Abstract
No body can deny the important role World Wide Web (WWW) has played in providing users from all around the globe with loads of information and data bases. At the same time the emergence of the web caused a lot of obstacles in searching for the needed information. In this regard AI experts developed agent systems that provide users with assistance and advice for the efficiency of the information search road.
Agent based systems evolved to support other and endless applications, also multi agent systems were designed to ensure more accuracy in performing the tasks and to support distributed AI. Like any other invention, problems appeared especially when trying to relate agent system in any application, but at the same time the fast acceleration of those systems development didn’t slow down since they are useful in a lot of fields. Now days a number of projects are taking place to improve their functionality and capabilities.
1. Introduction
Online information retrieval at its beginnings used to be done by knowledgeable information seekers called intermediaries. Those intermediaries meet with users and individuals to help them find their desired information since they have the experience in finding user interest areas, but due to the emergence of the World Wide Web (WWW).
And the amount of information it provided which caused overload and difficulty in finding the needed data, both experienced and inexperienced users need help in the search process to save time and effort, Borgman (1986).
The Essay on Management For Information Systems
MANAGEMENT FOR INFORMATION SYSTEMS Today there are a lot of current trends and challenges in information management. The digital world is the linking of people, decisions, tasks and processes via computers and computers with other computers. Cyberspace represents the real time transmitting and sharing of text, voice, graphics, video and the like over a variety of computer-based networks. ...
To solve this problem, Artificial Intelligent (AI) experts developed software called intelligent agents or agents. The developed software can help users in browsing the related data bases provided by (WWW) more conveniently, and to ensure the relevance of the retrieved information, Tecuci (1998).
According to Russell and Norvig (2003), agent is anything that responds to the environment where it exists through sensors to receive the signal or command and corresponds to the environment via actuators. This means that agent communicates with its environment by receiving tasks through sensors and by implementing them and displaying results to the environment by actuators. Intelligent agent according to Tecuci (1998) is a knowledgeable based system that can interact with the user through natural language. It helps the user to accomplish tasks without absolute obedience from the agent but with the guidance from the user. It has the ability to distinguish between different tasks and decide which tasks to take.
The main difference between intelligent agent and agent is the word intelligent. In the term agent, the word intelligent means according to Jennings and Wooldridge (1998) the following:
• Reactivity: this means that intelligent agents can perceive their environment and respond to changes that take place.
• Proactiveness: intelligent agents exhibit goal-directed behaviour by taking the initiative.
• Social ability: this means that intelligent agents can communicate with both agents and human to satisfy their design purpose.
According to Sdsu (2002), there are other characteristics for intelligent agents such as:
• Mobility: means that the agent can travel from one machine to another using the advantage of the internet.
• Autonomy: the agent can perform a set of tasks without the user notification or confirmation. This means agent can control the actions it takes and adjust itself to the condition that is taking place. Such as night backup.
Intelligent agents are agents provided with improved capabilities to perform the purpose behind their design. And the way agents behave depends on the correspondence between external and internal agent domain, D’inverno and Luck (1998).
The Homework on A healthy and safe home based environment
It is my responsibility as a professional childminder to do everything possible to keep all children safe at all times and prevent accidents, to do so: •I will ensure I keep appropriate supervision of children at all times. •I have in place a thorough risk assessment which I enforce and revise regularly. •I use only equipment with children that is age appropriate and in safe working order and ...
2. How agents work
The way agent operates and behaves depends on the environment it takes place in. the basic way of their operations is that they help users and individuals in doing their tasks. The main functionality of the agent is based on the correspondence between external application of the agent and internal domain consisting of knowledge base and an interface engine. The knowledge base contains the data structure representing the application domain such as entities, objects, relations. While the interface engine consists of programs to manipulate the data structure in order to solve the problem the agent was designed for. Tecuci (1998).
Agents can learn from users or other agents either directly or by observing their behaviour or by own experience Muller et al (2003).
They are called learning agents and are defined by Tecuci (1998:2): “Agent that is able by itself to acquire and maintain its knowledge”. Agents can improve their performance and increase the work they can achieve through learning and this depends on how the agent can communicate and adjust with the nature of the environment it exists in Russell and Norvig (2003).
Learning from previous experiences might be crucial and important in some applications and at the same time undesirable in other applications, Wooldridge (2002).
This means that an action could be taken if a specific condition existed, but if the condition that caused the action didn’t exist in the new situation, then there is no need to repeat or use the experience learned.
In order to create a learning agent, it should include together with knowledge base and interface engine, a learning engine that is capable of updating data in the knowledge base. Learning engine can learn from sources in the environment that surrounds the agent such as other agents, users, data bases or own experience, Klusch (1999).
According to Tecuci (1998) if an agent can learn then building the knowledge base is an easy task, but the difficult part is to build and create learning engine due to the lack of understanding of the learning process. So in this case the building of learning agent should at least provide the agent with some knowledge base with the ability to customize and correct during learning process. This means that making a learning agent is very difficult task and learning methods are used to create the knowledge base.
The Essay on Learning Without Dimensions Knowledge Work
LEARNING WITHOUT DIMENSIONS In their essay, respectively, The Age of Social Transformation by Drucker and The New Humanities: Readings for the Twenty-First Century, Miller and Spellmeyer outline and discuss on some improvements on the contemporary education. Education in the United States has always been mainly in service by the public. Should or would education still be given by the public or ...
To sum up, agents include knowledge base and interface engine. While learning agents include in addition to what mentioned a learning agent. The main influence on the functionality of the agent is the environment that surrounds it. And the influence from the environment plays a big role in building the agent’s knowledge base.
3. Type of Environment
Wooldridge (2002) lists the most common environments where agents operate as following:
• Accessible vs. Inaccessible: in the accessible environment, an agent can get updated information about the status of the environment, but most environments are in accessible.
• Deterministic vs. Non-deterministic: this means that every action has a guaranteed effect. But at the same time there is no prediction of the action result.
• Static vs. Dynamic: the static environment changes only by the agent action and work it performs, while dynamic have other processes and is out of complete agent control.
• Discrete vs. Continuous: discrete environment has a fixed number of actions and reactions (effects).
While in continuous environment there is no limit of action or reactions.
• Episodic vs.non-Episodic: in episodic environment, the agent experience is divided into episodes. Each episode includes the agents receiving and then acting. Episodic environment are simple since agents doesn’t need to think ahead because the experience is divided through the whole task and next actions to be taken.
4. Multi-Agent System (MAS)
According to Ferber (1999), MAS is a system where all agents communicate with each other by sending different messages or signals. In MAS different agents can share the same goal and interact together or each agent can have a special interest area, Aiii (2000).
Agents can communicate with each other whether they were positioned in environment or not. If agents communicate without existing environment, the system is called purely communicating MAS. While if agents communicate and were positioned in environment, the system is called situated MAS, Muller et al (2003).
The Essay on Kontora 2000 System Information Users
KONTORA 2000 PURPOSE The purpose of the information system KONTORA 2000 is to provide organizations with information, which is necessary to perform routine processes, planning, organizing, control and evaluation of work as well as improvement of personnel performance. The main goal during KONTORA 2000 development stage was to create and implement a system, which can provide help in collecting, ...
Beer et al (1999) discussed the form of interaction between agents in multi agent system. The interaction depends on negotiations between agents. Agents should agree on a goal or a plan. Agents can have influence on each other to convince each other to do so and so.
The negotiations depend on three factors which are:
• Negotiation Protocol: protocols are set of rules that organize negotiations between agents. It decides the negotiators and the related third party. Also actions to be taken together with events that cause change are decided.
• Negotiation Objects: this decides the situations that should be discussed by which agreement is reached. This means the structure of agreement is fixed together with the level of acceptance or rejection. Here the action to be taken is discussed and all agents decide to agree on this action or not.
• Agents Reasoning Models: shows the decision making structure through which agents will achieve what was agreed. It means that the steps of action agreed on are displayed.
5. Applications of agents
According to Wooldridge (2002), the basic and main applications of agents can be classified as following:
• Distributed Systems: in this case agents work as nodes in distributed system. This is most commonly used in multi agent systems.
• Personal Software Assistants: agents work as assistants to users in some applications. This is commonly used in individual agent systems.
The widely distributed applications where agents are used are as following in more details:
• Agents for Business Process Management: in this case agents interact to achieve organizational objectives on behalf of individuals or companies. In the agent based business process management, the organization is treated as a society of agents providing services. It is based on transforming each department in the organization into an agent, and each employee in the department is considered as an agent. These agents communicate between each other to discuss the best way of providing services inside the institution Jennings (1998).
The Term Paper on Performance of Information Systems through Organizational Culture
The objective of this paper is to determine the importance of the connection between the organizational culture and the information system which can be vital to achieve essential business goals. However the proper definition of information system (IS) is important, as different people create confusion in this respect, which according to Anderson (1992) it is the system which captures, records, and ...
• Agents for Distributed Sensing: it is considered as the main application of multi agent systems. The main idea was to have a number of sensors (agents) that track a certain event to make the work easier. The sensors (agents) can communicate with each other and provide predictions to inform each other when the event will occur.
• Agents for Information Retrieval and Management: the agent that is responsible of helping users to track and search desired information is called information agent. The user types a query, and the information agent searches various browsers and data bases since it has authority to access various information on (WWW).
According to Wooldridge (2002), agents can help in this task because of the following characteristics that the web has:
o The web allows access to plenty of information resources without restrictions on any user.
o The web has specific interface for different multimedia.
o The web allows different documents to be linked in a good and meaningful way.
o The web interface is user friendly that is simple and professional.
Despite all these characteristics, users usually get bored from browsing the web due to information traffic jam and overload. Different search engines tried to solve this problem by arranging documents and then retrieving them if the query matches metadata. But still it is not an efficient way.
While other websites tried to attract people by providing overwhelming animations and colours, but that doesn’t change the fact that users are looking for specific information rather than the interface and appearance.
Durfee et al. (1997) stated that in order to solve those problems, different agents are used such as:
• Personal Information Agents: Maes (1994a) developed E-mail assistant software that takes actions according to the user usual actions. The agent learns from users and repeats their actions when a new event happens. In the case of new event that didn’t occur before, the agent compares the event with previous actions and tries to guess what the user will do in this case. If the agent is confident of what to do, it will perform the suggested action. Else it will suggest an action to the user.
The Essay on Information Systems 7
INFORMATION SYSTEMS In a general sense, the term information system (IS) refers to a system of people, data records and activities that process the data and information in an organization, and it includes the organization's manual and automated processes. In a narrow sense, the term information system (or computer-based information system) refers to the specific application software that is used ...
• Web Agents: Etzioni and Weld (1995) proposed the idea of web-based agents. Now web agents are seen in different ways as:
o Tour Guides: this kind of agents usually answers questions as where to go next. It either depends on learning from user preference or suggests a link that is related to the previous link.
o Indexing Agents: the idea here is to use the information provided by search engine and the user goals or interests to display a customized service to the user.
o FAQ Finders: it aims to direct users to frequently asked questions to get their answers. Since FAQ’s are learning intensive, there will be a lot of potential to find automated FAQ finder.
o Expertise Finder: tries to find what users really want in order to provide better information retrieval service.
Despite all that, search engines suffer from partial coverage of information. To solve this problem Etzioni (1996) suggested the use of Meta search engines to gather information from different search engines.
• Agents for E-commerce: agents are widely used in different applications in E-commerce such as:
o Comparison Shopping Agents: agents are used to help users find the most suitable product while buying online. It recommends product on other products and compare prices to find the best deal available. It also monitors new products and offers. This agent makes the process of online purchase more convenient and saves time, money on consumers.
o Auction Bots: agents here are created to participate in online auctions for goods. According to Chavez and Maes (1996), there is a system called Kasbah where users create agents to sell or buy good according to their desires. Agents track the sell and purchase that is going on and on a certain time they consider selling or buying.
• Agents for Human-Computer Interfaces: users work with program interfaces by deciding the action to be taken by clicking icons, buttons etc. the goal here is to build an interface that can work with the user to achieve a task together. The idea of agent interfaces is to change the function of interface from servant to the user into assistant that cooperates with the user. Also this helps the user in working with new interfaces easily before getting used to them.
• Agents for Virtual Environments: agents here are used in film, entertainment and cinema industry. The key here is to provide real emotions to agents so that they won’t appear as featureless characters. Agents are provided with the ability to understand human behaviour. By this, agents can act and react in a way that looks like real and more convincing to the audience. This is widely seen in different movies.
• Agents for Social Simulation: agents are used as experimental tools in social sciences. The aim here is to use agents to represent individual people to simulate the behaviour of human societies. Different projects were done in the area of social simulation that aimed to use agents as human societies and see what kind of reaction will happen in different circumstances. This was used in measuring the effects of new policies on human societies. As an example a model of agents representing people was done to measure the reaction of putting the policy of asking people to consume less water in drought times.
• Agents for Industrial Systems Management: deployment of multi-agent systems in several industrial domains. In this case the system is enabled to plan the action it takes and communicate with other agents. Agents here include both domain component which shows the domain functionality, and wrapper component which shows the agent functionality.
• Agents for Spacecraft Control: it provides real time advices to astronauts in the event of malfunction. It was very hard to use agent here since the system works in real time environment.
• Agents for Air-Traffic Control: agents are used to assist air-traffic controller in managing the flow of aircraft at an airport by monitoring previous estimates.
6. The Drawbacks and Limitations of Agent Solutions
Wooldridge (2002) discussed a number of drawbacks that individuals and users might encounter while experiencing an agent-based system. The basic three drawbacks are:
• Users have to tell every customization service what they want to know. This becomes a tedious job after a couple of time.
• Users need sometimes to fill a questionnaire or provide some personal information. This information will be used by the service provider to sell ads. This is a kind of personal information theft unless the user is aware that his information will be used.
• Users need to remember endless list of ID’s for each service. Since there is a big number of services, sometimes its easy to forget your ID easily
On the other hand Jennings and Wooldridge (1998) stated that agent-based systems provided a number of solutions to different applications and domains. Nevertheless, those systems should not be overrated and used for any application. Using agent-based system will cause a number of problems if it was used regularly with no big need for it. The problems can be summarized as:
• No overall system controller: this means that agent-based systems must not be used in domains that have constraints to be maintained or in real-time applications.
• No global perspective: different decisions are taken by the local state of the agent. This leads to incomplete global knowledge.
• Trust and delegation: agents work on behalf of users. So at the first a sense of trust should be build between users and agents and this might take a long period of time. During the building trust period, the agent must know its limitations and must not exceed its authority.
The industry of agent system is facing the lack of standards for inter-agent communication language. Currently agents are limited by communication difficulties despite the development of a standard language called Knowledge Query and Manipulation Language (KQML), Seipp (2001).
The future now is to develop standard communication languages that enable efficient communication means between agents.
7. Agent Communication Languages (ACL)
Agent communication languages provide the agents with the mean of interaction and knowledge exchange, the main two languages used for communication between agents are as following according to Alonso (2002):
• Knowledge Query and Manipulation Language (KQML): KQML can carry any representation language. It includes three layers: content, communication and message. The content layer stores the actual content in the programs own representation language. The communication layer encodes some features of the message such as sender and recipient ID and unique ID associated with the communication. The message layer encodes the message that one application would send to the other. The basic design feature in KQML is to produce a language that is able to support different and multiple interesting agent architectures. One of the important things KQML introduced is an agent that performs a lot of communication services such as forwarding messages to named services.
• Foundation for Intelligent Physical Agents (FIPA) ACL: the FIPA specification consists of message type and description of attitude for the sender and receiver. It also includes protocols and the action requested. FIPA is similar to KQML in separating the outer language from inner language. The outer language is the meaning of the message, while the inner language describes beliefs and desires. In conclusion, the two ACL’s are the widely used communication languages, but till this moment there is a lack of universal agreement on semantics foundation, Hon (1999).
8. Current Research
According to RMIT (2005), a lot of research is taking place now to improve and find new ways to develop software systems such as:
• The reason behind the lack of widespread agent technology is the lack of appropriate software engineering methodology. The research aims to support design processes in agent systems.
• Since the World Wide Web (WWW) includes loads of information. A range of agents are hosted by the web to serve users in information search. The research will be looking at the technical issues of adding new agents, and how the new agents can be located by other agents and work together with them.
Those two are a sample of a large number of research and projects done now days to improve agent technology and add new applications that can be supported by this technology.
9. Conclusion
The emergence of the World Wide Web (WWW) provided people around the globe with the facility of information access; however the amount of information caused a lot of disturbance and difficulty while browsing.
AI experts developed software called agents. Agents were used to help people to find the accurate information they look for. Since knowledge and improvement never stop, agents were developed to cover different application whether they are related to the web or not. The different applications that agents are used in are: Agents for Business Process Management, Agents for Distributed Sensing, Agents for Information Retrieval and Management, Agents for E-commerce, Agents for Human-Computer Interfaces, Agents for Virtual Environments, Agents for Social Simulation, Agents for Industrial Systems Management, Agents for Spacecraft Control and Agents for Air-Traffic Control .
The way agents operate depends basically on the environment they are located in. agents are provided with a knowledge base and interface engine. Learning agents include learning engine that enable them to learn from users or other agents
Multi-agent systems are systems that include more than one agent, and those agents interact between each other to achieve a shared tasks or each agent can have a separate task. Different protocols arrange their interaction that is called as negotiations. Agents use Agent Communication Language (ACL) such as Knowledge Query Manipulation Language (KQML) or Foundation for Physical Intelligent Agents (FIPA).
Agent based systems can not be used in any application due to the difficulty of imposing the characteristics of agents on some applications that are limited by different constraints.
Nowadays a lot of projects and research are taking place to improve and develop new features to agent systems to make the best and efficient use of this technology that is day by day spreading all over the world.
Word Count 3,503
References
1. Aiii (2000), Multi-Agent System: a Subtopic of Agent. [Online]. Available from:
[Accessed 8th May 2005].
2. Alonso, Eduardo. (2002), Agent communication slides, City University, London. [Online]. Available from:
http://www.soi.city.ac.uk/~eduardo/ [Accessed 13th May 2005].
3. Beer, Martin et al (1999), Negotiation in Multi-Agent System. [Online]. Available from:
http://www.ecs.soton.ac.uk/~mml/papers/ker99-2.pdf [Accessed 16th May 2005].
4. Borgman, C. L. (1986), “Why are online catalogs hard to use?,” in Journal of the American Society for Information Science, 37(6): 387-400.
5. Chavez, A. and Maes, P. (1996) Kasbah: an agent marketplace for buying and selling goods. In proceedings of the 1st International Conference on the Practical Application of Intelligent Agents and Multi-Agent Technology (PAAM-96), London, UK, pp. 75-90.
6. D’inverno, Mark and Luck, Michael. (2001), Understanding Agent Systems, New York: Springer, c2001.
7. Durfee, E.H. et al. (1997), The agent architecture of the University of Michigan digital library. IEEE proceedings on Software Engineering, 144(1), 61-71.
8. Etzioni, O. (1996) Moving up the information food chain: deploying softbots on the World Wide Web. In Proceedings of the 13th National Conference on Artificial Intelligence (AAAI-96), Portland, OR, pp. 4-8.
9. Etzioni, O. and Weld, D.S. (1995) Intelligent agents on the Internet: fact, fiction and forecast. IEEE Expert, 10(4), 44-49.
10. Ferber, Jacques. (1999), Multi-Agent Systems: an Introduction to Distributed Artificial Intelligence, Harlow: Addison-Wesley, 1999.
11. Hon, Haw. (1999), Agent Communication Language, [Online]. Available from:
[Accessed 11th May 2005].
12. Jennings, Nick. (1998), Negotiating Agents for Business Process Management. [Online]. Available from:
http://www.ecs.soton.ac.uk/~nrj/download-files/adept.pdf [Accessed 11th May 2005].
13. Jennings, N.R and Wooldridge, M. (1998), Applications of Intelligent Agents [Online]. Available from:
http://agents.umbc.edu/introduction/jennings98.pdf [Accessed 5th May 2005].
14. Klusch, Matthias (Ed.) (1999), Intelligent Information Agents: Agent-Based Information Discovery and Management on the Internet, Berlin: Springer, 1999.
15. Maes, P. (1994a) Agents that reduce work and information overload. Communications of the ACM, 37(7), 31-40.
16. Muller, Jorg. et al. (2003), Multi-Agent System and Applications |||: 3rd international central and eastern European conference on multi-agent systems, CEEMAS 2003 Prague, Czech Republic, June 2003, Proceedings, Germany: Springer-Verlag Berlin Heidelberg 2003.
17. RMIT (2005), School of Computer Science and Information Technology. RMIT University. [Online]. Available from:
http://www.cs.rmit.edu.au/agents/#dest [Accessed 6th May 2005].
18. Russell, Stuart and Norvig, Peter. (2003), Artificial Intelligence: a modern approach, 2nd edition, Upper Saddle River, N.J.: Prentice Hall, 2003.
19. Sdsu (2002), San Diego State University. [Online]. Available from:
http://map.sdsu.edu/geoagent/agent_intro.htm [Accessed 15th May 2005].
20. Seipp, Anna. (2001), Intelligent Agents, [Online]. Available from:
http://www.slais.ubc.ca/courses/libr500/01-02-wt1/www/A_Seipp/index.htm [Accessed 9th May 2005].
21. Tecuci, Gheorghe. (1998), Building Intelligent Agents: an apprenticeship multistrategy, tool and case studies, London; San Diego, Calif, Academic press, c1998.
22. Wooldridge, Michael. (2002), an Introduction to Multiagent Systems, Chichester: Wiley, 2002.
