Citation Context

...ased on CYC). Such an approach is compatible with ours and can be used to add traditional structured database external sources to our basic information agents. The ontological mismatch problem (e.g. (=-=Gruber 1993-=-)) is still a difficult one and is outside the scope of this paper. Figure 1 shows an abstract view of a single information agent in relation to its environment. An information agent has three concept...

Citation Context

...to make themselves known as information providers to the larger agent society. When an agent is created, it advertises itself to some entity such as amatchmaker or broker (Genesereth & Katchpel 1994; =-=Finin et al. 1994-=-; Kuokka &Harada 1995; Decker, Williamson, & Sycara 1996b). This advertisement, expressed in terms of an agent’s informationbase schema, specifies the information services that the agent is making ava...

Citation Context

...es et al. 1995), are common to all classes of information agents as well as to the task and interface agents we have built. The architecture presented here is consistent with formal BDI agent theory (=-=Cohen & Levesque 1990-=-; Rao & Georgeff 1995). The larger part of the architecture (outside of the local infobase cache) is shared by all of our software agent classes. This differentiates our approach from approaches such ...

Citation Context

...tion in a vacuum. They have to make themselves known as information providers to the larger agent society. When an agent is created, it advertises itself to some entity such as amatchmaker or broker (=-=Genesereth & Katchpel 1994-=-; Finin et al. 1994; Kuokka &Harada 1995; Decker, Williamson, & Sycara 1996b). This advertisement, expressed in terms of an agent’s informationbase schema, specifies the information services that the ...

Citation Context

...). Thus, these simpler agents can work effectively with one another as well as with more complex agents, such as task agents, that reason about commitments explicitly to produce coordinated behavior (=-=Decker & Lesser 1995-=-). In particular, in the WARREN portfolio management multi-agent system, basic andmulti-source information agents work alongside more complex task- and interface-agents. We have used (and reused) the ...

Citation Context

...nteracting with an individual user, task agents involved in the processes associated with arbitrary problem-solving tasks, and information agents that are closely tied to a source or sources of data (=-=Sycara & Zeng 1996-=-). These three broad agent categories share common architectural components but have different sets of agent behaviors. The behaviors in each class of agent are reusable. We have identified the behavi...

Citation Context

...ion agents and the information that they provide, is itself an information agent. One of the reusable behaviors inherited by a matchmaker is the ability to process persistent queries as described in (=-=Kuokka & Harada 1995-=-). The agents in our system communicateusingKQML (Finin et al. 1994). Our focus on long-term behaviors, such as periodic queries and information monitoring, has required us to extend the language with...

Citation Context

...roduce inconsistencies between the external source and the cache. A recent discussion of some of these tradeoffs in the context of higher-level multi-source information agent caching can be found in (=-=Arens & Knoblock 1994-=-). Agent Architecture: Building Blocks for Agent Behaviors An information agent’s reusable behaviors are facilitated by its reusable agent architecture, i.e. the domain-independent abstraction of the ...

Citation Context

...schema, and a set of generic software components for knowledge representation, agent control, and interaction with other agents. The generic software components, based on DECAF (Decker & Lesser 1995; =-=Oates et al. 1995-=-), are common to all classes of information agents as well as to the task and interface agents we have built. The architecture presented here is consistent with formal BDI agent theory (Cohen & Levesq...