4 Filtering Agents 10 6 2 example essay topic

Index Abstract 2 1. Introduction: Problems being solved 2 1.1 What is an intelligent agent? 3 2.12 layer Business model 4 2.1 Vision 4 2.2 Business model 4 2.3 Business strategies 4 2.4 Business processes and rules 5 2.4. 1 Retrieving user's requirements 5 2.4. 2 Before starting to search 6 2.4. 3 Filtering Information 6 2.4.

4 Packaging Information 6 2.5 Organizational design. 7 3. Benefits of the solutions 7 4. Contrast with traditional solutions.

7 5. My opinion 8 6.1 Types of software agents 8 6. Annex 8 6.2 Another useful classification 9 6.2. 1 Interface agents 9 6.2. 10 Multiagents 11 6.2. 2 System agents 9 6.2.

3 Advisory agents 10 6.2. 4 Filtering agents 10 6.2. 5 Retrieval agents 10 6.2. 6 Navigation agents 10 6.2. 7 Monitoring agents 10 6.2. 8 Recommender agents 10 6.2.

9 Profiling agents 10 7. References 12 Intelligent agents: tools for students and teachers Abstract Through this paper it is analyzed the possibility of developing intelligent agents that help students and teachers have a specialized assistant that searches and retrieves information from the on their behalf. These agents are also analyzed to be developed and offered by a company named Netuniversitaria. com as its new product / service. 1. Introduction: Problem being solved New technologies have strongly impacted on the education segment.

Through the Internet, students and teachers have access to any kind of information for free. This unbounded availability of contents and knowledge has no precedents in human history. Educational institutions, scientists, investigators, government, companies are feeding the Internet with the information they produce. "With world knowledge doubling just about every three to five years and the Internet being on the verge of becoming the primary source for information of all kinds, one question becomes ever more important: How can we properly handle this highly powerful online medium by finding the information and services we need and avoiding the most pressing problem of the Information Age: Information Overload". Hundreds of million pages available on the are actually being consulted for more than 400 million users all over the world. All this data is stored in web servers distributed worldwide and can be located by using search engines.

Search engines are useful tools for finding needed information but they have some limitations. They don't look for information in "all" web servers of the Internet, they just consult a few of them. That means that the consulting user is not having access to the whole network. Queries not always end in successful results for many reasons. How to evaluate millions of documents? How to find what is needed if the engine is showing that there doesn't exist information about a certain subject?

These are some of the unanswered questions that every Internet user frequently experiences when looking for specific information. This limitation also affects teachers and students in their intention to take advantage of such a huge variety of contents that can be used for investigating, studying, teaching and sharing knowledge". As the amount of content available over the Internet multiplies, there will be an increasing need for teachers to find places where they can conduct "one stop shopping" to obtain educational materials. We cannot expect teachers to spend the time "surfing" the net and weeding through countless material to find the subset that will be useful to them. We do not believe teachers will spend the time surfing the information superhighway looking for educational content. ".

What is needed is a tool that facilitates the searching and retrieving process. A tool that can be oriented in its searching abilities by receiving the user's conditions and preferences, capable of continuing its searching off-line and acting on behalf of the user. By having a tool with these features, students and teachers could reduce the time used for searching issues and concentrate on learning. That is the task of intelligent agents.

1.1 What is an intelligent agent? Some definitions, "An intelligent agent is a software that assists people and acts on their behalf. Intelligent agents work by allowing people to delegate work that they could have done, to the software agent. Agents can, just as assistants can, automate repetitive tasks, remember things you forgot, intelligently summarize complex data, learn from you, and even make recommendations to you". (Gilbert, 1998).

"An agent is a computer program that simulates a human relationship by doing something that another person could otherwise do for you. It combines artificial intelligence and system development techniques. Agents can learn from and about the user". (Amy Babor Phd, Florida State University and Ali Jafar i, Phd Indiana University Purdue university) Pattie Maes, founder and director of the Software Agent Group from the MIT Media Laboratory defines agents as follows, "A computational system which: o Is long lived o Has goals, sensors and effectors o Decides autonomously which actions to take in the current situation to maximize progress toward its (time-varying) goals. A software agent is a particular kind of agent that inhabits computers and networks assisting users with computer-based tasks".

Though the definitions may vary, there are a number of aspects that most people agree upon a program must posses to qualify as an agent. An agent should be: o Autonomous. The agent must have control over its own actions and be able to work and launch actions independent of the user or other actors. o Reactive. The agents can detect changes in its environment and react to these in a timely matter by answering to events and initiate actions. o Communicative. The agent is able to interact and communicate with users and other agents. o Goal-driven.

Agents have a purpose and act in accordance with that purpose until it is fulfilled. Other aspects often also mentioned are dynamic (agents should be able to operate depending on time and space), adaptive (agents learn and change their behavior based on previous experiences), temporal continuous (agents should not be started or stopped for explicit tasks but rather be a continuously running process), and / or mobile (agents should be able to move themselves from one machine to another, and across different architectures and platforms). 2. Twelve layer Business model 2.1 Vision Netuniversitaria. com offers to Latin American universities brand new technology (specially PCs) with Internet connection for free. Universities have to install on the Pcs offered by Netuniversitaria its proprietary browser and set up its web site as a default page (web).

Students and teachers have to register in order to access to Netuniverstaria's site and, if wanted, access to the rest of Internet. Today, Netuniversitaria has signed contracts with the most prestigious Latin American universities which gather together almost a million students. Having such a big community of students and teachers permanently using Netuniversitaria's web site and considering that one of the company's objectives is to offer the best educational services on the web, Net has to provide its users with tools that help them find the information they need in the shortest time and the best accuracy. And these kind of services can be offered by intelligent agents. 2.2 Business model Netuniversitaria has a captive market. Students and teachers from all universities that have a contractual relationship with Netuniversitaria are daily visitors of its web site and its browser.

This Latin American university network created by Netuniversitaria is the perfect target to offer Net's new product / service : agents that help users to find any information they need. Developing such a solution would not only increase Net product's usage among users but also would incorporate all those universities that don't need new hardware (PCs) but surely need a tool like this. As Netuniversitaria does today, these tools would be offered for free and universities would be asked to install net's site in their own equipment. Users must be registered in order to have complete access to all the features of the offered tool, .

It is important to indicate that Net's site has a classified directory with updated educational contents, useful applications like a calendar, on line files storage, calculator, e-mail, community chat, etc... It is developing an e-commerce platform for offering to users the possibility of buying everything from everywhere (from next corner's pizza to a souvenir from Greece). 2.3 Business strategies Promotion of these tools should be done by inserting banners in the most famous and visited portals of the Net. At the same time, this effort should be accompanied by a big off-line campaign (through radio, Tv, streets, newspapers, magazines). Asking for Net user's opinion (teachers and students) about some details of the developing tools would be a very good alternative for knowing users' needs, customs and troubles on their way to finding data. Despite the intervention of some teachers and students, the participation of a specialized librarian will be indispensable.

The developing agent is expected to do the same tasks as a librarian does. Both act as information mediators for the end user: both negotiate information spaces and retrieve information relevant to a particular user or goal. The development of software agents can be informed by looking at how human information agents do their work. For a complete effectiveness, Net should plan a kind of intelligent agent basic structure that can be reusable for generating new other agents, each of them specialized in searching for different subjects.

To achieve this goal, it a searching core method has to be defined, composed by generic steps that every agent has to respect. What is going to differ are the consulted databases. 2.4 Business processes and rules As any project, all steps of this work must be included in a plan where every detail must be taken into account, especially those related to financial matters and funding. What this new tool tends to be is an intelligent virtual librarian that acts as a real person, orienting users on how to get to the needed information. Both librarians and software agents do the same tasks. Work is delegated to them by a user; both apply some expertise to the user's problem or need; and both work in the background, completing tasks for the user's information needs.

Trying to imitate a librarian's behavior and the way he faces a search will help us to define the logical searching method that is going to be used by our agents. To achieve this it is necessary to define a set of carefully studied steps and rules to be followed. An important approach to be considered is presented next. 2.4. 1 Retrieving user's requirements Librarians use what is called a reference interview to learn about and assess a client's needs. This interview is often an interactive process with much feedback between librarian and client as they work together to refine the information needed.

This process can be quite delicate: many clients are unclear about their needs and a skilled librarian can help them specify and communicate their need. The librarian works for relevance feedback during the interview, asking the client "Is this what you mean?" and "More like this?" The librarian must learn about the user's context: knowledge of client's situation, history, and preferences. All these data provide context for the inquiry. ii Our agent will have to make many questions at first in order to understand what the user is looking for. After having all user's answers, the agent should begin the process of defining the information requested. This can be done by asking the user more details about his request such as dates, kind of information (articles, papers, books, common texts, web sites, etc. ), geographic details, language, etc... By asking all these questions, the agent can define the context of the information asked and establish a profile of the consumer.

Just as librarians do. "The librarian contributes to the client's activity and to do so effectively, he creates a representation of the activity that guides and focuses the search. This representation goes beyond understanding the client's task, simply conceived, to a broader contextual sketch of the client, including the client's preferences, constraints, and environment". (Nardi and O'Day, op. cit., p. 75). 2.4. 2 Before starting the search The librarian may sketch the search steps on paper, choosing databases, search terms, and operators or he may just form this representation mentally. He considers the cost of the search, the quality of the databases at his disposal, the customer's profile, and the ultimate use of the information requested.

In the case of our developing agent, things are quite different. All gathered information serves as conditions that orientate the agent to choose some searching strategy. After the first retrieval of information, the agent will have decided three main aspects: a. the general subject of the search b. the initial searching criteria to be used for consulting Netuniversitaria's database c. other agents with whom it will share information After a few questions, the agent will have a clear idea about what the user is looking for. Although our agent will obtain requested information from, Netuniversitaria should have its own databases permanently updated (by a simple retrieving agent) and evaluated, classified and catalogued by education and content specialists. This database will be a good starting point for the agent to search until it learns about the user it is working for.

With Intelligent Agents there is no use of queries in the traditional sense. After they have been created by the user, they have to be trained before they can perform their first task (search). This training involves the input of plain English text describing the topic to be investigated. Once this has been done, the agent will independently navigate the Web, rating the material it comes across according to its likely relevance. 2.4. 3 Filtering Information Information filtering is a process in which a filtering agent reads every document in an information stream and compares it to a set of interest profiles. If the document matches a profile, the filter sends it to the appropriate user's inbox or stores it somewhere for the user.

The rest of the documents are filtered out. (Williams, 1996 p. 174). Agents should consider all user's given conditions for filtering information. At this point, the agent can still make contact with the user in order to ask him for giving more guidance and clues for adjusting the searching process. 2.4. 4 Packaging Information Librarians do more than connect people to raw information: they use their expertise to help clients make sense of information. Librarians usually arrange search results and other information products into customized sets for their clients.

Generally, librarians give its customers a report with all the obtained results. Agents tend to do the same and much more. Our agent will communicate with the user in different ways. The agent may display suggestions, or perform direct-manipulation actions on objects in the displayed interface, based on input implicitly collected from the user.

Other kinds of interface agents may criticize the user's behavior, or augment the user's direct-manipulation actions with extra computed information that the user may find helpful. As seen in the above paragraphs, it is very important to work with on line searching specialized professionals (librarians or information brokers) so that a group of agents that imitate human behavior for making choices can be created. 2.5 Organizational design. Netuniversitaria will have to create a new project group for investigating and developing these tools. This group is expected to be integrated by knowledge engineers, content specialists, librarians, advanced programmers, and specialists. 3.

Benefits of the solutions By developing these tools Netuniversitaria will once more gather the attention of the higher education community. Students and teachers will finally have access to a set of tools that assist them on the searching information process in such a way that they will find what they need in very short times and with a high rate of precision. Their task will just be to study and learn. Signing a contract with Netuniversitaria will not only be synonymous of having numerous benefits by accessing to new hardware and web-based applications, but will also be the possibility of accessing a set of tools that will make the process of finding information a fast, efficient and effective process. Netuniversitaria will be offering an innovative and useful solution that will confirm its leader position as the most important educational site for universities of Latin America. 4.

Traditional solutions for finding information are based on the use of search engines and directories. Searching engines use robots (also known as crawlers, bots or spiders) that crawl the looking for web site pages. They collect information about new, old (deleted) and modified web pages. All this information is stored in a database, the one that is consulted with the user's searching criteria. Each robot has its own method for indexing that database. That's why it is very common to use the same word as searching criteria in two different engines and obtain different results.

This also happens because not all robots crawl the same web servers over the so the information they gather is not the same. Retrieved documents used to be so many (some times millions) that they are impossible to evaluate nor even to read their titles! Some times engines can't find any document related to the given criteria. What could have happened? Doesn't that information really exist or does another searching clue have to be used? Maybe this is the explanation and the user has to think of another combination of words and probably connect those words with Boolean connectors (AND, OR, NOT), signs (+, -, ") or other special connectors (IS NEAR).

These features are used by experienced users and the ones that don't know about the existence of them keep surfing to the drift. Knowing how to take advantage of advanced search features doesn't guarantee the finding of the information needed. Directories show all the information classified by humans. Directories are useful when the user is looking for easy-categorized information. For example, if the user looks for universities he will select the directory's Education link. Things are not that simple when looking for contents that are not that easy to classify, such as "disorders in the teenager's behavior at school".

Which links of a directory should be consulted? Education, Humanities, Social Sciences or Health? With intelligent agents things are quite different. The user can count on the assistance of an expert that works for him even when the PC is turned off. Agents do all the time consuming work following the user's directions. They also learn about the user, his customs and preferences so that it can make recommendations and suggestions.

They collaborate with other agents in order to access many information sources and interchange data. By using these new tools users save time and the teaching-learning process becomes more effective and efficient. 5. My opinion In order to prepare this paper I have used search engines. I have spent three working days finding updated and exact information. I have used two of the most used engines such as web and web I also used meta-engines such as: web web and web Besides these five, I had to think of other engines because I couldn't find what I was looking for (web and web).

Many of the documents retrieved, many titles that seemed to be what I needed but when I opened the documents most of them were about anything else but what I wanted. Many other times, documents contained the right information but they weren't updated! I lost precious time reading, filtering and evaluating all materials I got but just found a very few updated documents. And, at last, I am still not sure of having covered all existing materials about what I am interested in. Academic, scientific, educational and qualified information available on the Internet and we, the users, can't take advantage of it because we don't have the right tools. It's like being inside the biggest library of the world with lights turned off and we have a small lantern in our hands.

If I could have had access to a tool like the agent we have been talking about through this paper, I could have done my work in less than half the time I spent and could have reached to more consistent conclusions because I could have had a more realistic, complete and comparative vision of what I was looking for. I think intelligent agents are the future of the Internet. Every company that expect to lead in the near future its market segment needs to work with them. 6. Annex 6.1 Types of software agents Dimensions along they would differ: a. Nature of task performed User agents: they assist a user, know his interests, preferences and habits.

It may act on user's behalf. For example, personal news editor, personal e-shopper, personal web guide. Service agents: they perform more general tasks in the background. For instance, web indexing, info retrieval, phone network, load balancing. b. Nature and source of intelligence: User programmed: the person provides rules and criteria directly. These are the simplest agents, not very smart (that relies on user's programmer skills).

Artificial intelligence engineered: created by traditional knowledge based AI techniques. They are complex, programmed by a knowledge engineer. Learning agents: They program themselves: Patterns in user's actions and among users are detected and exploded. They observe and imitate and can also learn from other agents. They have a medium complexity and are smart in key areas (where user concentrates). c. Mobility / Location: Mobile and Wireless agents.

6.2 Another useful classification Agents can be separated in user oriented agents and collaborative agents. A user oriented agent focuses on the individual, making sure that the user's needs are fulfilled. Every agents is then in itself enough, and the owner of the agent benefits from the start. Collaborative agents are of course also useful to the individuals but the real benefit comes from the cumulative knowledge of several cooperative agents.

A critical mass of agent users must be established before anyone may notice any benefits. As with the definition of an agent, there are a number of way to classify agents. Let's see one of the classifications: 6.2. 1 Interface agents Interface agents are used to decrease the complexity of the more and more sophisticated and overloaded information systems available. They may add speech and natural language understanding to otherwise dumb interfaces, or add presentation ability to systems. Henry Lieberman defines an "interface agent" as a program that can also affect the objects in a direct manipulation interface, but without explicit instruction from the user.

The interface agent reads input that the user presents to the interface, and it can make changes to the objects the user sees on the screen, though not necessarily one-to-one with user actions. The agent may observe many user inputs, over a long period of time, before deciding to take a single action, or a single user input may launch a series of actions on the part of the agent, again, possibly over an extended period of time. An interface agent could be considered to be a "robot" whose sensors and effectors are the input and output capabilities of the interface, and for that reason are sometimes also referred to as "soft bots" 6.2. 2 System agents System agents run as integrated parts of operating systems or network protocol devices.

They help managing complex distributed computing environments by doing hardware inventory, interpreting network events, managing backup and storage devices, and performing virus detection. These agents do not primarily work with end-user information. 6.2. 3 Advisory agents Advisory agents are used in (complex) help or diagnostics systems 6.2. 4 Filtering agents Filtering agents are used to reduce information overload by removing unwanted data, i.e. data that does not match the user's profile, from the input stream.

Simple versions are built-in in many email clients and Agentware and InfoMagnet provide a more general kind of server-based filtering capabilities. 6.2. 5 Retrieval agents Retrieval agents search and retrieve information and serves as information brokers or documents managers. Many products claim to be retrieval agents, including the client-based AT 1, BullsEye, Go-Get-It, Got-It, Surfboat, and Web Compass, and the server-based Agentware and InfoMagnet. 6.2. 6 Navigation agents Navigation agents are used to navigate through external and internal networks, remembering short-cuts, pre-load caching information, automatically book marking interesting sites. IBM's Web Browser Intelligence (WBI - pronounced Webby) is an example. 6.2. 7 Monitoring agents Monitoring agents provide the user with information when particular events occur, such as information being updated, moved, or erased.

Enterprise Minder does this but is no agent. WBI from IBM has this as a feature, as do BullsEye and Smart Bookmarks. 6.2. 8 Recommender agents Recommender agents are usually collaborative; they need many profiles to be available before an accurate recommendation can be made. Examples are Agentware, Firefly, and GroupLens, which are all server-based. Learn Sesame is an exception that is user-oriented and bases its conclusion on the user's previous behavior. 6.2. 9 Profiling agents Profiling agents are used to build dynamic sites with information and recommendations tailored to match each visitor's individual taste and need.

The main purpose is to build customer loyalty and profitable one-to-one relationships. Available examples are Agentware, Firefly, and GroupLens on the server side. Learn Sesame and IBM's Knowledge Utility also do this but on a user-oriented level. 6.2. 10 Multiagents Much of the current research related to intelligent agents has focused on the capabilities and structure of individual agents. However, to solve complex problems, these agents must work cooperatively with other agents in a heterogeneous environment. This is the domain of multi-agent systems. 7.

Bibliography

John Leddo. "A Virtual Schoolhouse". Dr. John Leddo is President of Research Development Corporation, 2875 Tower view Rd., Suite A 4, Herndon, VA 20171.
E-mail: . 1997.
web Dick Stern mark. "Intelligent Software Agents. A study of what is available". Viktoria Institute. Knowledge Management Group. February 2000.
Laura Z ick, The Work of Information Mediators: A Comparison of Librarians and Intelligent Software Agents, May 2000.
web Tobias Rade mann, "Information Unlimited: Employing Internet Resources in Education", , Ruhr University, Bochum, Germany, 1996-97.
web Gerhard Fischer, Andreas C. Lemke, Thomas Masta glio and Anders I. March, Critics: An Emerging Approach to Knowledge Based Human Computer Interaction, International Journal of Man-Machine Studies, p. 695-721, 35 (5), Nov 1991.
Pattie Maes is an Associate Professor at MIT's Media Laboratory, where she founded and directs the Software Agents Group, and is principal investigator of the e-markets Special Interest Group. "General tutorial on software agents", 1997, prepared for the CHI 97 conference.
Other consulted material: "Agents That Reduce Work and Information Overload", 1994.
Henry Lieberman, "Autonomous Interface Agents", ACM Conference on Human-Computer Interface [CHI-97], Atlanta, March 1997.