|
This page contains a brief overview of UCLA's Multimedia
Stream System (MMSS) research project. Read it to get a general idea
of what MMSS is and what it aims to achieve. |
|
 |
This project investigates and develops stream-based
data model constructs and the corresponding querying facilities, in
response to the growing requirements of advanced multimedia database
applications. |
|
 |
Streams represent temporally ordered sequences of data,
and are a natural means by which to capture the temporal nature of
multimedia information. Using the basic concept of a stream, we
describe new data modeling constructs: substreams, aggregated streams,
and derived streams. Substreams enable users to characterize and
manipulate subsets of information within a stream based on its logical
contents. Aggregations of streams represent the conglomeration
of multimedia temporal sequences, dependent on a specified aggregation
behavior (e.g., temporal union vs. temporal
intersection). Multimedia data that are often operated upon
through a variety of methods (e.g., image filters) to generate
new data, are represented using derived streams. These stream
constructs complement existing object-oriented database models, and
increase the overall flexibility in multimedia data
representation. Our proposed work will formalize these
constructs, and develop tools and a database to facilitate creation of
schemas based on this stream data model. |
|
 |
The modeling of relationships between multimedia objects is
investigated. The various types of relationships between and
within streams, and with standard database entities, are explored and
developed as part of the data model. In addition, due to the
growing wealth of information appearing on the World Wide Web (WWW),
our model should support the establishment of links from the database
to these external data sources. |
|
 |
Query facilities are needed to support these new stream
and relationship constructs. We propose to develop a stream
algebra to express queries involving stream constructs and explore
methods for effective stream query processing. A visual query
language is described, allowing end-users to make use of the
underlying data modeling constructs and stream algebra. Certain
queries may be better expressed using a declarative syntax, rather
than with a visual metaphor. Thus, we will also extend and
develop additional constructs for the Object Data Management Group (ODMG)
data model and the Object Query Language (OQL) in further support of
the data modeling advances in this proposal. |
|
 |
Our work has been inspired by multimedia medical applications and geophysical domain applications. We have developed an image stack multimedia data model and an end-user interface, called TimeLine, to aid in the temporal visualization of a patient's medical history and of geophysical data behavior. A version of the TimeLine application predicated upon the problem-oriented organization of patient information stored in streams and the related constructs has been implemented in a clinical test bed environment. Similarly, the TimeLine advances have been applied in a geophysical domain test bed environment. Patient data and physicians from the UCLA Medical Center formed the foundation of a test bed to evaluate and validate the effectiveness of these software technologies developed. Likewise geophysical and environmental specialists at UCLA evaluated and validated the effectiveness of the technologies.
See the various co-authored publications, and screen shots and some demos of the image stack and Timeline resulting from this multidisciplinary R&D project. |
|