Attempting to close down L594. Mail to Ralf and Kathryn.

---------- Forwarded message ----------
From: Chris Dent <cdent@burningchrome.com>
Date: Fri, 9 Aug 2002 02:38:43 -0500 (EST)
Subject: Short summary of L594 work


Below is a summary of the work associated with the (revised) L594. I
sent out a message earlier this week asking if there was anything else
that needed to be done to seal up the work to be rid of the
incomplete. I didn't hear back (as far as I can tell), so I'm sending
out this message as insurance. Please let me know if there is anything
else that needs to be done. Thanks.

-=-=-
To reflect unexpected changes in the goals of projects related to this
L594 project the requirements for project completion have been
adjusted to the following three:

- Migration of existing data and tools from an Oracle database
  installation to a system using MySQL. As a result a more complex
  data schema is supported (for future use) as well as more rigorous
  free text searching and retrieval.

===This has been accomplished.

  - The system now resides on servers in my home, using a recent 4.0+
    version of MySQL.
  - This version supports two styles of free text searching:
    - TfxIDf
    - What MySQL calls boolean. Searching in this mode will return
      messages that contain the keywords in the query, regardless of
      frequency count.
      - There is a bug with this style of searching which presents
        itself when you search on a word shorter than the index
        allows. The server crashes. However, because of the way the
        system operates, it starts right back up, so the error is only
        present for the user attempting the search.
  - Data is stored in a more full featured data scheme that will
    support future goals of identified users, evaluations for all
    inputs, and abstracts. The descriptors and evaluations added
    in this round (described below) do not use this more complex
    scheme.

    The schema is here:

      http://ella.slis.indiana.edu/~cjdent/unrev/htmlreport2/

  - All of the command line input and output tools created for the
    Oracle database have been replicated on the new server. Similar
    tools, for use with the new features, have been added.

- Extensive revision of the paper submitted, with Kathryn La Barre,
  to the 2002 PORT Pragmatic Web Workshop. The paper initially described
  this work but was revised to explicate the intellectual context
  that supports the design decisions we have made.

===This has been accomplished:

  - Version one of the paper:

      http://www.burningchrome.com/~cdent/uviz/portpaper1.html

  - Version two:

      http://www.burningchrome.com/~cdent/uviz/portpaper2.html

  - Version two was the result of extensive conversation with the
    members of the peer review process:

      http://lab.bootstrap.org/port/

  - Writing the paper had a powerful influence on my interpretation of
    the works in the Augury readings group:
 
      http://www.burningchrome.com/archives/augury/

- Implementation of numeric evaluations and textual annotations in
  the database system and interface. These will allow community
  members to participate in the structuring of the data that allows
  for flexible retrieval. Community members will be able to give a
  numeric score to groups of messages (based on author, subject,
  keywords, or date range) or individual messages reflecting the
  relative value of the message. Text annotations will be based around
  a limited set of meaningful phrases, selected by the community, that
  reflect the "aboutness" of the messages.

===This has been accomplished:

  - Working prototype versions of evaluations and what came to be
    called descriptors have been added to the database structure and
    the CGI interface to the system:

      http://www.burningchrome.com/~cdent/uviz/cgi/index.cgi

  - With both the evaluations and descriptors the implemented
    interfaces are considered "basic". This is because it was
    necessary to create something quickly to ensure we did not loose
    the attention of the Bootstrap Alliance community members that we
    hope to use for building and testing the access structures we
    develop.
    - Code and database elements in the system are marked with "basic"
      to distinguish them from the more flexible and considerably more
      complex original design.
  - Numeric evaluations maintain a running average of scores that
    visitors to the archive provide. The evaluations can be used to
    limit the four styles of search (body contents, subject contents,
    author, descriptor (see below)). 
    - Visitors may evaluate a single message or a set (see below for a
      description of sets) of messages.
  - Descriptors are made up of a word or phrase selected by the
    administrators of the system (responding to input from the users).
    - When reading a message, a visitor
      - may add one or more descriptors to the message
      - may view all other messages that have one of the same
        descriptors
        - all the descriptors associated with a message are listed and
          "clickable"
    - A visitor may select messages from the search interface based on
      one or more descriptors

===Other changes:

  - Message sets
    - When a user retrieves a collection of messages, the system
      remembers, temporarily that list of messages so that the user
      can operate on and traverse the set of messages.
  - A simple free text interface to the unrev-ii archive. 
    - To compare with the CGI/Database search interface, I created a
      mhonarc

        http://www.mhonarc.org/

      archive of the mailing list with a search interface using FFW

        http://www.burningchrome.com/~cdent/uviz/ffwdate.cgi

  - I collected all these resources (including the papers) under a
    single page

      http://www.burningchrome.com/~cdent/uviz/

    linking it all back to the official starting page at 

      http://ella.slis.indiana.edu/~klabarre/unrev_firstpage.html

</warp>