, some of which are hard to represent using the RDF stack without invoking very heavyweight techniques. Despite this, there is far less tool support available for Topic Maps. However, there are enough tools available – including open source ones - to make it worth considering Topic Maps for business use.

Why? I discussed in the previous blog entry how RDF-based tools have two particular disadvantages for business use – that they tend to categorize information into a hierarchy rather than a more flexible network structure, and that it is difficult to keep track of “old” knowledge (necessary to justify decisions or re-use knowledge, for example). Topic Maps have neither of these limitations.

The basic elements of Topic Maps are Topic (a subject or category), Association (a relationship between two or more specified topics), and Occurrence (information relevant to a specified topic). It turns out that this humble basis is enough to create complex webs of knowledge – mainly because the concept of Topic is general enough to represent any “thing” whatsoever. A topic is an information resource in its own right.

Hence, Topic Maps are not created via annotating each document in a corpus of information. Rather, a separate XTM file, embodying the network of relationships interwoven into the documents, is generated to represent the knowledge they contain. This means that it is simple both to categorize data into a network (as opposed to a tree), and to preserve previous states of enterprise knowledge (just keep the old XTM files you generate). Both are significant advantages in knowledge work, and are easier to implement with Topic Maps than with the RDF stack.

A nearly complete list of open source Topic Map software shows tools with various different applications, including graphical visualization, omitting only one other tool that I know of. A complete Content Management System based on open source Topic Map tools and an XML database is described by the IVS research group. A freeware (but not open source) mind mapping tool that imports and exports Topic Maps is ThinkGraph, and this is an area to watch – there are interesting analogies between concept mapping and Topic Maps, though mind mapping tools tend to view topics in a tree structure rather than as a network.

TAKE AWAY

The enterprise looking to develop a KM strategy and/or reduce its current expenditure on high-end commercial KM applications would be advised to consider Topic Maps as a representation technique, either as an alternative or a complementary approach to the RDF stack.

In future blog entries I will discuss open source tools that can be used to automatically generate Topic Maps from unstructured data – including the Web. Before doing so, however, I will be turning to the world of Business Process Management, looking at deep changes that are on the way, and showing how to prepare for them.

Posted by keithhb in Knowledge Management |Digg This|Add to del.icio.us