The Melampus Project: Toward an Okmniscient Computing System

Software technology has not evolved to accommodate the rapidly increasing volume and diversity of on-line data. Finding data is often very difficult, and once found, data from different sources is often difficult to combine because of a lack of common formats and semantics. The advent of distributed systems has highlighted the problem by multiplying the amount of data that is hard to find and hard to use. A fundamental problem contributing to these shortcomings is the lack of a comprehensive model in which to describe and manipulate the range of entities in a system. This paper introduces Melampus, a system designed to address this problem.

The overall goal of Melarnpus is to define a powerful data model, to build a prototype system that implements the model, and to build sample applications that highlight both strengths and weaknesses of the model. We hypothesize that by providing a system-wide framework in which to describe the structure and behavior of data and by providing efficient associative access to all entities, we will significantly improve the ability of users to exploit the system. By providing global system coherence, we will enable the use of data in unanticipated ways, allow rapid retrieval of data, ease the formation of new relationships among data, and promote the sharing of data between applications.