Based on our previous discussion, I modified the initial concept map of SI-DL information concepts. As before, documents and sub-documents are digital objects. I added a metadata catalog, metadata specification, and a collection for digital objects. Sub-documents and documents have their own metadata catalog, metadata specification and collection. I welcome comments and feedback. Thank you.

-Uma

Today, we had an interesting class discussion that got me relating what we were talking about in class to a potential SI application.

This semester I am taking a course called Personal Information Management (PIM). The course is taught by Dr Manuel Perez and deals with management of personal information including emails, files, calendar, to-do lists, pictures, music, and even laundry! It is a very interesting class where I am learning not only about PIM theory but also about practical problems faced by people in managing their personal information.

Today in class, Manas (a classmate) brought out an interesting problem. He said that if there were some way of being able to link different personal information on his computer, that would really help. Here is an example - linking notification about a conference (email) with submissions to that conference (files), with conference-related meetings (calendar), and with more information about that conference (web page). This would significantly improve our information re-finding activity.

Then it struck me that many times we are interested in pieces of information within these different personal information types. For example, if I subscribe to a listserv in daily digest mode, I may want to link a relevant message within the daily digest to some code on my computer or some documentation. Another example would be linking an email from someone with their contact phone number on a website. If these pieces of information and links would somehow be indexed, I could use a tool like Google Desktop to easily re-find information that I want. I think that this would reduce noise because links created by the user could be given higher weight while computing the rank of the results of a query.

Another problem people face is not being able to easily annotate over emails or other personal information. Many times when we receive or come across some information (via an email or a web page that we come across), we do not know where in our taxonomy of personal information this new information would fit. Initially, we do not have a tag or a folder for it. However, later when we use it for something, we want to remember that information using the context in which we used it. Adding an annotation with catchwords (or keywords) to that information may help us re-find that information more easily. Or simply highlighting some important text in that email will help in our re-finding task.

A potential SI application that I see in this discussion would provide the following functions:

• Allow linking of pieces of personal information (that may be of several different types)
• Allow annotations over personal information
• Organize and index all superimposed information, marks, and base information for better information re-finding functionality.

-Uma

Distribution Alternatives for Superimposed Information Services in Digital Libraries

Sudarshan Murthy, David Maier, Lois Delcambre

In: Lecture Notes in Computer Science; Pages 96-111; Volume 3664; 2005; Aug; ISBN: 3-540-28711-6.

(Selected papers from: The Sixth Thematic Workshop of the EU Network of Excellence DELOS on Digital Library Architectures; 2004; Jun 24-25; Cagliari, Italy.)

http://dx.doi.org/10.1007/11549819_7

By definition, a mark is a selection of a (continuous) region in a base document. We can infer that a mark refers to sub-document information. Using this idea, I built a concept map to represent information concepts in, a digital library (DL) that would support SI. The concept map was derived from a figure that originally described DL concepts from the information perspective. Feel free to comment on or modify this figure.

We still need to think about where SI tools/applications. The DL metamodel in Marcos' dissertation (see page 62) may help with some ideas.

p.s. I acually created a mark pointing to the figure caption in Marcos' dissertation. However, the blog editor does not accept non-http URIs and gave an error when I used it. If you want to try the mark, please go here. You will also need Mozilla Firefox to use it (since IE has an restriction on the length of a URI). For some reason, the page in Marcos' dissertation is resolved correctly, however, the selection is not seen.

- Uma

It seems to me that there are a few options about how marks are modeling in a DL.

1. Marks (into a base document) are stored in a metadata field for the base document
a. this means that ALL marks into a base document are in one place
this is sort of the opposite of Sidepad (which doesn't mean it's not desirable)
b. this means that a base document "knows" about it's marks (also new; also interesting)
c. can we have multiple collections of marks? Perhaps some are personal, some are shared.
Heretofore, we (at PSU) haven't
considered having a set of marks, by themselves. We think of marks as existing in a
superimposed document. So...maybe we also want to allow superimposed documents
like Sidepad documents, to be placed into the digital library.
2. Marks (into other (base) documents) are stored in a metadata field for the superimposed
document. (Maybe they are also stored in-line. The interpretation of in-line marks
as in a Sidepad document) is left to the base application. Whenever a document is
deposited or redeposited into the DL, then this metadata field holding outbound marks
would have to be updated.
3. Marks are first-class elements in a DL; they have their own metadata.
4. Collections of marks are first-class elements in a DL; they have their own metadata.
Note: it seems that marks that are in a DL should ALWAYS indicate the document that they are marking into - in a way that allows the DL to retrieve the document.

That's it for now. Perhaps we can consider this during our conversation about DL architecture. Note this note was inspired by Ed's comment on Monday where he said that he and Boots were considering storing marks in a metadata field. Perhaps Ed/Boots have something else in mind other than the options I've listed here.

lois