Topic Maps

The Topic Maps standard didn't take off as expected and is mostly forgotten. What didn't work was the fact it was a standard, not the Topic Maps model.

The standard was an attempt to get competitors to agree on a common way to interchange interconnected data through a common syntax. This goal was hampered by the fact that nobody wanted to interchange their interconnected data networks. Not because they did not the value of their interconnections, but for the exact opposite reason. They value the interconnections so highly that they did not want to exchange them with anybody else. Even non-profit organizations, including the US government, did not want to give away the work they did to third parties. In the case of the Department of Energy, the data that were topic maps were extremely sensitive and therefore highly confidential. In the case of the IRS, there was no incentive to give away all the work done internally by highly qualified experts to private companies who would benefit from their work without having to invest themselves in building high quality data connections. They were proud to claim that they own the trust that was produced among the communities of professionals using their data directly.

During the period where the standard was created, product vendors took the lead over information owners. They insisted on focusing almost exclusively on providing editors for the common interchange syntax, which was a narrow view that presented no clear advantage to their customers. Their efforts to provide free tools and to train their potential users on how to use them was misguided and eventually failed, after a few years of trying. This is comparable to the efforts from the other standard community, focusing on RDF, the World Wide Web Consortium's semantic web foundation, to train highly specialized technicians with the various layers needed to implement successfully a full data management system based on that standard. Many businesses that were initially interested by these standards were eventually discouraged, and the sector remained marginal, even if the RDF community has enjoyed a broader user base than Topic Maps.

What worked instead were numerous attemps to build interconnected data sets that provided immediate value to their users. The compliance with the Topic Maps standard was not an issue. In the case of the IRS, the availability of a standard is what allowed the architecture to be known. Once IRS understood that the information architecture defined in the Topic maps standard could be applied to their case and enhance the value of their information, the compliance with the standard became quite secondary and became a non-issue. In the course of the 17 year project, many improvements were made, and the way they were designed was not on the basis that there was a need to get more compliant with the standard, but on the contrary, that they were adapted to the subtleties and idiosynchrasies of their particular application context. Apart from the project leader, no person who was involved in the development of the information network knew it was an instance of a topic map. The name itself was not mentioned.

In 2010, Google bought a company called MetaWeb, that had created a knowledge base, called Freebase, which was created as a topic maps application, as Veda Hlubinka-Cook told me when we met. Google used this information to publish what was called the Google Knowledge Graph.

I had a meeting in one of the largest banks in America where I learned that they were using Topic Maps but they didn't want to advertise it because that very fact gave them a competitive advantage.

Similar approaches exist and are sometimes designated under the name of Concept Maps or Subject maps. Mind maps are a simplified versions of a similar concept, limited to a hierarchical structure. They are also related to neural networks.