introduction multimedia
[] readme preface 1 2 3 4 5 6 7 appendix checklist powerpoint resources director

talk show tell print

2.1 information spaces

Current day multimedia information systems distinguish themselves from older day information systems not only by what information they contain, that includes multimedia objects such as images and sounds, but also by a much more extensive repertoire of query mechanisms, visual interfaces and rich presentation facilities. See  [Spaces].

S.K. Chang and M.F. Costabile -- Visual Interfaces to Multimedia Databases


The Handbook of Multimedia Information Management


Preceding the advent of multimedia information systems, which include networked multimedia systems as discussed in section 6.3, we have seen advances in

multimedia information systems

Now, the class of multimedia information systems is, admittedly, a large one and includes applications and application areas such as:

multimedia applications

geographical information systems, office automation, distance learning, health care, computer aided design, scientific visualization, information visualization.

Nevertheless, irrespective of what technology is used for storage and retrieval, multimedia information systems or multimedia databases impose specific requirements, with respect to: the size of data, synchronisation issues, query mechanisms and real-time processing.

multimedia databases

  • the size of data,
  • synchronization issues,
  • query mechanisms, and
  • real time processing.
Partly, these requirements concern the efficiency of storage and retrieval and partly they concern aspects of usability, that is the way information is presented to the user. In particular, we can think of a great number of query mechanisms that our multimedia information system of choice is expected to support: free text search, SQL-like querying, icon-based techniques, querying based on ER-diagrams, content-based querying, sound-based querying, query by example, and virtual reality techniques.

query mechanisms


  • free text search
  • SQL-like querying
  • icon-based techniques
  • querying based on ER-diagrams
  • content-based querying
  • sound-based querying
  • query by example
  • virtual reality techniques

logical information spaces

 But before thinking about the optimal architecture of multimedia information systems or the way the information is presented to the user, let's consider in what way a multimedia (information) system or presentation may be considered an information space.

As a tentative definition, let's assume the following

definition: an information space is a representation of the information stored in a system or database that is used to present that information to a user.

This may sound too abstract for most of you, so let's have a look at this defininition in more detail.

First of all, observe that when we speak of representation, and when we choose for example a visual representation, then the representation chosen might be either

In principle the same holds for a text-based representation, but this is far less interesting because the options in choosing a representation and presenting it to the user are much more limited.

Unfortunately, the phrase representation is also somewhat vague. To be more precise,

we must distinguish between a visual information space (for presentation), a logical information space (in which we can reason about abstract information objects) and a physical information space (where our concrete multimedia objects are stored).

Summarizing we have:

physical information space

  • images, animations, video, voice, ...

logical information space

  • abstract database objects

presentational information space

  • to present information to the user
Our visual information space, our presentation space, as you may prefer to call it, might reflect the logical information space in a symbolic manner by using diagrams, icons, text and possibly visualzations, or, going one step further, it may also mimic the logical information space by using virtual reality, as discussed in chapter 7. 


  mimic                           -- virtual reality
  
   or
  
  reflect                         -- diagrams, icons, sketches

Now we can give a more precise definition of the notion of information space, in particular logical information spaces:

definition: a logical information space is a multidimensional space where each point represents an object from the physical information space (read: database).

First of all, observe that when we speak of dimensions we might also speak of attributes that can take either continuous, numerical, discrete or logical values.

unified view of the database

dimensions = attributes

  • continuous, numerical, discrete or logical
So, concretely, these attributes may be directly or indirectly related to information stored in the database, and hence we can give a more precise definition of the notion of (multimedia) information objects, queries and cues (in the logical information space).

(multimedia) information object

  • or example, is a point in the (logical) information space

query

  • is an arbitrary region in this information space

clue

  • is a region with directional information, to facilitate browsing

both examples and queries are clues


The notion of clue is actually quite interesting, since both examples and queries may be regarded as clues, that facilitate browsing through the contents of an information space. As an example, just think of the situation that, when looking for another notebook, you want something that is similar to the the thing you've previously seen, but that has an additional video output slot that may be connected to your TV.

Also, clues are needed to allow for query by example. In this case you need to help the user to define a query in the logical information space, so that the system can construct an optimal query to search for the desired object(s) in the physical information space.

When we regard the information retrieval problem to be

information retrieval problem


the construction of the optimal query with respect to the examples and clues presented by the user,

then we may characterize the optimal query as

optimal query

the one that will retrieve the largest number of relevant database objects within the smallest possible region in the (logical) information space.

visual reasoning may help the user ...


extensions

Given the stratification, that is levels or layers, of information systems discussed above, we can think of improvements or extensions on each level.

physical layer

  • networked multimedia -- client/server (6.3)

logical layer

  • information hyper space -- chunks and hyperlinks (2.2)

presentation layer

  • virtual reality interface -- physical location of student records (7.2)

VU Campus


Each of these improvements or extensions can be regarded as a technological or scientific adventure in it's own right. (See the sections indicated inbetween the brackets.)

research directions -- universal interchange

Technology changes rapidly. Just think about the development of the PC in the last two decades of the previous century. And applications change rapidly too. At the time of writing the web does barely exist for ten years. Information spaces, on the other hand, from a sufficiently abstract perspective at least, should be rather stable over time. So the question is, how can we encode information content in an application-independent way? As a remark, application-independence implies technology-independence. The answer is, simply, XML. The next question then should be, what is XML and why is it more suitable for encoding information then any of the other formats, such as for example relational tables.

The first question is not so difficult. There are many sources from where an answer may be obtained. Perhaps too many. A good place to start is the XML FAQ (Frequently Asked Questions) at the Web Consortium site:

www.w3.org/XML/1999/XML-in-10-points


XML is a set of rules (you may also think of them as guidelines or conventions) for designing text formats that let you structure your data.

More specifically, XML may be characterized as follows:

XML in 10 points


  1. XML is for structuring data
  2. XML looks a bit like HTML
  3. XML is text, but isn't meant to be read
  4. XML is verbose by design
  5. XML is a family of technologies
  6. XML is new, but not that new
  7. XML leads HTML to XHTML
  8. XML is the basis for RDF and the Semantic Web
  9. XML is license-free, platform-independent and well-supported
Perhaps not all of these points make sense to you at this stage. So let me first indicate that XML has in fact quite a long history. XML is the successor of SGML (the Structured Generalized Markup Language) that was developed in the 1980s to encode documents (such as airplane manuals) in an application-independent manner. SGML is not a language itself, but a descritpion of how to create a content description language, using tags and attributes (as in HTML). In fact, HTML is an application of SGML, using tags with attributes both for formatting and hyperlinks. In other words, SGML is a meta language. And so is XML. Since everything got messy on the web, XML was proposed (as a subset of SGML) to make a clear distinction between content and presentation. Presentation aspects should be taken care of by stylesheets (see below) whereas the content was to be described using and XML-based language.

Now, why is XML a suitable format for encoding data? That question is a bit harder to answer. One of the reasons to use XML might be that it comes with a powerful set of related technologies (including facilities to write stylesheets):

related technologies


  • Xlink -- hyperlinks
  • XPointer -- anchors and fragments
  • XSL -- advanced stylesheets
  • XSLT -- transformation language
  • DOM -- object model for application programmer interface
  • schemas -- to specify the structure of XML documents
These technologies (that are, by the way, still in development) provide the support needed by applications to do something useful with the XML-encoded information. By itself, XML does not provide anything but a way to encode data in a meaningful manner. Meaning, however, comes by virtue of applications that make use of the (well-structured) data.

In summary, XML and its related technologies provide the means to

XML


  • separate data from presentation
  • transmit data between applications
Actually, the fact that XML was useful also for arbitrary data interchange became fully apperent when XML was available. To get an impression of what XML is used for nowadays, look at www.xml.org.

This leaves us with the question of why XML is to be preferred over other candidate technolgies, such as relational databases and SQL. According to  [XSLT], the answer to that question is simply that XML provides a richer data structure to encode information. In the multimedia domain we see that XML is widely adopted as an encoding format, see section 3-2. For an example you might want to have a look at MusicXML, an interchange format for notation, analysis, retrieval, and performance applications, that is able to deal with common Western musical notation as used from the 17th century onwards. In appendix XML we will explore how XML might be useful for your own multimedia application by treating some simple examples.
[] readme preface 1 2 3 4 5 6 7 appendix checklist powerpoint resources director
eliens@cs.vu.nl

draft version 1 (16/5/2003)