In December 2006, James Kim and his family were stranded, their car stuck in heavy snow. After a week, fearing for their survival, James headed out into the wilderness to find help. In the dead of winter, he was wearing nothing but street clothes. His family was found in their car two days later, alive and well. Two days after that, James was found in a ravine. He had died of exposure.
When I first heard this story, I said to a friend, “That man was a hero.” My friend disagreed. I found that a little disturbing. James Kim was clearly a hero. I was wondering what in my friend’s character was preventing her from acknowledging that fact. After some debate, our differences became clear. We were just arguing semantics.
For me, a hero is one who acts selflessly for the benefit of others. Those qualities of nobility and action are what define my personal meaning of “hero”. For my friend, “hero” embodies all those attributes, but one additionally: success. She certainly wouldn’t disagree that James Kim was noble and brave. He just didn’t succeed in his act of bravery. Hers was merely a more restrictive notion of heroism.
Everyone loads their thoughts and language with these types of deeply personal associations. At Primal Fusion, we refer to these subtleties of meaning as “personal semantics.” We can share a common definition of the word, but we will never share a common definition of its semantics. A tiny word like “hero” cannot possibly contain it all.
Is technology coming to the rescue? Can machines help surface and mediate these types of miscommunications? They should, but not until semantic data is played to its strengths. In this post and the last, I’ve been arguing for semantics as first-class data, as the data of information consumption and personal meaning. It’s far more than metadata in the service of search and retrieval. When you’re looking for a killer app for semantic technologies, why not embrace the most killer aspects of semantics?
I can’t explain precisely why I think James Kim was a hero, but I know one when I see one. If I asked my friends to share their personal perspectives on heroism, I could whittle it down pretty quickly. I could see on which points we’re aligned and where we disagree. And in that process, I think I’d learn a tremendous amount about the subject, about the people I asked, and a whole lot about myself. We may still argue, but it won’t be about the semantics.
>At Primal Fusion, we refer to these subtleties of meaning as “personal semantics”. We can share a common definition of the word, but we will never share a common definition of its semantics. A tiny word like “hero” cannot possibly contain it all.
Hi Peter – agree but I thought sentence diagramming with sophisticated NLP, data mining techniques, and machine intelligence to generate human centered linguistic expression was the means to that end
Secondly, the article you linked ot Semantic Web: What Is The Killer App? was composed Jan 9, 2008 and last May at http://www.semantic-conference.com while we saw indicators and hints we still saw no REAL killer apps.
Thirdly, The Real Reason Powerset Sold (Out) is an interesting read i.e. the need for scale and big $$$ because Powerset, noted that their search “requires 100 times more processing than simple keyword searching and indexing (about one second per sentence is required for processing)”.
http://gigaom.com/2008/07/02/the-real-reason-powerset-sold-out/
Cheers….Steve
Steve, thanks for the comments. I agree, semantic extraction tools are evolving rapidly and that’s a great thing. The issue I’m addressing here is more a reflection on the highly individualized and personal nature of semantics. It’s the distinction between objective/universal representations of meaning vs. subjective/personal. Semtech needs to confront that quality of semantics.
Your point about scale is well-taken. It should be question number-one for any semtech venture. It’s also a big topic, so I’ll leave that discussion to a future post. But as it relates to the topic at hand, I think scale and semantic interoperability are clearly driving a bias towards universal semantics at the expense of the personal.