Week 9 Flashcards
Explain what is meant by an intelligent agent.
Ans: An intelligent agent is a computing entity to whom a task can be delegated by the user and the agent gets the task done for the user.
There is a difference between the ‘traditional software’ that we mean and the ‘agent’. Take an example of ‘printer software’. You need to print a document, you use the software to select a printer connected to PC or in the network and issue the printing command. It will get the document printed, if the printer is not working, you might get an error message. If you want to choose another printer
when one fails, you have to do it manually. But a ‘printer’ agent will do more that. You simply assign the task of printing to the agent. Then it will find the nearest printer (or other printer of you preference that you knows), get the printing done and notify that the printed is done. If that printer is not working, it will continuously look for other printers which are currently available, may be in another room or even in another floor, maybe with the one with shorter printing queue (so that it can be printed quickly), ie., in other words, it will apply it judgment and intelligence to find out the right printer for you. Once the printing it done, it will notify you that the printing is done and in which printer. So you can see that a ‘printer agent’ is something to whom you simply delegate the task just like a human agent and then it is upto the agent to get it done. In the process of doing that, in may interact with the user just like human being does. The principle key here is the intelligence. However, the agent technology is still developing, useful practical applications are expected in future.
List and briefly explain the defining properties of intelligent agents.
Delegation: the user will delegate the task to the agent to carry out.
Communication: It will receive instructions from the user and also will inform about task status to the user. An agent may also need to communicate with other agents in order to perform the delegated task.
Autonomy: operates without direct intervention of humans or other agents, ie, it will apply its own intelligence. It has control over own actions and internal states.
Monitoring: it has the ability to check attributes of the environment.
Actuation: it has ability to act independently and interact with its environment.
Mobility: an agent may need to be executed on a remote server.
Outline the kinds of knowledge and skills that an agent may need to possess.
Task Level Skills: skills needed to perform the allocated task.
Knowledge: just like human beings it needs knowledge and intelligence to do things. An agent is built to do certain tasks, so knowledge and intelligence can be put at the development phase (mostly as in a rule based system, an example of simple rule “if it is raining, carry your umbrella”,
ie., you sense you environment first and then act accordingly based on known knowledge). But for
such an intelligent system to work, it must also learn over time through new situations it faces, i.e, through experience (just like we human being learn through experience). There many artificial intelligence based learning techniques that are employed to train an agent over time.
Communication skill: it also needs to communicate with the user and other agents during performing the task.
Provide classifications of agents according to their operating environment, task capabilities or
architecture.
Ans: It can be classified as follows:
Operating environment: eg: desktop, operating system, network, Internet, application.
Task capability: what task the agent is supposed to do, eg: retrieval, filtering, searching.
Architecture: depends on internal knowledge architectures, eg: learning agents, neural agents.
Outline five application domains for intelligent agents.
Industrial: Process control, manufacturing, air traffic control.
Commercial: Electronic Commerce, business process management, data-mining.
Medical: Patient monitoring, health-care.
Entertainment: Game, interactive theatre and cinema.
Personal task: organize personal activities and building communities.
Briefly discuss three different applications of intelligent agents in electronic commerce.
Automated market research: one of major usages of agents in electronic commerce would be to employ them to conduct market research for a business.
Prepurchase Interaction: it can search, select and negotiate on behalf of the user. There are some prototypes built in research lab, for example, Kasbah by Media Lab, MIT and AuctionBot by AI Lab, University of Michigan.
Purchase consummation: On behalf of the user, a buyer agent can place order, organize payment etc. A seller agent, on receiving payment, can organize delivery of services and products. Since payment is involved here, security issues need to be considered if a forged payment is shown by a buyer agent.
Post-purchase interaction: A seller agent can create a user profile of buyers when a purchase is complete. It can also organize personalized after sale service. For example, a company sells a heavy-duty printer to an organization and a printer can contain an agent. If something is wrong with the printer, this agent can automatically contact the seller agent to organize a check by the seller’s technicians. Similarly, it can contact seller agent when a service is due.
Discuss what benefits agents can offer in electronic commerce.
The biggest benefit is productivity and efficiently. A human agent can’t work with same efficiency all the time, but the software agent can work with then same efficiency all the time.
Similarly, agent to agent communication would be cheap using internet.
An agent can work 24 hours a day, seven days a week, 365 days a year.
Agent to agent communication can be done at any time, irrespective of the time zone.
When agent technology matures, small business organization can also have agents.
As indicated in Q.7, better and personalized after sell service can boost business reputation for a company.
Discuss some current impediments to the widespread adoption of agents in electronic commerce.
Agent technology has not been matured yet. At the current state, an agent’s performance is not up to the level that a business can fully rely on it.
It limits user’s direct involvement and reduces impact of advertising investments.
Malicious mobile agents may pose security threats. Agents may need private information from the users to work which also creates a risk of privacy breach.
What are the future prospects for agent technology in electronic commerce?
Agent technology can certainly have enormous impact on the way electronic commerce, or business in general, is done. However, the technology is not so mature at the moment to fully rely on it. Once matured and security issues are resolved, it is expected that agents will receive wide acceptance.
