No abstract

Analogous to the meta-data that composes "data about the data" for library/information collections, there is a growing interest in employment of "bits about the bits" for compressed audio and video documents. The idea is to provide content description, or meta-data, of compressed multimedia documents, so that users could make use of such description to search and retrieve multimedia data/documents, as they do with Dublin Core bibliographic information to library collections. An ongoing endeavor undertaken by the Motion Picture Expert Group (MPEG) is to call for a new international standard, known as MPEG-7. When it is finalized in 2001, MPEG-7 is expected to provide standardized schemes for concise and unambiguous multimedia content description for different types of multimedia data/documents in a wide range of applications. Such content description should be denoted in the standardized description language, so as to support machine parsing and interoperability. In this paper, we report our recent work related to MPEG-7. We will present the Multimedia Description Definition Language (MD 2 L), which provides a set of formal notation for defining Description Scheme Definition (DSD) for various types of media data such as audio, video, and etc. We will also discuss our on-going implementation of Java-based mechanisms for parsing and indexing/searching MD 2 L description.

A practical and effective data model for the interactive multimedia is provided in our project. It is shown in this paper that our data model is not only adequate to represent various types of media entities including the audio, video, image and text, but also competent for facilitating hypermedia relationships among different entities. Based on such a data model, we have designed and developed an object-relational multimedia database infrastructure, with a number of useful schema's for supporting multimedia services. Apart from standard authoring, editing, and searching/querying features, advanced options such as hypermedia navigating and browsing are also supported in our object-relational multimedia database (ORDB). A prototype of the proposed ORDB is being implemented in our project for interactive multimedia applications.

No abstract

This paper provides a formal speci cation in Z of a new intelligent hypertext model called the soft-link hypertext model (SLHM). This model has been implemented and extensively tested, and provides a new methodology for constructing the future generation of information retrieval systems. Its core is the adoption of a data structure called the conceptual index, which allows hypertext structure to be built automatically upon conventional Boolean systems. The functionality of resulting systems using this approach is then extended from Boolean search to more sophisticated information retrieval applications, including associative searches and information browsing. Compared with other similar projects, SLHM has the following three major advantages. First, the conceptual index is automatically formulated. Second, powerful neural learning mechanisms are applied to the conceptual index, thereby improving its e ciency and applicability. Third, machine intelligence installed on the conceptual index can be utilised for online assistance during navigating and information browsing. This speci cation has been developed by application of an existing formal framework for specifying hypertext systems. The speci cation presented here provides: a formal account of the state and operations of this new model; a sound basis for instantiations of the model to be built; a case study in the application of an existing formal framework; and an environment in which further re nements of new learning and hypertext strategies can be presented and evaluated.