Background

Consulting

Publications

Reading

Research

Contact

Home

Patrick Durusau, patrick@durusau.net

Reading

The listing below contains new books or articles that I have found of interest as well as older gems that I have discovered while researching a variety of subjects. Only brief comments are included here, but occassionally I will post more extensive comments on a particular work and those will be linked to from this listing.

  • The Topos of Music: Geometric Logic of Concepts, Theory and Performance, by Mazzola, Guerino. Birkäuser Verlag, ISBN 3-7643-5731-2. Erratta Mazzola's homepage

    This is an amazing work of European scholarship that develops a formal mathematical concept of music. Along the way it considers a wide range of topics, including category theory as the basis for that theory. The work runs just over 1,300 pages, with a bibliography of 595 entries. A CD is included with the volume that has the full text as a PDF file (and software), which makes it possible take this work along while traveling.

    I became interested in this work because category theory looks like a useful tool for developing theories of subject identity that can then be used in Topic Map Applications (not software, the rules by which subjects are identified), also referred to as TMA's.

  • What is Knowledge Representation? by R. Davis, H. Shrobe, P. Szolovits, AI Magazine, 14(1):17-33, 1993.

    I actually found this resource while running the citations in John Sowa's Knowledge Representation. Compare this paper to Sowa's report of it on page 134 of Knowledge Representation. I find the original paper more persuasive than Sowa's interpretation of it.

    The fundamental issues in knowledge represenation have not changed since this paper was published and have often been discussed and debated with less insight that is found within. Highly recommended.

    The phrase, "running the citations" refers to the practice of tracking the citations of an author back to their sources and then the citations in those sources back to their sources, etc. It is quite instructive when, for example, a recent CS paper had one citation to a non-existent article, fifteen citations that were incorrect or incomplete, out of a total of twenty citations.

    At the very least ones gets an idea of how carefully the author considered their sources, when citing articles that simply don't exist or were not considered important enough to merit correct citation. Such a practice also aids in finding when a cited source does not appear to support the proposition for which it is being cited.

  • HyTime ISO/IEC 10744:1992 (in pdf)

    Actually I am re-reading the material on groves. Recall that a grove (graph representation of property values) is defined as: An abstract data structure consisting of a directed graph of nodes in which each node may be connected to other nodes by labeled arcs. (ISO/IEC 10744: 1992, 3.35) That being the case, it appears that the notion of a node having only one parent is an artifact of the SGML property set. See directed graph and simple directed graph, both at MathWorld. Had a grove by its very nature been restricted to a single parent for each node, the standard could have easily defined a grove as simple directed graph.

    I am interested in the notion of groves, separate and apart from the architectural form trappings of ISO/IEC 10744 both as a means of handling overlapping hierarchies, supporting interchange between divergent information systems and as a mechanism for robust implementations of topic maps.