The Taxonomy Tango
Jennifer Zaino
SemanticWeb.com Contributor
“See a need, fill a need.” So says Mr. Bigweld, founder of invention factory Bigweld Corp. in the animated film “Robots.”
Well, Mobile Content Networks has seen one in the semantic web area and may well fill it. The company has filed a patent for its multi-taxonomy management method and tool. The application runs against multiple semantic web taxonomies to facilitate the exchange of information between classification systems within ontologies. That functionality is a core requirement for its own business, as a provider of real-time mobile search services that support wireless operators and content providers worldwide. It recently acquired the assets of Caboodle Networks Inc., which developed a multimedia, multi-service semantic recommendation platform.
“I’ve been working with taxonomies for a long time, and it seems no one really recognized the issue of multiple taxonomies except as a problem, but since it is the backbone of our system we actually had to manage that,” says CTO Phyllis Reuther.
“Because we are a federated search management platform, our taxonomy must map to the taxonomy of all our sources. We are basically a global application integrator in the sense of search.”
Its management platform, underpinned by OWL, puts an integration layer called a reference taxonomy between its source-level taxonomies and the customer-facing taxonomy that an operator would show to its subscribers. The operator then connects to MCN’s relatively rigid taxonomy, which it keeps in sync with the source data.
Mapping among taxonomies is going to be an important issue as the semantic web pushes forward, says Reuther. To some extent, industry and trade associations and library systems have tried to manage that by having their own reference taxonomy for their own content domain. But, “people are just getting comfortable with their own taxonomies and now they are realizing the world is full of taxonomies,” she says. Reuther says that she could easily imagine someone building a business on taxonomy mapping — say, for example, mapping the National Institutes of Health taxonomy to that of every new WebMD-like company that hits the market.
That potentially includes MCN, although it doesn’t have a productization plan for its management platform at this point.
The emergence of taxonomy exchange houses that will handle the complexity of dealing with multiple sources and the process of managing multiple taxonomies could be a bright one, if the semantic web takes hold. “They’ll just connect your application to that and then you have access to the taxonomy structures of everyone, but mapped to something you can deal with,” Reuther notes.
While Reuther suspects that what MCN has done may be a first patent in the realm of semantic technology, she wouldn’t be surprised to see others stake their own patent or commercial claims.
“I do think that in any new technology there is that discussion of how much of that is public domain and how much proprietary,” she says. “RDF and OWL have to stay open source, so that the point is that I need to have some standard transmission media so I can exchange information with others. We have to agree on what that format is, but the tools that sit on top of that that manipulate the taxonomies themselves, that’s every vendor for themselves.”
More immediately, MCN plans this quarter to fold Caboodle’s semantic technology into its services. Caboodle used agents to implement its search engine surrounding the recommendation engine, but the key point for MCN was that the recommendation was based on the semantic web, not on the standard collaborative behavioral model. “What’s interesting about Caboodle is that it lets us prime the pump, so to speak, on a collaborative behavior system by having an underlying semantic relationship among the types of information available,” she says.
With its hybrid approach, MCN should enable its customers to provide services that give those operators an edge, via semantic relationships that contextualize the query for cross-selling and up-selling opportunities, no matter how obscure or esoteric a request, or how sparse existing data for it may be. Additionally, it can help semantically define relationships that are one step away from the highest usage model — for example, getting to “yurts” following a search on “hiking” that semantically draw a relationship to tents, and ultimately leads you to yurts because tents are semantically related to other types of temporary structures. “It broadens the scope,” says Reuther.
Reuther says MCN has been getting a positive reception from customers, particularly in the U.S. and Europe, where enterprise search has conditioned consumers to want recommendations and personalization. But the Asian market is catching on too.
“Personalization and recommendation doesn’t have the same hype factor in Asia,” Reuther says, but they still find the functionality to be appealing.
The 
Eric Franzon
VP Community
Jennifer Zaino
Contributor
Angela Guess Contributor
semanticweb.com Twitter feed loading...
