For your own deeper look, write me for a 24-page whitepaper on using ISO Topic Maps to hold, query, and exchange charts. It also explains how we came to this mission, our related products, future plans, etc.

Company Name: Lexikos Corporation
Proposal #: A5.01-8468
Project Title: Charting Paragraph Topics
Principal Investigator: Dan Corwin
Factor 1: Scientific/Technical Merit and Feasibility

Reviewer 1:

The research proposed has some interesting possibilities and is clearly related to previous work done at Ames. It is a challenging problem but, given the previous work they have done and the tools they plan to leverage, they could be successful. They seem to understand the difficulty of the task they face and have a plan for dealing with it.

Reviewer 2:

The proposed work is to develop a semantic and discourse parser to complement the offeror's pre-existing syntactic parser. The proposal is written clearly with a charming touch of informality. While the introduction implies that this work is well beyond the "intro to AI' assignments in NLP, it still relies heavily on a human operator to shape and interpret inputs. No NLP product is "magic" and indeed, a strong philosophical argument is that humans are the ones who inherently create meaning, not computer representations (see Herbert Clark's work (especially his book "Using Language"). Good reliance on existing and emerging standards.

Factor 2: Experience, Qualifications and Facilities

Reviewer 1:

The PI seems to have a strong background for the work described. While the need for subcontractors is indicated no actual subcontractors have been found. Finding someone with design and HCI skills for the UI developer would be important.

Reviewer 2:

The offeror is well qualified for the work. Personnel appear to have high-quality and relevant experience. No reliance on external facilities is noted.

Factor 3: Effectiveness of the Proposed Work Plan

Reviewer 1:

The work plan described is comprehensive but does seem to be somewhat of a stretch goal. However, given the PI's background, hours of work planned and work already done, there is a fair chance of success.

Reviewer 2:

The work plan is quite detailed and comprehensive. It is clearly written and quite logical to rely on existing Java freeware. Good discussion of technical methods. Schedule of the planned work needs to be more explicit. No discussion of cost management, but the proposed costs are reasonable.

Factor 4: Commercial Merit and Feasibility

Reviewer 1:

Should the work succeed there is clearly a growing need both within NASA and commercially for this work. Other tools would have to be designed to take advantage of this technology, however given the benefits I believe they would do so.

Reviewer 2:

Natural language processing is a technology with very broad commercial potential. It has been a difficult problem in AI for some time and while this proposal is not entirely persuasive that Lexikos has "the magic bullet", if successful, the proposed work will be quite valuable and commercializable.

Overall Comments:

Reviewer 1:

The work described, while complex, has a fair chance of success backed by a PI with a strong background and personal interest in the field. It the project is successful it would be a powerful tool both for NASA and others.

Reviewer 2:

While the whole premise of NLP is questionable depending on one's view of context and meaning, the proposed work is well-designed and offers new possibilities for the automatic analysis of text.