Yesterday was a bit of an odd day. Decided to get a later train from Bath and rolled into work after eleven, which partly makes up for the extra time I’ve put in over the last few weeks. Ran a viva for one of my mentor’s PhD students (we viva our PhD students three times, as a check on progress at nine months, for the transfer from MPhil to PhD at eighteen months, and the final thesis defence) which went well. Needed another teaching observation for the lecturer training course, so got my mentor to assess me assessing his student.
And then there was the crank. Just before lunch, a chap in his thirties stuck his head round the door and asked if I was busy because he’d like to talk about Semantic Web technologies. Well, fair enough. It’s not uncommon for Glorious Leader or Eternal Leader to send visitors round to me at little or no notice, so I asked him to come back after the viva.
For the first fifteen minutes he seemed normal enough; we talked about the difficulty of building ontologies that were reusable and applicable to more than just one domain. Then he pulled out a sheaf of papers and started to explain his idea, which was a universal ontology that could be used to represent all knowledge, that was grounded in natural language yet worked across languages, and which did not suffer from encoding biases due to underlying assumptions. I started to realise that I had a problem when he said that the problem with current ontologies is that they are only two dimensional, whereas they need to be three dimensional (this is not a meaningful statement). You see, “syntax is one-dimensional, pragmatics is two-dimensional whereas semantics is three dimensional”. I pointed out that computational linguists place pragmatics at a higher level than semantics, and that I was confused as to what the dimensions might be.
He had a chart of cells which represented concepts, where the relations between concepts were indicated by adjacency. The problem was that the concepts were two or three letter groups like “ish”, “ork”, “erf” and “und”. When I asked what the concept labelled “ish” was, he said “it’s the concept linking words with ish-ness, like issue, repetition, volition and so on”. Ri-ight. I forget the exact details, but when I asked how “ish” related to the cell on its right, labelled “erv”, there was some long explanation that involved semiotics and the fact that the ish-erv relationship was indicated by the “erf” rule at the bottom of the chart.
I can’t remember where I read this, but a year or so ago I read an account by a physicist that talked about how one deals with members of the public who wander into universities with Theories of Everything that manage to be overly complex, factually incorrect, fail to describe the observable facts, and plain crazy. As I sat with this man, I was frantically trying to think of a polite way of getting rid of him without having to say that his ideas were nonsense and that he seemed like a loon, albeit a well-mannered one. He’d admitted that he’d already gone to Oxford and presented his ideas to someone in the linguistics department, but that they’d been uninterested in his systematic means for determining the meaning of any word.
To cut a long story short (and to be honest, my mind turned off after a while as I tried not to think of Time Cube and giggle), I got rid of him after the best part of an hour by saying that I thought his ideas were interesting, but that I wasn’t properly qualified to offer him advice on them because they were outside of my field of expertise.
Very, very strange.