Databases Reference
In-Depth Information
Figure 7: Transformation of a Web page into a TV-program-like content
the screen for the passive viewing, where the transformed Web page is played like
a TV program. The original image appearing in the Web page is placed on the
panel, and two character agents explain the image. The two characters also behave
some performances (e.g. point out to the panel, walk around the panel etc.) during
their speech. Basically, text portion in a Web page is used as a character's lines,
and is spoken by the character. Images are basically placed on the panel appearing
in the TV studio set. Since the Web pages have their own structures that are
represented by HTML tags, our system automatically analyzes the structure, and
transforms the content of a Web page into a TV-program-like content according
to the user-specified TV-program style.
Our approach requires to transform the content of a Web page into a
synchronized data streams by time code. When nothing considered, images and
it's explanation text in the Web page are not automatically synchronized. This is
because HTML itself does not have any tag concerned with synchronization of
components. Therefore, it is necessary to guess which text synchronizes with an
image. We explored an algorithm to discover possible synchronizable regions, in
which a text describes an image. In finding such synchronizable regions, we mainly
consider to find the
minimal tagged region,
which intuitively means a minimal
portion of a Web page that is surrounded by a tag
< X >
and
< /X >
and that
contains a specified image and some sentences. Usually, table tags of HTML are
used to represent the synchronization between an image and its explanational
text. Our algorithm to find minimal tagged regions can cope with the case of using
those table tags.
3.2
Scripting-XML (SXML)
In order to create TV programs, we need to consider not only the content of the
TV program, but also the presentation styles such as casting, camera movements,
light ing, and characters' performances. Conventional markup languages such as
