4.7 How Agents Work

Regardless of whether it is an open or a closed type of agent, or of the function that it performs, its benefits are usually in automating some type of repetitive behavior that is either time-based or event-based. They can automate repetitive tasks, such as performing a common query against a database. More advanced agents can notify specific users of the arrival or creation of new data ready for their analysis; they can assist users with more advanced analyses, guiding them in processes they are not knowledgeable about; and lastly, they can perform messaging tasks, such as notifying users when a model has been completed. Some data mining tools incorporate agents to automate the process of model construction and analysis.

Specialized database or network agents can, on the basis of user requests, go out and perform queries, assemble the data found into a predesigned template, or process the data through a designed analysis. Database agents provide valuable functions in making information available to users in the most useful form and context. Once the data has been retrieved and assembled or the analysis is complete, such as the creation of a data cube or a data mining model, the results can be transmitted to a designated group of users. The entire process can be done in real time or overnight; the agent can be programmed to perform the task as required by the agency or department needs. The benefits are clear: agents reduce the workload of investigators, lead to faster decision making by the analysts, and increase the productivity of everyone involved.