In the software and knowledge engineering literature we can find maintenance strategies offered to maintain seven main types of knowledge: words; sentences; behavioral knowledge; and meta-knowledge. Meta-knowledge divides into problem solving methods, quality knowledge, fix knowledge, social knowledge and processing activities. There are five main ways in which these seven knowledge types are processed: acquire, operationalise, fault, fix and preserve. We review systems that contribute to these 7*5 = 35 types of knowledge maintenance to make the following conclusions. First, open issues with the current maintenance research are identified. These include (a) areas that are not being addressed by any researcher; (b) the recursive maintenance problem; and (c) drawbacks with rapid acquire systems and the operationalisation KM assumption. Secondly, a process is described for commissioning a new maintenance tool. Thirdly, a general common principle for maintenance (search-space reflection) is isolated.

Dealing with context seems to be a major challenge which the field of Artificial Intelligence (AI) will have to face in the next few years. The importance of the challenge is borne out by the various scientific events focusing on context which have been held since 1995. This does not mean that context has not been studied before—it has in such domains as Natural Language Processing—but it has yet to be studied in depth since previous research has only covered few aspects of the problem. We present in this paper a survey of the literature dealing directly and explicitly with context whatever the domain. This makes it possible for us to have a clear view of context in AI. One of the conclusions of this survey is to point out the existence of different types of context in areas such as the representation of knowledge in a computer system, the reasoning that the system carries out using the knowledge, and the interaction the system has with people.

Both the advent of large communication networks and the development of agent models designing more and more sophisticated (in some cases autonomous) behaviours, are producing what can be called an Agent Society. Agent societies are constituted by both artificial and human agents (in perspective indistinguishable from each other) communicating with each other (not necessarily each with all the others), with individual or collective tasks, different resources, different skills, and so on. In other words, these agent societies will become more and more similar to the human ones. As in real societies, in these virtual societies it will be necessary to consider all the problems connected with secure and reliable interaction, with trusting the other agents, with the possibility to be frauded, cheated, deceived in some exchange or relationship.