The structure of an agent:
Independent of the choice of attributes and the problems of definition that result from this, every agent has a structure that makes it possible to characterize it as such. Diagram 2.11 enables us to understand the typical components of an agent better. The Caglayan and Harrison model encourages us to consider four levels of analysis.
In the next extract, we will deal primarily with agents that operate in the Internet and Intranet environments in order to perform strategic intelligence tasks (information retrieval and strategic watch).
- An agent is by nature task-oriented. Agents can therefore be analyzed according to their capacity to solve more or less precise tasks (information retrieval and filtering, strategic watch...).
- An agent has a certain level of "knowledge", in other words, rules of operation which define its architecture. These rules can either be imposed by the developer, suggested by the user, or acquired in an autonomous way by the agent itself (learning agent).
- An agent acts in various environments. For example, there are agents that operate on your computer desktop (in other words, at the level of your operating system), others that work on the Internet, and others work on your Intranet.
- Last, an agent has a more or less developed capacity for communication, which enables it to interact with the user or with other agents.