ORB Visualization
4/1/2004
9:06 AM
A group of individuals have identified some surprising technical advances that could be integrated. The problem is in establishing a means to advance scientific work in spite of these five types of general systems problems.
http://www.bcngroup.org/beadgames/techInnovation/twentytwo.htm
The integration of these surprising technical advances has been discussed with at least one person in each one of eleven companies. In most cases, the companies do not understand or they cannot see immediately how the surprising technical advances would be incorporated.
As an example, one of many, consider the very good work on automated generation and use of subject matter taxonomy. This work has great difficulty in being accepted by government agencies for two reasons. First is the cultural resistance and absence of understanding regarding what automation of taxonomy/ontology construction from text would do if easily available.
The second is the prohibitive cost of current deployments. The high cost is linked to a bundling of many patents into one product. In each case, these patents are based on mathematics (which is not patentable) and if re-implemented based on the surprising technical advances would produce new intellectual property that could replace the existing patents. InOrb Technologies could completely reproduce all functionality of taxonomy generation software using technology that has 1/10 the memory and processor requirements and is 10 – 100 times faster. Moreover, as this functionality was reproduced, new IP would be generated so that no dependency on existing patents would be needed.
Whereas this would be great for the client, it would require an adjustment by the now established leading companies. A Technology Collaboration [8] is a plan to fix this problem.
We claim that business CEOs have not taken the time to understand the issues of science in this respect. As argued below, we feel that current new investments in knowledge-management type technologies fail in the market because the science cannot be addressed within the current business models.
As a broad generalized statement, the relevant scientific work can move forward with economic support.
http://www.bcngroup.org/investment/
On a positive note, we have proposed a conjecture on why the intelligence failures are systemically re-enforced.
http://www.bcngroup.org/beadgames/techInnovation/twentytwo-two.htm
The new innovations will directly address the failures.
One can observer that the standard venture capital criterion is shaped by well-understood IT and biotechnology patent aggregation behaviors, as identified in business science literatures and government studies:
http://www.bcngroup.org/investment/pdf
On a larger scale than [8] the BCNGroup (www.bcngroup.org ) is proposing a National Project to develop a knowledge science curriculum and to map the IP space in this area so as to bring some transparency on patents in this area. The purpose of this work is to assist in building a new marketplace based on a public understanding of the role that human cognitive acuity MUST play in machine processing of human linguistic variation, whether based on statistics, look up tables, categorical computation; or a combination of these methods.
The dual problems of technology evaluation and technology procurement are addressed at the same time.
Our HIP (Human-centric Information Production) paradigm requires a separation of the interpretation of patterns of words from the instrumentation/measurement of linguistic variation in text. This, unlike Tim Berners Lee’s notion of the Semantic Web, is the grounded architecture for action-perception cycles with two sides, the machine side and the human side. A criticism of the Artificial Intelligence / Semantic Web paradigms are given at { * } .
To illustrate how the community of scholars is developing a clear understanding of HIP, I point to the work by William Rapaport. In his work, a model of the reader’s interest is used to assist in the algorithmic process related to retrieval of subject matter. We did something similar in my distance-learning prototype for the State Department in 1998. More generally, an extension of Russian quasi-axiomatic theory provides a means to make the machine translation, and summarization, process work within an action perception cycle. The separation of personal awareness form the external realities is central to the socio-biological perspective of Maturana and Velara (as expressed with their term “autopoiesis”).
The way in which HIP supports action perception cycles in event awareness was discussed in our proposal to the NIMA Glass Box project (2002) – deemed fundable but not funded. This is good work and should have been funded in 2002.
In stratified theory, categorical abstractions are developed at several levels of organization and then used in the support of real time decision support that involves a human at each step.
Quote (25 March 2004) from the In-Q-Tel web site:
"In just a few short years, In-Q-Tel has evaluated nearly two thousand proposals, 75% of which have come from companies that had never previously considered working with the government. To date, we've established strategic relationships with more than 20 of these companies." (this information is considerably out of date.)
Is the amount of funding that
goes to actual breakthrough innovations sufficient to cause a shift in the
recognized failure of machine-only based text understanding systems.