This material is based upon work supported by the National Science Foundation under Grant No. 0082817
Last updated on September 30, 2007

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.

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.