> Jack Park wrote:
> This is precisely the nature of collaborative information filtering. At some collective web site, users just drop by with
> comments on various things, eventually accumulating the knowledge structures necessary to be of increasing value
> to all who would partake of that web site. No one should be expected to do it all, alone.
> OTOH, there remains the context of *personal* knowledge management. Large chunks of that, I would imagine,
> would never be done at a public site. That is the context in which you get to do it all, yourself.
Yeah, it's called collaborative treasure hunting! E.g., UC Berkeley's
Multivalent Document(MVD) JAVA applet Browser : Informed Web-Document
Access through Shared Annotation Tracker
MVD 1.0a3 is a Java applet, implemented in JDK 1.0.2:
MVD tour guide, including a GIS viewer:
Once you feel comfortable with the MVD applet, you can run it directly
via the following examples:
Since the bootstrapping process that Doug is advocating for the grand
challenge for the next decade, "The Information Utility: Make it easy
for everyone to store, organize, access, and analyze the majority of
human information online", am wondering if UCB's Multivalent Document
(MVD) Browser would be of any help for us to "improve on what we have
now and would provide a tool with which we could experiment"?
Apparently, the MVD browser has the ability to browse web objects with
the associated annotations to enable users to share each other's
ByungHoon Kang, a third year graduate student (510-642-8468,
"Multivalent documents (MVD) represent an open, extensible,
network-centric document model. Conventional documents are monolithic
entities, generally of a particular type (ASCII, HTML, XML, PDF, scanned
image files, RTF, word processor formats, etc.). To view, edit or
otherwise manipulate such a document, you need a particular viewer or
editor; moreover, functionality is limited by the nature of the
particular format and associated programs.
With MVD, you can, in principle, open and manipulate any document
format. Moreover, functionality (including what document formats are
supported) is completely extensible. If you want a new feature, you can
add it yourself, and share the resulting capability with others. Also,
because the architecture is intrinsically distributed, you can easily
perform functions with MVD that are not readily available elsewhere.
... We have designed an annotation-tracking service using the MVD
(Multivalent Document) model and developed a prototype for finding web
objects more expeditiously through parallel collaboration, efficient
sharing/storing/retrieving of annotation mappings using DB, and querying
the annotations with various criteria such as keyword, author and base
Collaborators can use the annotation tracker to track their
collaborator's previous works and notes in hunting the treasure on the
web. Since the collaborator can notice the existence of their
collaborator's annotation through implicit retrieval when they browse
the same document, the overlapped and redundant collaborative tasks are
Also, ... "In the annotation viewer (MVD viewer), a user can query the
annotation with parameters such as base URL, author name and keyword
with menu driven methods. Query results are collected into the list
window of the annotation tracker. The annotation tree viewer can be
used to visualize the author list in a way where the history of the
follow-up annotations is visualized in a tree format."
>> [Garold L. Johnson]
>> As an example, consider the chaordic site that Bernard referenced in
>> the earlier message. There is a lot on that site that is of definite
>> relevance to the bootstrapping process that Doug is advocating, and
>> that is possibly also relevant to the attack on complex problems .
>> Should Bernard (in an ideal world) have been expected to cross link
>> all of the relevant information on that entire site into the
>> structure of our total discussion on the process of bootstrapping
>> and the solution of complex problems? That seems to be a totally
>> unrealistic expectation.
>> More realistic is what actually happens here we now have a link.
>> Several people will look at the material and comment on various
>> parts of it (which they would link into the system). This gradually
>> ties the material into ever more places as it is digested by those
>> collaborating on it. Assuming that the information continues to
>> prove relevant, it will eventually be woven into all sorts of places
>> in the knowledge structure. Every now and then someone will
>> undertake to do a more thorough job of integration and link large
>> amounts of information into the structure.
This archive was generated by hypermail 2b29 : Wed Apr 18 2001 - 12:43:29 PDT