May 2, 2005
On the discussion of data
regularity à
On the Stephenson Cyber Attack
Taxonomy à
Our strategy is to develop communications to senior individuals at IBM, LM and at SAIC about the Roadmap and about the possibility of using the technology that is indicated in the Roadmap to bring a positive solution to some existing IT projects.
We also have communications that are being developed with policy makers. In all cases, there is nothing critical except in regards to the "First School of Semantic Science", which we represent as mistaken in regards to the limitations of computers. As scientists, we point to a limitation that algorithms have in exhibiting intelligence behavior. This is not a philosophical issue, but one where the mainstream Semantic Web community has "left the barn".
The Second School is positioned as a new solution to deployment efforts that always seem to end in some type of train wreak. We represent that the failures in advanced IT deployments are traced to First School assumptions about the nature of intelligence. Management issues (anywhere) are compounded by a fundamental confusion that Artificial Intelligence continues to create.
The Roadmap is supported by a Advisory Board having membership now of just under 40 PhDs and lead visionaries. Etc etc.... (sorry, I am just trying to develop the political scope of this Roadmap.)
Success in a first deployment will lead to a series of scientific publications and to additional discussions about adjustments to funding orientation at NSF, NIST and DARPA. It is this scientific success that motivates me, but the scientific success is predicated on business success.
It seems reasonable that this project be first developed as Selectivity and Targeting and the Advance Targeting System deployments at US CBP. The team is set, and the current proposed budget could be cut from $750 K to fit within the current eCP budget expenditure expectation. It this moment, I am willing to bring the technology ownership to LM (subject to license agreements), if LM will agree to allow future deployments outside of DHS to be un-inhibited by LM. (This is the IP issue.) SAIC is currently moving too slowly on this for us, and I am looking to find a Prime that will make agreements over exclusivity that allows the BCNGroup the rights to develop non-governmental markets.
However, we need some visibility at LM if the discussion is to be properly developed. As you can imagine, setting meetings with senior VPs at any of these organizations is quite difficult.
The Roadmap integrates 10 COTS systems at low costs. The functions that the Global Information Framework will have are there due to principled reasons that are only seen when one has many years of experience with artificial intelligence and related attempted deployments.
The human centric is key for reasons:
1) the cultural transitions
2) the limitations to machines as inference mechanisms
We binding the Roadmap (version 8.0) into a book (one that fits into a pocket) so that it can be distributed in certain circles. I am wondering if a committee might meet with you and give one of these to you. I suggest meeting with Dr Art Murray (GWU Knowledge Management Institute), Doug Weidner (founder KMInstitute) and I. Doug and Art are long time BCNGroup associates of mine and have no Industry affiliations.
Tom Bush is aware of the evolution of the Roadmap, and looks to you to provide a go ahead on doing something that gets the ontology (MIKE) prototype back on track. I understand that Chris is leaving. This is a positive development because he had a very limited understanding of what could be done.
It is time to get this project put together.
Ah, please one other thing. I have put together budgets and indicated who is in and who is out in regards to a trade study. I can imagine that Convera still thinks that it has or will have a 3 M contract to provide semantic extraction tools to eCP. (This information is on their web site, and my discussions lead me to understand that CTO Claude Vogel’s business staff felt that they already had a eCP Contract in hand.) My belief is that this is an illusion that the business staff has lead Claude into. Time will tell if I will have technical discussions with Claude.
A similar discussion has occurred with Larry Levy, CEO at Semagix. The Semagix product is an excellent one, but has the problem that all semantic extraction tools have - the need to manually focus and certain blind sights based on internal algorithms. This is why the GIF has three quite different semantic extraction processes, NdCore, Readware, and Recommind.
The discussion I have made are not laid out in stone, but are based on the companies willingness to provide to me low cost evaluation copies and technical support during the integration of the GIF.