While reading An Ontological Representation of Biomedical Data Sources and Records by Michael Bada, Kevin Livingston, and Lawrence Hunter, I realized an essential difference between ontologies and topic maps.
Bada and colleagues developed:
an an OWL-based model for the representation of these database records as an intermediate solution for the integration of these data in RDF stores.
That is to say they transformed the original records into a representation in OWL.
Which then allowed them to query consistently across the records, due to the transformation into a new, uniform representation.
Contrast that to topic maps, which offer an additive solution.
Topic maps enable the creation of an entity and the addition to that entity the equivalent identifications from all 17 databases.
Any other databases that become of interest can be added to the topic map in the same way.
Another way to say the difference is that ontologies set forth “a” way to make any statement, whereas topic maps collect multiple ways to say the same thing.
Which solution works best for you will depend on your requirements, existing efforts in your field, data that you wish to use, etc.
None of those considerations involve the software being sold by a vendor, advocated by devotees or similar considerations.
Any solution should fit your needs or you should simply walk away.
This brings to mind something I’ve heard the CTO at my place of work say more than once: OWL ontologies are snapshots. (Which, incidentally, is only something I heard him say after I introduced Topic Maps here — we use OWL in a very topic-mappy way.)
Comment by marijane — November 12, 2012 @ 8:57 pm
I am posting this for Steve Newcomb:
Some cultures and movements that habitually assimilate other cultures, while other cultures prefer not to assimilate, thus to maintain their purity. There are advantages and disadvantages to both attitudes, but over the long term, the incorporation of diverse elements is inevitable, and delaying such incorporation is often seen in retrospect as amounting to missed opportunities. (I observe that, as the world shrinks and the population grows, the purists tend to become more shrill and less influential.)
Some Americans vociferously oppose a multilingual United States, insisting that the official language must be English and that all government business must be transacted in English. For others, this is a non-issue; a multilingual society needs multilingual government. In fact, both sides have good arguments. The English purists want everybody to learn and use English, and when confronted by the enormous inconvenience this would create for a substantial minority of U.S. residents, they can argue that it is pure coincidence that such an outcome will be most convenient for those who already speak English. At least in theory, the net productivity of the entire economy would be higher if everyone spoke English. It’s possible that this theory is
correct!
Similarly, the idea that all information must be transformed as necessary in order to make it conform to some specific formalized world-view has good arguments in its favor.
However, here’s the rub: if everyone spoke only English, there would be a net *loss* of cultural diversity. And a net loss of culture. The world views with which English is imbued would retain their influence; the conflicting world views that imbue other languages would lose their influence. Survival depends on diversity, because the influences that will turn out to be the key to survival cannot be known before they are
actually needed. (My favorite example is the genetic oddity that bubonic plague survivors conferred on a few of their distant descendants, who are now the small present-day minority that can be exposed to HIV without developing any resulting illness. In more cultural terms, consider the many tantalizing hints of both common and
diverse origins in each of them. Consider the diverse origins of alphabets, numbers, and other symbols. Nobody involved in any of those developments knew where we’d be today, much less where we would be if only they had made different choices.)
An insight can be only imperfectly preserved apart from its cultural context, and therefore it’s best to keep it there, rather than moving it elsewhere and discarding any baggage that won’t be understandable in the new context. The original cultural context deserves respect and honor — just as much as any other cultural context where the insight might be useful. The Topic Maps attitude is, as Patrick puts it, “additive”: an insight’s cultural context need *not* be discarded in order to exploit
it in some other cultural context (such as a formal ontology that offers processing convenience within the scope of its own purity). Instead, what’s really required is to face the fact that no fewer than *three* cultural contexts are involved in making one culture’s insight available in the context of another:
Culture #1: the culture wherein the insight originated,
Culture #2: the culture that needs access to the insight, and
Culture #3: the world view of the person(s) who take responsibility for honoring Culture #1 in its own terms, for honoring Culture #2 in its own terms, and for faithfully translating the insight from the terms of #1 to the terms of #2.
I say, “no fewer than *three* cultural contexts” because carrying the insight-cargo from #1 to #2 is not exclusively the province of Culture #3. Cultures #4, #5… may also undertake the same responsibility.
Once we recognize the importance of cultural diversity to the survival of humanity, we seek ways of avoiding the brutality of insisting that everyone adopt any single mode of thought. Then the question becomes, “How can we honor, defend, preserve, and *exploit* Cultures #1 through #n simultaneously, and yet support the identification and communication of insights within and among them all? So that everybody can benefit
and so that many enterprises of culture can prosper?” That’s the question that Topic Mapping seeks to answer.
The “subject proxy”, as described in the, uh, culture of Topic Maps, is the best answer I know of.
P.S. During the 1975-76 academic year, I was an Assistant Professor in a tiny rural college in the southern United States. One of my friends there was a professor of Religion. During a relaxed moment, he told me that, over the years, he had experienced the same little drama repeatedly. An enthusiastic student would bring his/her campus-visiting parents into his office, and during the ensuing conversation, the
parents would address him as “Reverend”. This bothered him, so at least in his early years, he would attempt to explain that “the study of Religion is a Humanistic discipline”. It never made any difference; he continued to be addressed as “Reverend”. Eventually, he shrugged, gave up, and answered to “Reverend” when that form of address was used.
I tell this story because it’s very similar to what I experience when I try to explain topic mapping. Almost everybody assumes that I’m an exponent of some particular religion — the “Topic Maps” religion — rather than one who seeks to honor and provide for the peaceful and mutually exploitative co-existence of all religions. By “religion” I mean all of the following: “data model” (including the so-called “Topic Maps Data Model” and all other data models and schemata), ontology, taxonomy, language (including formal and natural languages for all purposes including data modeling), etc. If I *am* an exponent of a religion, it’s simply this:
Ideas are real. They exist independently of any symbols that may be used as proxies for them.
If you’re willing to believe, or to pretend to believe, that ideas are real, then you’re able to understand the idea of “subject proxy”. (And not otherwise.)
Comment by Patrick Durusau — November 13, 2012 @ 7:16 pm