[Top]

[Back]

[Next]

[Bottom]

7 Conclusions

"Include me out" - Sam Goldwyn

The growth of inter-networking that has recently taken place has demonstrated that the electronic information systems of today are providing the user with more electronic information than has ever been previously available. As more organisations and users `get on-line', then not only will their needs and expectations increase, but the burden on current and new systems will increase, expecting them to provide more information resources.

Internet users of today are given relatively few tools to assist them in navigating through the plethora of distributed information systems that exist. The most rapidly expanding protocol currently within the Internet is the World Wide Web (Web) and the tools that people have developed for it are firmly centered around the resource discovery area; there are almost no management tools that exist for controlling and manipulating distributed Web resources.

Distributed information management is a term that is being used more frequently to describe a set of tools that provide four basic functions; resource discovery, information integrity, navigation assistance and system integration. These functions assist the user in navigating and managing distributed information resources, such as the Web, FTP, Gopher and Hyper-G. It is proposed that if these tools were available, the user would be better served by distributed resources.

However, most current tools are purely manipulation tools which require the user to maintain some form of interactive control over them, for example, an FTP session. Agent technology is a step away from direct and interactive tools into the realm of task delegation; agents are essentially processes that possess characteristics that enable them to perform tasks that would normally be carried out by the user.

These characteristics define the nature and limitations of agents and also enables the user to predict the behaviours of their agents. One of the most important features that agents must exhibit is trustworthiness; an agent that performs its task badly is of little use to a user if they not only have to perform the task by hand, but also have to repair the damage caused by the agent in the first place.

To assist agents in completing the tasks that users assign to them, it is proposed that they should possess the meta-characteristics of intelligence and learning. Intelligent agents can make reasoned and informed decisions based upon current knowledge; intelligent, learning agents can make reasoned and informed decisions based upon current knowledge and past experiences. These two facets are seen as crucial in agents gaining the trust of users.

Mobile agent technology is receiving much current interest and is seen as the next evolution in programming after the object-oriented paradigm. Mobile agents make a significant contribution to accessing distributed information in a natural fashion and for building applications and then integrating them into a distributed environment. However, contemporary mobile agent systems have put aside the more traditional view of agents (intelligence, learning, etc.) and have concentrated on the aspects of agent migration, agent to agent communication and the security, authentication and validation of agents. This technology, whilst promising for the future is not inherently realisable, since an important layer is missing between integrating mobile agent systems with distributed information systems, the desktop and (ultimately) the user.

The research presented in this thesis has advocated building a two-tiered approach to bringing mobile agent technology and distributed information management to the user; a lower strata dealing with process-level concepts (migration, communication, security) and an upper strata to handle the distributed information management tools, represented as component mobile agents. This approach allows mobile agents to reside beside the application and across distributed applications.

The framework that has been described attempts to provide a solution to the upper strata; it details how various agent types can integrate with distributed resources, users and other agent to provide the fundamental requirements of distributed information management. It is the future work of this research to develop a prototype lower strata upon which an upper strata can be developed and the integration of distributed information systems tested. It is further hoped that the development of mobile agent-based distributed information management tools will provide the user with the methods necessary to manage, navigate and discover distributed information resources that exist within the global information space that is developing.

[Top]

[Back]

[Next]

[Bottom]