Another Word For It Patrick Durusau on Topic Maps and Semantic Diversity

Get a cup of coffee and set your timer!

*****

The best way to explain topic maps is to say why we want them in the first place.

I have a set of food/recipe/FDA documents.

Some use the term salt.

Some use the term sodium chloride.

Some use the term NaCl.

Walmart wants to find information on salt/sodium chloride/NaCl, no matter how it is identified.

Not a hypothetical use case: Walmart Launches Major Initiative to Make Food Healthier and Healthier Food More Affordable

Topic maps would enable Walmart to use any one identification for salt, and to retrieve all the information about it, recorded using any of its identifications.*

How?

Topics represent subjects and can have multiple, independent identifications of the same subject.

That sounds cool!

Can I talk about relationships too? Like they do in the tabloids? 😉

Well, sure, but we call those associations.

Let’s keep with the salt theme.

What about: Mark Kurlansky wrote “Salt: A World History”.

How many subjects to you see?

Mark Kurlansky (subject) wrote “Salt: A World History” (subject).

Two right?

But Mark is probably in a lot of associations, both professional and personal. How do we keep those straight?

Topic maps identify another three subjects:

Mark Kurlansky (subject) (role – author) wrote (subject – written-by) “Salt: A World History” (subject) (role – work).

Mark is said to “play” the role of author in the association. “Salt: A World History” plays the role of a work in the association.

Means we can find all the places where Mark plays the author role, as opposed to husband role, speaker role, etc.

One more thing rounds out the typical topic map, occurrences.

Occurrences are used to write down places where we have seen a subject discussed.

That is we say that Mark Kurlansky (subject) occurs at:

http://www.bookbrowse.com/reviews/index.cfm?book_number=960

(How am I doing for time?)

That’s it.

To review:

1. Topics represent subjects that can have multiple identifications.
2. Use of any identifier should return information recorded using any identifiers for the same subject.
3. Associations are relationships between subjects (role-players) playing roles.
4. Occurrences are pointers to where subjects are discussed.

Exercise:

Choose any subject that you like to talk about. Now answer the following questions about your subject:

1. Name at least three ways it is identified when people talk about it.
2. Name at least one association and the roles in it for your subject.
3. Name at least three occurrences (discussions) of your subject that you would like to find again.

Congratulations! Except for the syntax, you have just gathered all the information you need for your first topic map.

*****
* The salt example is only one of literally hundreds of thousands of multiple identifier type issues that confronts any consumer of information.

Walmart could use topic maps to collate information of interest to its 140 million customers every week and to deliver that as a service both to its customers as well as other commercial consumers of information.

A contrast to the U-Sort-It model of Google, which delivers dumpster loads of information, over and over again, for the same request.

PS: There are more complex issues and nuances of the syntaxes but this is the essence of topic maps.

The New York Times, reported Sunday, Hiding Details of Dubious Deal, U.S. Invokes National Security extraordinary fraud in defense contracting.

Gen. Victor E. Renuart, Jr. of the Air Force, former commander of the Northern Command is quoted as saying:

We’ve seen so many folks with a really great idea, who truly believe their technology is a breakthrough, but it turns out not to be.

OK, but the technology in question was alleged to detect messages in broadcasts from Al Jazeera. (I suppose something more than playing them backwards and hearing, “number 9, number 9, number 9, ….”)

The fact that some nut-job wraps up in a flag and proclaim they want to save American lives should not short circuit routine sanity checks.

Here’s one where topic maps would be handy:

Build a topic map interface to the Internet Movie Database that has extracted all the technologies used in all the movies listed.

Give each of those technologies a set of attributes so a contracting officer can check them off while reading contract proposals.

For example, in this case:

• hidden messages
• TV broadcasts
• by enemy

Which would return (along with possibly others): Independence Day.

That should terminate the funding review with a referral to the U.S. Attorney for attempted fraud or the local district attorney for competency hearings.

New technologies are developed all the time but non-fraudulent proposals based upon them can be independently verified before funding.

If it works only for the inventor or in their lab, pass on the opportunity.

BTW, a topic map of who was being funded for what efforts could have made the Air Force aware other departments were terminating funding with this applicant.

Caution: It is always possible to construct topic maps (or other analysis) in hindsight, that avoid problems that have already happened. That said, topic maps can be used to navigate existing information systems, providing a low impact/risk way to evaluate the utility of topic maps in addressing particular problems.

I was puzzled by a recent comment concerning the “..uncertain future of topic maps…?”

Some possible sources of an “uncertain future” for topic maps:

1. Unchanging one world language and semantic adopted universally, plus conversion of all existing data into that language/semantic.
2. Universal agreement to ignore semantic differences. Increase in death rates stabilizes the world’s population at 17th century levels.*
3. Extinction of human race.

I hate to be disagreeable, ;-), but I really don’t think any of those three causes are likely to imperil the future of topic maps.

To be honest there could be a fourth cause:

4. As a community we promote topic maps so poorly or create such inane interfaces that users prefer to suffer from semantic impedance than use topic maps.

And there is a fifth cause that I don’t recall anyone addressing:

5. There are people who benefit from semantic impedance and non-sharing.

I think we can address #4 and #5 as a matter of marketing efforts.

Starting by asking yourself: Who benefits from a reduction (or maintenance) of semantic impedance?

*****

*No need for alarm. That is just a wild guess on my part. Has the same validity as software and music piracy figures. The difference is I am willing to admit that I made it up to sound good.

This makes my 1,000th post to Another Word For It.

I wanted to take a moment to think about where I would like to go next with the next 1,000 posts.

First, I want to become more systematic with the academic literature that is relevant to topic maps. As you have seen, it is spread from bioinformatics and computer science to library science and semiotics.

One of the things that makes articles/books/presentations slow to add is that I read/view all of them before actually posting them to the blog.

I suppose I could go just on titles/abstracts but then you would have to duplicate my wading through stuff that never makes it onto the blog. That doesn’t seem like a value-add.

Second, along with that, I want to provide more in the way of assistance in that jump between, “ok, topic maps sound great,” and having an operational topic map that provides a meaningful result.

Being more systematic about the literature isn’t going to be easy and providing generalized assistance for topic map authoring is going to be even harder.

My proposal, subject to your comments and suggestions, is to create what I am calling starter maps that have a lot of the basic infrastructure topics, types, etc., plus topics, associations, etc., for a particular domain.

For example, I might want to offer a starter map for say NASCAR racing, that has all the usual structural topics but also all the racetracks, races and competitors for the last decade. Plus relevant associations. Not everything someone would want but enough that getting visible results would not be all that hard.

A boost over the topic map fence as it were.

At least initially, those are mostly going to be in topic map format type data resources. Doesn’t really scale for semantically diverse resources but everyone has to start using topic maps somewhere.

Third, in addition to bare bones starter maps, I would like to create outlines of data sources that look particularly interesting.

Not nearly as easy to use as the starter maps but easier for me to author.

The sort of thing that points out subjects and relationships with subjects in other data sets that may not be readily apparent.

Fourth, I want to continue to discover interesting approaches and resources to bring to your attention.

Those will range from new technologies, such as NoSQL and graph databases, to new algorithms for data processing, to new ways to think about subjects and their identifications, etc. Some of them will prove to be very useful in connection with topic maps and others will prove to be less so, if useful at all.

The key criteria for that last item being that it is interesting material. It is impossible (IMHO) to say what information will or will not spark the next great idea about topic maps in my readers.

Finally, in order to devote the cycles necessary to make all of the foregoing happen, I need donations/sponsorships for these activities.

If you like what you see here on a daily basis and this sounds like a good plan, please use the donate button.

Sponsors welcome as well, please inquire. patrick@durusau.net

PS: Topic map consulting/teaching/training also available.

Astonish Us has to be one of the more persuasive pieces of any genre I have read in a very long time.

The advice you find there is applicable to funding for topic maps research, marketing topic maps to potential customers, promoting topic maps among fellow researchers, in short, every aspect of spreading the word about topic maps.

Go read the post and then come back here (using the back button) to post your comments on what astonishes you with a topic map on some particular subject? How do you want to convey that astonishment to others?