Formal and Semi-Formal Summaries

(Workshop at ICCS 2005, July 17th, 2005)

The workshop has now occurred. Among other issues raised in this workshop and in a subsequent meeting with people from the "Pragmatic Support for Collaborative Inquiry" was the need for an application context in order to motivate people to engage in structured discussions (see below). However, the ideas of summarizing the content and debates from the CG mailing list were very welcome.


Topic and Goals

In the Conceptual Graph related part of an archive of computer science knowledge or a Digital Aristotle, how should the techniques, tools, opinions, reached status-quos, ... be organized and presented to help researchers or students learn about them or find them, compare them and complement them quickly? That is the central question of this workshop. Such a knowledge repository (cooperatively-built semi-formal state of the art) would also permit the authors of articles or messages in mailing lists to reference existing ideas or techniques instead of having to summarize their advantages and drawbacks, and reduce recurring discussions on mailing lists.

Many document databases specialized in certain fields (e.g. subparts of medecine or mathematics) already exist but a base/Web of documents cannot by itself support a "knowledge repository": ideas and techniques from articles or discussions should be organized into a (semi-)formal very structured semantic network. Nowadays, many domain/discussion summaries exist but virtually none is structured into a semantic network or into an argumentation structure. Many domain related ontologies exist but they do not compare ideas, tools and techniques. This workshop seeks the contribution of examples of such formal or semi-formal summaries. Each author is free to re-use or invent the notation(s) that s/he feels the most adequate and readable for entering the knowledge and then ease its retrieval via queries and conceptual navigation. This workshop has two goals: 1) collecting and comparing different viewpoints on what the kind of content and notation should be for "knowledge repositories" (as illustrated above), and 2) collecting core knowledge for a "knowledge repository" on CG related subjects. This core knowledge will be available for re-use and incremental completion via a wiki (see for example the pages about Structured discussion and CG tools) and via the shared KB of AnyKB (and WebKB-2). In this wiki and this shared KB, for each piece of knowledge or annotation, a reference to its author is kept; thus, the knowledge providers and reviewers will be acknowledged and subsequent refinements will likely refer/re-use their contributions.

From a tool designer perspective, the collected knowledge may also simply be seen as a body of knowledge representation examples, the particular nature of which is likely to motivate extensions in CG tools or knowledge servers.


Submissions, Reviews, Discussions and Outcomes

Authors are invited to submit HTML documents presenting a formal or semi-formal summary about a technical domain, preferably a domain related to CGs (for example, a summary can represent and organize the main ideas of one or several papers published at an ICCS conference). Each HTML document should be divided into sections and subsections. One section should quickly explain the adopted (semi-)formal notation and the others should mostly contain knowledge representations or argumentation structures. Each (semi-)formal part should be enclosed within the tags <SCRIPT TYPE="text/knowledge/name-of-the-notation"> and </SCRIPT> in order to permit the (semi-)formal parts to be automatically distinguished from the informal parts. Graphic notations are discouraged unless a textual notation is also provided and it is shown that the use of the graphic notation improves both readability and conciseness for the particular content provided. There is no restriction on the volume of the provided knowledge and hence no restriction on the length of the HTML document. However, since authoring very structured information is difficult and time-consuming, it is unlikely that very long documents will be provided.

As for the PORT workshop in 2002, each person submitting an article will become part of the reviewing committee and the reviews will be conducted via a mailing list and/or directly via the structured wiki. Hence, the reviews will not be anonymous (for PORT, this proved to lead to detailed and careful reviews and this allowed interesting feedback and revisions by the authors). Other researchers are also encouraged to participate to the reviewing and discussions.

The deadline for article submission is June 30th, 2005: an HTML document (or whenever possible, a URL for this HTML document) should be sent to "pm at phmartin dot info" (sorry for using this anti-spam technique). Thus, reviews and discussion on these articles officially open on June 1st, 2005, but articles (or parts of articles) can be proposed ealier, and hence reviews, discussions and incremental revisions can occur ealier. Thanks for sending expressions of interest to "pm at phmartin dot info" as soon as possible.

If before the workshop, printed copies are asked for, some printed copies of the selected articles and of the best parts of the integrated result will be available at the workshop.


Schedule

This workshop will take place on Sunday the 17th of July from 10:00 to 13:00. Its planned schedule is currently the following:
- "Semi-formal summaries: some rationales, approaches and examples" by Philippe Martin (30 minutes);
- informal discussions (30 to 90 minutes; within that period, paper presentations can also be made);
- group exercises (collaborative creation of (semi-)formal summaries with about 15 to 30 minutes per subject): 1) Structured discussions (proposed subjects: "XML for knowledge exchange", "Design criteria for CSCW tools" and "abortion"), 2) other kinds of (semi-)formal summaries (proposed subjects: "CG tools", and "sport reports");
- informal discussion to ponder on the results of these exercises (15 to 30 minutes).


Intended audience

The researchers of the CG community have various goals but many of them are interested in representing the content of natural language discourse in clear, precise and normalized ways. Thus, they seem more likely than other communities to contribute knowledge for "knowledge repositories", appreciate the potential of such repositories, and exploit them. However, participants from other communities are also very welcome.

The participants of this workshop will have the chance to 1) make contributions to the "core" of the first cooperatively-built semi-formal state of the art, 2) learn and discuss about the issues related to that enterprise and its potential, and 3) orient developments related to that goal.

This workshop has several relations with the ICCS 2005 Tools workshop (which takes place immediately after this workshop) and the Pragmatic Support for Collaborative Inquiry workshop. Therefore, participants to these two workshops are invited to submit a (semi-)formal summary of their tools or research.



Organizer: Dr Philippe MARTIN (email: pm at phmartin dot info)
The subject of this proposal is directly related to his article for ICCS 2005.

Dr Martin got his Ph.D in Oct. 1996 at the INRIA (Sophia Antipolis, France, ACACIA project) where he developed CGKAT, a Conceptual-Graph-based and structured-document-editor-based knowledge acquisition tool. During his postdoc in 1997 at the University of Adelaide and then as research fellow at Griffith University (Australia) until June 2000, he developed WebKB-1, a knowledge server using, retrieving and generating (parts of) Web documents and superseding CGKAT (this work was sponsored by the DSTO). Then, as senior research fellow employed by the DSTC, he developed WebKB-2 (a knowledge server complementing WebKB-1 by allowing Web users to tightly interconnect their knowledge into a large shared knowledge base without having to agree on terminology or facts) and its default Multi-Source Ontology (MSO) (voted as a candidate ontology by the SUO working group in May 2004). Dr Martin was invited at the LOA (Trento, Italy) in Oct.-Nov. 2004. Now back at Griffith Uni, besides doing some teaching and looking for full-time research grants, he dedicates his research time to AnyKB (the open source version of WebKB-2) and its extension to support semi-formal "knowledge repositories" and discussions (updates of the MSO are also on his agenda).

Contact: