13th International Conference on Conceptual Structures
Conceptual Structures: Common Semantics for Sharing Knowledge

ICCS05 Tools Workshop

July 17, 15:00-18:00

Tools Topic and Goals

The tools workshop will consist of presentations of tools designed for developing and supporting applications built using conceptual structures. The presentations will include reports on the status of the more familiar tools for CGs and FCA as well as presentations on new tools that are currently being developed and deployed. We will also have a panel discussion for the last hour to discuss where we need to "go from here". It will include John Sowa as moderator and as many invited speakers from the conference that are available.

Panel Discussion

Tools for Intelligent Applications
Where are we now, and where should we be going?


The movie 2001, which appeared in 1968, featured an intelligent computer named HAL 9000. That computer not only spoke and understood English, it could read lips and devise complex plans on its own initiative. Marvin Minsky, one of the consultants on the project, said that HAL's abilities were a conservative estimate of what would be possible by 2001. Yet today, nothing remotely similar to HAL has been developed. The Cyc project, whose goals were somewhat more realistic, has been under development for over 20 years with over $70 million of research funding, but it still cannot generate enough revenue from commercial applications to pay its own expenses. This panel discussion will address the questions of why AI applications have not lived up to the hopes and promises of the 1960s and '70s, whether there were fundamental flaws in the research goals, what new kinds of technology are available today, whether any of that technology can deliver better results, and what new directions might be taken in the future.

Schedule

This workshop will take place on Sunday the 17th of July from 15:00 to 18:00.
Its planned schedule is currently the following:

15:00 Introduction - John F. Sowa, Chair

15:02 Formal Concept Analysis Tools - Philippe Martin, Chair

15:03 Jon Ducrou, D-SIFT


15:15 Conceptual Graph Tools - Mary Keeler, Chair

15:16 Alain Gutierez, COGUI
15:28 Harry Delugach, CharGer and OpenCG
15:40 Heather Pfeiffer, CPE
15:52 Philippe Martin, Comparison of CG-related tools
16:04 Jan Uhlir, DNAT
16:16 John Sowa, VivoMind Technology


16:28 Coffee Break

16:45 PORT Project - Heather D. Pfeiffer, Chair

16:46 Mary Keeler, PORT's Testbed Partnerships


17:00 Panel Discussion - John F. Sowa, Moderator

Abstracts

D-SIFT (Dynamic Simple Intuitive FCA Tool)
Jon Ducrou


Abstract. D-SIFT is a Web-based browser application that provides untrained users in Formal Concept Analysis with practical and intuitive access to Formal Concept Analysis's core functionality.
D-SIFT is an information systems architecture that supports natural search processes over a predefined database schema and their attribute values. This enables the user to build concept lattices interactively through the selection and refinement of dynamic definitions of search boundaries and dynamic selection of search scales based on the attribute value features contained within the database. This demonstration will show the tool, its usage and some techniques for achieving different goals from a conceptual information system.

COGUI
Alain Gutierez


Abstract. Cogui is a java standalone application offering user interface for graph editing in CG context. Writing a graph editor with a minimal set of features require some development effort. Using an existing API also require some adaptations to keep edited graphs and back-end CG objects in synchronization. The purpose of Cogui project is to provide a scalable workbench interacting with the CG model.
Through an event-driven approach, it is possible to handle editor behavior regardless of real components in use. Every editor is decorated with different handlers depending on graph type, user status, or applicative context. To preserve compatibility with future releases of external APIs, the public package is intended to work with any API implementing some fixed interfaces. Actually a subpackage do this ugly task in adapting Jgraph a LGPL licensed API.

Actually cogui supplies a limited but extensible set of predefined actions to work with CG objects including :
* cogitant connectivity based on xml protocol
* most common editing features such as copy/paste, drag&drop;, undo management
* in-place property editing, ontology browsing
* different layout algorithms to arrange diagrams
* debugging features, message browsing, object highlighting
* xml i/o.

CharGer and OpenCG
Harry Delugach


Abstract. CharGer, a stable conceptual graph editing platform, will be briefly described. Now that CharGer is an open source project, we can envision further cooperation with other open source efforts, such as the new OpenCG effort being initiated by Bevilacqua Research Corp. in Huntsville. We will also discuss coordination with the new ISO Common Logic standard that is being developed.

Conceptual Programming Environment, CPE: Efficient Data Structures and Algorithms for a Knowledge Base
Heather D. Pfeiffer


Abstract. It is known that world knowledge cannot be directly stored in the computer; therefore, must be transformed to a format that the computer can use. This transformed knowledge is saved in the machine using a dynamic data structure that is suitable for the style of knowledge representation (KR) being implemented. This data structure, together with the contents of the transformed world knowledge, is known as the knowledge base. The KR allows the system to manipulate the knowledge in the data by using reasoning operations.
Data structures and algorithms used in reasoning operations for operating over knowledge bases must be flexible, fast and efficient. The reasoning operation is also the basis for querying the knowledge base, and uses the same knowledge representation (KR). This knowledge base applies a particular style of semantic network KR known as Conceptual Graphs (CGs). The reasoning operations make up a complete conceptual programming environment with multiple language API interfaces; therefore, the system is called CPE.
CPE addresses both the efficiency of the underlying algorithms for knowledge reasoning given a particular knowledge representation where the data structures and algorithms for the operations of projection and maximal join are examined, and provides a modular flexible generic interface for using these efficient reasoning operations for storage and retrieval to a knowledge base.

Formal and semi-formal Comparisons of CG-related tools
Philippe Martin


Abstract. This talk will show (i) an ontology that represents some aspects of some CG-related tools, (ii) the way tables can be generated from user' selected criteria to compare tools, (iii) the way the ontology can be updated in WebKB-2 by the author of each tool, and may be (iv) the way an informal but structured discussion/feedback on a tool aspect can extend the ontology. A very quick comparison with a static and superficial comparison of ontology editors will be made.

DNAT (Dynamic Narrative Annotation Tool)
Jan Uhlir


Abstract. DNAT was designed to support users in creating printable knowledge-intensive content and the corresponding knowledge-base at the same time. DNAT allows creation of semantic annotations in Conceptual Graphs (CG). CG is a human readable notation for First Order Logic (FOL) and were developed by John Sowa. The tool uses this CG notation as an interface between annotated content and ontologies that store definitions of concepts.

VivoMind Technology
John F. Sowa


Abstract. VivoMind Intelligence, Inc., has been developing software based on conceptual graphs and related logic-based languages. The foundation is a system called PrologIKS, which consists of a compiler for ISO Standard Prolog plus Intelligent Knowledge Services. The output of the compiler is either C code for optimum performance on any platform or Java bytecodes for portability across all platforms. The Flexible Modular Framework, which was presented at ICCS'04, is the basis for interconnecting modules implemented in any language, including PrologIKS or any other hardware or software technology. Among the modules of the FMF are the VivoMind Analogy Engine, which was presented at ICCS'03, and other modules for processing various languages, accessing and serving the WWW, and doing reasoning by induction, deduction, or abduction. This talk will show how the FMF provides an open-ended framework for designing and connecting modules of any kind on any platform, ranging from modules running on a cell phone to legacy software or the latest research prototypes.

PORT's Testbed Partnerships
Mary Keeler


Abstract. The difficulty of establishing effective testbed partnerships, among users of computer-based tools and builders of those tools, has has been an obstacle to the success of science collaboratories. Testbeds were intended to support partnerships between users and technologists in exploring the utility of various technical approaches by which a scientist user-community could take advantage of emerging technology. Many promising knowledge tools are stranded at the end of the research and development cycle, waiting to be found by some user community. Meanwhile, user communities are struggling to find tools powerful enough to augment their increasing collaborative, intellectual work. Testbed partnerships are as essential for knowledge-tool advancement as that advancement is to the support of effective partnerships, because knowledge tools must be instituted in a context that truly augments (rather than merely tries to replicate) human inference in collaborative contexts. Participants' ability to institute self-critically controlled conduct as the directing agency of user-tool co-evolution will determine whether they can engage in effective partnerships. Based on game-theoretical findings of how such cooperation and coordination capabilities evolve, we propose a game framework for instituting the self-critical conduct required in testbed partnerships.