Friday, September 24, 2004
Background discussions on a proposed
Anticipatory Technology Challenge Problem
Tom and Paul
I see the challenge problem as producing a better theory proving infrastructure for Analysts. This they admit they do not have, because the hard problems have led to a disparate collection of methods and frameworks, each with independent, representations, parameters, logic and requirements, addressing some piece or part of the hard problem. The real hard problem now, after billions have been invested, is why the sum and whole of all the parts is not greater that the parts alone.
Our conjecture is there is no glue: no coherent and universal conditions capable of bringing them or holding them together. (At least no one believes there is and no one worked on that part.) We will investigate whether this is true or not and whether combining some independent IP and work product provides a pragmatic and near term solution.
The workshop will challenge existing state-of-the-art in indexing and text analysis technologies, and knowledge frameworks, with the notion that a more universal and stratified architecture that takes care of parsing, expansion, thematic analysis, conceptual roll-up, and performs dynamic linking and association, using more sophisticated measures than co-occurrence or noun phrase alone, would provide better visualization for the analyst that will enable hypothesis testing (over the same, new or old data sets) using newly acquired anticipatory insights.
Also the fact that most of us will be bringing IP to the table that should be spelled out in the proposal too.
Now this is my entire understanding and filling ten pages of a proposal should not have to say what or how we will do anything other than the above. We do not have to fill it with explanations we just need to propose it in statements that cohere with areas of the ARDA program.
We do not have to talk about management or about how we can commercialize the technology. We can talk about how we recommend program managers from the specific program areas that we expect could benefit should attend our workshop and add to or orient the challenge problems that will be set out in the first workshop.
We should propose an unclassified workshop and seek champions for anything else. i.e., let's proceed with Aesop's fables as the generic text that we can start with. Maybe workshop participants will agree on other texts.
Ken