Previous comments. make a comment Research note 26

Research Note 21 October 16, 2003

When business trumps science

A systemic cultural problem is illustrated.

The problem is not with one company, but with a business culture in which business reasoning, of the narrowest nature, steps in and controls research interests often simply for the sake of ego, out of simple ignorance, or because of an unwillingness to understand what the issues of science are. It is often that this behavior is expressed by individuals who are in a type of self-denial that they are involved in this.

One might credit this behavior as creating the cultural barriers that keep the intelligence community stove piped in spite of the best efforts of the Congress and the Administrations, past and present. In fact, the evidence for the self-denial includes a close examination of what are largely irrational institutional behaviors.

It must be stated that certain lines of research are funded consistently and some of this is excellent research. Some of it is very poor research, but has funding.

Over our recent history, particularly in the software industries, this inhibition of certain lines of innovation has been so much part of the culture that it is recognized to be something that others are involved in, but never recognized as being something that is a controlling factor over “my” behavior. We have a blind eye to how the culture is expressed through us, and we have denial that we have this blind eye.

Our culture could more broadly understand this and act to balance the influences so that business was not always the rule making side and science the rule following side. This would benefit everyone, including the business-oriented persons.

We have a situation very similar to the fable about the Emperor’s court where everyone will agree that the Emperor is clothed in a fine cloth, when in fact the Emperor is naked. (See Sir Roger Penrose’s book, The Emperor’s New Mind” )

Proposed near term research

1: An evaluated technology is based on a construction and methods for finding connections between type:value pairs. This technology was judged, by several scholars, as being widely available from scientific and mathematics literatures. Moreover, the methods disclosed in a patent covering this technology are do not take into account any theoretical framework that might justify why the connections found and the higher level constructions developed by the patented method might have validity.

1.1: We, OntologyStream and scholars that communicate to us about these types of issues, have judged that the type:value pairing in the construction suffers because no method exists for developing an situational ontology that describes the class of types specific to an investigation. Several approaches might be made to address this deficit. These approaches have been offered to anyone who will continue to fund the basic research, while making an agreement that the results are to be openly shared and if patented will be licensed at a reasonable cost.

1.1.1: The object-oriented paradigm was developed as a software paradigm from work beginning in the early 1980s. It is grounded in theories of type, realized as “classes”, and instantiation of type as “objects”. These theories go back centuries. We suggest that the use of generalFramework theory provides a return to deep roots in scholarship that predate computers.

1.1.2: The replacement of the type:value pair with a richer class:object pair is suggested and delineated in the Notational System for Ontology Referential System. Nathan Einwechter and Paul Prueitt could build, within four weeks, an experimental system that uses a variation of the Ballard Framework . One could then have direct evidence that a complex class:object “Ontology Referential Base” would provide an immediate quantum leap in intelligence technology.

Figure 1: The Ballard Framework

This experimental system would show how classes might be defined from the set of 18 “semantic primitives” in the Ballard Framework, and objects produced from an automatic parsing of text. The experimental system would use a data encoding mechanism that is demonstrated in the existing experimental system’s Referential Base. It would have a “controlled” terminology reconciliation processes allowing members of a community to make collective modifications to an underlying category theory, as expressed in the nature of the classes.

The automated parsing of text could be produced using Amnon Myers’ system for developing text analysis systems. In the same effort, this group of three individuals could demonstrate a Parts of Speech tagger where the tagger produced class:object pairs having classes from a framework having 4 primitive classes.

Figure 2: A Parts of Speech Framework

These four primitive classes are extended so that noun roles, verb roles, noun phrase roles and passage roles each have a theory of type and functional role behavior included in the class structure.

One additional issue needs to be immediately addressed by this team. This has to do with how connections are made between localized class:object pairing.