alpha ontologist

02 Oct 2022 05:11 - 07 Oct 2023 07:12
Open in Logseq
    • Any ontology project has to deal with the fact that our formalisms for representing knowledge are impossibly inadequate to their given task. The standards in vogue nowadays (OWL, the official ontology standard of the semantic web) are particularly hard to use. (See Politics and Pragmatism in Scientific Ontology Construction )
    • I came up with the phrase "alpha ontologist" during a short stint working with Doug Lenat's Cyc project. At the time, they were trying to encode all of botany and had a small staff of professional botanists doing knowledge entry. Naturally it was quite difficult for the botanists to try to translate their knowledge into the formalisms required by Cyc, and they would regularly puzzle over various questions (I wish I had written some of these down) and if they could not come to a consensus, would have to take it before the Master, Doug Lenat, who would think for a bit, maybe draw some diagrams on a whiteboard, and come up with the Right Representation.
    • Because I was somewhat alienated from the goals of the project, I thought this process was a mix of alarming (because it obviously couldn't scale), hilarious, and interesting as a really solid example of the process of formalization, which is sort of an essential task of any computational scientist but has no good theory to say how it should work (IMO – there are plenty of theories, but they aren't very good). If we understood this better, maybe it could be made to scale, you wouldn't have to be Doug Lenat to do it.
    • image.png
    • It՚s unremarkable I guess that any group doing cognitively difficult work would have one person at the center who held the team together. It rubs against my anarchist sensibilities, which lead me to dislike leaders. But that՚s just me, for normal people having somebody In Charge seems pretty unproblematic.
    • My objection to the dynamics with Lenat՚s project was not that it was authoritarian, but that it couldn՚t possibly scale to the level demanded by its goals, which were to represent basically everything. Whether or not that՚s possible or sensible, it seemed unquestionably true that you couldn՚t do it if every representational difficulty had to be adjudicated by one person.
    • I think others were aware of this problem, and were attempting to address it by various means. Lenat could train others to be master ontologists and could then fan out to manage larger-scale efforts. He did have a protege, R V Guha, who played that role, but I don՚t know if it ever scaled out past that.
    • Or you could leverage the computational tools to enforce a degree of internal consistency and guide the knowldge-building process.
    • image.png