an inventory of currently offered functionality in the information society & a prediction of (near-)future developments

Chia sẻ: Ledung Ledung | Ngày: | Loại File: DOC | Số trang:96

0
66
lượt xem
3
download

an inventory of currently offered functionality in the information society & a prediction of (near-)future developments

Mô tả tài liệu
  Download Vui lòng tải xuống để xem tài liệu đầy đủ

Big changes are taking place in the area of information supply and demand. The first big change, which took place quite a while ago, is related to the form information is available in. In the past, paper was the most frequently used media for information, and it still is very popular right now. However, more and more information is available through electronic media.

Chủ đề:
Lưu

Nội dung Text: an inventory of currently offered functionality in the information society & a prediction of (near-)future developments

  1. Intelligent Software Agents on the and further into the information Internet: age, any information-based an inventory of currently offered organisation which does not functionality in the information invest in agent technology may society & a prediction of be committing commercial (near-)future developments hara-kiri." Hyacinth S. Nwana in [NWAN96] by Björn Hermans __________ __________ "[...] Agents are here to stay, not least because of their diversity, their wide range of applicability and the broad spectrum of companies investing in Tilburg University, Tilburg, The them. As we move further Netherlands, the 9th of July 1996 http://www.hermans.org/agen ts Table of Contents 1 Preamble.........................................................................................................................................3 1.1 Abstract...............................................................................................................................3 1.2 Introduction.........................................................................................................................3 1.2.1 Problems regarding the demand for information................................................4 1.2.2 Possible solutions: Search Engines and Agents..................................................6 1.2.3 Agents as building blocks for a new Internet structure......................................8 1.2.4 Thesis Constraints.................................................................................................9 1.3 Two statements...................................................................................................................9 1.4 Structure of the thesis.........................................................................................................9 2 Intelligent Software Agents Theory.........................................................................................12 2.1 Introduction.......................................................................................................................12 2.2 Definition..........................................................................................................................14 2.2.1 The weak notion of the concept "agent"............................................................15 2.2.2 The strong(er) notion of the concept "agent"....................................................15 2.2.3 "Agency" and "Intelligence"...............................................................................17 2.3 The User's "definition" of agents.....................................................................................17 2.4 Summary...........................................................................................................................18 3 Intelligent Software Agents in Practise...................................................................................19 3.1 Applications of Intelligent Agents..................................................................................19 3.2 Examples of agent applications and entire agent systems.............................................21 3.2.1 Two examples of agent applications..................................................................22 3.2.1.1 Open Sesame! 22 3.2.1.2 Hoover 23 3.2.2 Two examples of entire agent systems..............................................................23 3.2.2.1 The Internet SoftBot 23 3.2.2.2 The Info Agent 25 3.3 Summary...........................................................................................................................27 4 The Three Layer Model.............................................................................................................29 4.1 Introduction.......................................................................................................................29 4.2 Definition..........................................................................................................................30 Intelligent Software Agents on the Internet 1
  2. 4.3 The functions of the middle layer....................................................................................32 4.3.1 Middle layer (agent) functions...........................................................................32 4.3.2 An example of a future middle layer query.......................................................36 4.4 Computer and human Intermediaries..............................................................................37 4.4.1 Introduction.........................................................................................................37 4.4.2 Intermediary/Broker Issues.................................................................................39 4.4.3 Human versus Electronic Intermediaries...........................................................40 4.5 An example of a middle layer application: Matchmaking.............................................40 4.6 Summary...........................................................................................................................43 5 Past and Current Agent Trends & Developments.................................................................45 5.1 Introduction.......................................................................................................................45 5.2 Computers and the agent-technique................................................................................45 5.3 The User............................................................................................................................46 5.4 The Suppliers & the Developers......................................................................................47 5.5 The Government...............................................................................................................49 5.6 The Internet & the World Wide Web..............................................................................50 5.7 Summary...........................................................................................................................54 6 Future and Near-Future Agent Trends & Developments.....................................................55 6.1 Introduction.......................................................................................................................55 6.2 The Agent-technique........................................................................................................56 6.2.1 General remarks...................................................................................................56 6.2.2 Chronological overview of expected/predicted developments.........................60 6.2.2.1 The short term: basic agent-based applications 60 6.2.2.2 The medium term: further elaboration and enhancements 61 6.2.2.3 The long term: agents grow to maturity 62 6.3 The User............................................................................................................................62 6.3.1 General remarks...................................................................................................62 6.3.1.1 Ease of Use 63 6.3.1.2 Available applications 65 6.3.2 Chronological overview of expected/predicted developments.........................66 6.3.2.1 The short term: first agent encounters 66 6.3.2.2 The medium term: increased user confidence and agent usage 66 6.3.2.3 The long term: further agent confidence and task delegation? 67 6.4 The Suppliers & the Developers......................................................................................67 6.4.1 Who will be developing agents, and how will they be offered?......................67 6.4.2 What kinds of agents will be offered?...............................................................69 6.4.3 Why/with what reasons will agents be developed and/or offered?..................70 6.5 The Government...............................................................................................................71 6.6 The Internet & the World Wide Web..............................................................................75 6.7 Summary...........................................................................................................................78 7 Concluding remarks, statement reviews and acknowledgements.......................................79 7.1 Concluding remarks..........................................................................................................79 7.2 Statement conclusions......................................................................................................80 7.2.1 The claim.............................................................................................................80 7.2.2 The prediction......................................................................................................83 7.3 Acknowledgements...........................................................................................................83 8 Used information sources..........................................................................................................84 8.1 Literature...........................................................................................................................84 8.2 Information sources on the Internet.................................................................................85 9 Appendices...................................................................................................................................89 Appendix 1 - A list of World Wide Web Search Engines....................................................89 Appendix 2 - General, introductory information about the Internet....................................93 Introduction...................................................................................................................93 Internet Services offered..............................................................................................93 Appendix 3 - Internet Growth Figures..................................................................................96 Intelligent Software Agents on the Internet 2
  3. Preamble 1 Preamble and the ease with which it can be obtained. Expectations are that these 1.1 Abstract developments will carry on into the future. Software agents are a rapidly A third important change is related to developing area of research. the supply and demand of However, to many it is unclear what information. Until recently the market agents are and what they can (and for information was driven by supply, maybe cannot) do. In the first part, and it was fuelled by a relatively this thesis will provide an overview small group of suppliers that were of these, and many other agent- easily identifiable. At this moment related theoretical and practical this situation is changing into a aspects. Besides that, a model is market of a very large scale where it presented which will enhance and is becoming increasingly difficult to extend agents' abilities, but will also get a clear picture of all the suppliers. improve the way the Internet can be used to obtain or offer information All these changes have an enormous and services on it. The second part is impact on the information market. all about trends and developments. One of the most important changes is On the basis of past and present the shift from it being supply-driven developments of the most important, to it becoming demand-driven. The relevant and involved parties and number of suppliers has become so factors, future trends and high (and this number will get even developments are extrapolated and higher in the future) that the question predicted. who is supplying the information has become less important: demand for information is becoming the most 1.2 Introduction important aspect of the information chain. "We are drowning in What's more, information is playing information but starved an increasingly important role in our of knowledge" lives, as we are moving towards an information society1 . Information has John Naisbitt of Megatrends become an instrument, a tool that can be used to solve many problems. Big changes are taking place in the area of information supply and demand. The first big change, which took place quite a while ago, is related to the form information is available in. In the past, paper was the most frequently used media for information, and it still is very popular right now. However, more and more information is available through electronic media. Other aspects of information that 1"Information society" or "Information Age" are both terms that are very often used have changed rapidly in the last few nowadays. The terms are used to denote the years are the amount that it is period following the "Post-Industrial Age" available in, the number of sources we are living in right now.
  4. Preamble 1.2.1 Problems regarding the can or cannot be found in the demand for information database, and - if available - where it Meeting information demand has can be found. become easier on one hand, but has also become more complicated and On the Internet (but not just there 3) difficult on the other. Because of the this strategy fails completely, the emergence of information sources reasons for this being: such as the world-wide computer • The dynamic nature of the network called the Internet2 (the Internet itself: there is no central source of information this thesis will supervision on the growth and focus on primarily) everyone - in development of Internet. Anybody principle - can have access to a sheer who wants to use it and/or offer inexhaustible pool of information. information or services on it, is Typically, one would expect that free to do so. This has created a because of this satisfying information situation where it has become very demand has become easier. hard to get a clear picture of the The sheer endlessness of the size of the Internet, let alone to information available through the make an estimation of the amount Internet, which at first glance looks of information that is available on like its major strength, is at the same or through it; time one of its major weaknesses. The • The dynamic nature of the amounts of information that are at information on Internet: your disposal are too vast: information that cannot be found information that is being sought is today, may become available (probably) available somewhere, but tomorrow. And the reverse often only parts of it can be retrieved, happens too: information that was or sometimes nothing can be found at available, may suddenly disappear all. To put it more figuratively: the without further notice, for instance number of needles that can be found because an Internet service has has increased, but so has the size of stopped its activities, or because the haystack they are hidden in. The information has been moved to a inquirers for information are being different, unknown location; confronted with an information overkill. The current, conventional search methods do not seem to be able to tackle these problems. These methods are based on the principle that it is known which information is available 3 Articles in professional magazines indicate (and which one is not) and where that these problems are not appearing on the exactly it can be found. To make this Internet only: large companies that own possible, large information systems databases with gigabytes of corporate such as databases are supplied with information stored in them (so-called data warehouses), are faced with similar (large) indexes to provide the user problems. Many managers cannot be sure with this information. With the aid of anymore which information is, and which is such an index one can, at all times, not stored in these databases. Combining the look up whether certain information stored data to extract valuable information from it (for instance, by discovering 2General, introductory information about the interesting patterns in it) is becoming a task Internet and its services can be found in that can no longer be carried out by humans appendix two. alone.
  5. Preamble • The information and information services on the Internet are very heterogeneous: information on the Internet is being offered in many different kinds of formats and in many different ways. This makes it very difficult to search for information automatically, because every information format and every type of information service requires a different approach.
  6. Preamble 1.2.2 Possible solutions: Search Engines and Agents "In the future, it [agents] There are several ways to deal with is going to be the only the problems that have just been way to search the described. Most of the current Internet, because no solutions are of a strong ad hoc matter how much better nature. By means of programs that the Internet may be roam the Internet (with flashy names organised, it can't keep like spider, worm or searchbot) meta- pace with the growth in information 4 is being gathered about information..." everything that is available on it. The gathered information, characterised Bob Johnson, analyst at Dataquest by a number of keywords (references) Inc. and perhaps some supplementary information, is then put into a large database. Anyone who is searching Using agents when looking for for some kind of information on the information has certain advantages Internet can then try to localise compared to current methods, such as relevant information by giving one or using a search engine: more query terms (keywords) to such a search engine5. Search Engine feature: Improveme Agents can Although search engines are a 1 An information search is done, based Agents are c valuable service at this moment, they . on one or more keywords given by a information also have several disadvantages user. This presupposes that the user is instance bec (which will become even more capable of formulating the right set of thesaurus) e apparent in the future). keywords to retrieve the wanted related term A totally different solution for the information. Querying with the wrong, concepts. A problem as described in section 1.2.1, too many, or too little keywords will to fine-tune is the use of so-called Intelligent cause many irrelevant information (on the basi Software Agents. An agent is ('noise') to be retrieved or will not user informa (usually) a software program that retrieve (very) relevant information as supports a user with the it does not contain these exact accomplishment of some task or keywords; activity. 6 4 For example, the gathering programs that collect information for the Lycos search engine, create document abstracts which consist of the document's title, headings and subheadings, the 100 most weighty words, the first 20 lines, its size in bytes and the number of words. 5 In appendix 1, a list of Internet search engines is given, to give an idea just what kind of search engines are currently being offered. 6 There are many different kinds of software agents, ranging from Interface agents to Retrieval agents. This thesis will be mainly about agents that are used for information also apply to the other kinds of agents. tasks (such as offering, finding or editing all However (for briefness' sake), we will only kinds of information). Many things that are concern ourselves with information agents in said about agents in this thesis do, however, this thesis.
  7. Preamble 2 Information mapping is done by . gathering (meta-)information about information and documents that are available on the Internet. This is a very time-consuming method that causes a lot of data traffic, it lacks efficiency (there are a lot of parties that use this method of gathering information, but they usually do not co-operate with others which means that they are reinventing the wheel many times), and it does not account very well for the dynamic nature of the Internet and the information that can be found on it; 3 The search for information is often . limited to a few Internet services, such as the WWW. Finding information that is offered through other services (e.g. a 'Telnet-able'7 database), often means the user is left to his or her own devices; 4 Search engines cannot always be . reached: the server that a service resides on may be 'down', or it may be too busy on the Internet to get a connection. Regular users of the service will then have to switch to some other search engine, which probably requires a different way to be operated and may offer different services; 7See appendix 2 for more information about Telnet.
  8. Preamble 5 Search engines are domain-independent . in the way they treat gathered 1.2.3 Agents as building blocks information and in the way they enable for a new Internet structure users to search in it 8 . Terms in gathered The Internet keeps on growing, and documents are lifted out of their judging by reports in the media the context, and are stored as a mere list of Internet will keep on growing. The individual keywords. A term like big threat this poses is that the "information broker" is most likely Internet will get too big and too stored as the two separate terms diverse for humans to comprehend, "information" and "broker" in the meta- let alone to be able to work on it information of the document that properly. And very soon even contains them. Someone searching for (conventional) software programs documents about an "information will not be able to get a good grip on broker" will therefore also get it. documents where the words More and more scientists, but also "information" and "broker" are used, members of the business community, but only as separate terms (e.g. as in are saying that a new structure should "an introductory information text about be drawn up for the Internet which stock brokers"); will make it more easily and 6 The information on Internet is very conveniently to use, and which will . dynamic: quite often search engines make it possible to abstract from the refer to information that has moved to various techniques that are hidden another, unknown location, or has under its surface. A kind of disappeared. Search engines do not abstraction comparable to the way in learn from these searches9 , and they do which higher programming languages not adjust themselves to their users. relieve programmers of the need to Moreover, a user cannot receive deal with the low-level hardware of a information updates upon one or more computer (such as registers and topics, i.e. perform certain searches devices). automatically at regular intervals. Searching information this way, Because the thinking process with becomes a very time-consuming regard to these developments has activity. started only recently, there is no clear sight yet on a generally accepted The precise characteristics of agents standard. However, an idea is are treated in more detail in chapter emerging that looks very promising: a two. Chapter three will focus on the three layer structure10 . There are practical possibilities of agents. quite a number of parties which, although sometimes implicitly, are 8 Users do not directly search the studying and working on this concept. information on the Internet itself, but the The main idea of this three layer meta-information that has been gathered model is to divide the structure of the about it. The result of such a search, is not the meta-information itself, but pointers to the document(s) it belongs to. 9 If a document is retrieved which turns out to be no longer available, the search engine does not learn anything of this happening: it will still be retrieved in future sessions. A search engine also does not store query 10 As opposed to the more or less two layer results, so the same query will be repeated structure of the current Internet (one layer over and over again, starting from scratch. with users and another layer with suppliers).
  9. Preamble Internet into three layers11 or when they are employed in concepts: the (future) three layer 1. Users; structure of the Internet." 2. Suppliers; and 3. Intermediaries. The function and added-value of the The second part of the thesis will be added middle layer, and the role(s) about current, near-future and future agents play in this matter, are agent developments. Questions such explained in chapter four. as "how will agents be used in the near future?", "who will be offering agents (and why)?", and "which 1.2.4 Thesis Constraints problems/needs can be expected?" There are agents in many shapes and will be addressed here. sizes. As can be concluded from the Because of the nature of this part, the preceding text, this thesis will deal second statement is a prediction: mainly with one special type of intelligent software agents, namely "Agents will be a highly those that are used in the process of necessary tool in the information supply and demand. process of information When, in the forthcoming sections of supply and demand. this thesis, the term "agent" is used, However, agents will not usually these "information agents" are yet be able to replace meant. However, many things that are skilled human information said, apply to the other types of intermediaries. In the agents as well. forthcoming years their role will be that of a valuable personal 1.3 Two statements assistant that can support all kinds of people with This thesis consists of two parts. For their information each of these two parts a separate activities." statement will be formulated. The first part of the thesis is an inventory of agent theory, agents in 1.4 Structure of the thesis practise, and the three layer model. In the next chapter, the theoretical The claim for this part is: side of agents will be more deeply looked at: what are agents, what "Intelligent Software makes them different from other Agents make up a techniques and what is the promising solution for the functionality they (will) have to current (threat of an) offer? information overkill on the After having looked at agents in Internet. theory in chapter two, chapter three The functionality of agents will give an idea of the kind of can be maximally utilised practical applications that agents and the agent technique are already being The term "layers" is perhaps a bit used in. 11 misleading as it suggests a hierarchy that is In chapter four a three layer model not there: all three layers are of equal importance. Thinking of the layers in terms will be sketched, where the agent of concepts or entities may make things technique is combined with the more clearer. functionality offered by the various
  10. Preamble Internet services. Together they can expectations. The parties and factors be used to come to a Internet that that have, are, or will be influencing offers more functionality, is more developments are looked at in more surveyable, and has a cleaner logical detail. structure than the current (two-layer) set-up. In chapter seven, the thesis will be concluded with concluding remarks The second part of this thesis, and a look at the accuracy of the two comprised by the chapters five and statements of section 1.3. six, is entirely about past, present and future developments, prediction and
  11. PART ONE - Theoretical and Practical Aspects of Agents and the Prospects of Agents in a Three Layer Model
  12. Intelligent Software Agents Theory 2 Intelligent Software Agents Agent producers that make unjust use of Theory the term agent to designate their product, cause users to draw the 2.1 Introduction conclusion that agent technology as a whole has not much to offer. That is - Intelligent software agents are a popular obviously - a worrying development: research object these days in such fields as psychology, sociology and computer "In order to survive for the science. Agents are most intensely agent, there must be studied in the discipline of Artificial something that really Intelligence (AI) 12 . Strangely enough, it distinguishes agents from seems like the question what exactly an other programs, otherwise agent is, has only very recently been agents will fail. addressed seriously. Researchers, the public and companies will no "It is in our best interests, longer accept things that as pioneers of this are called agent and the technology, to stratify the market for agents will be technology in such a way very small or even not that it is readily exist." marketable to consumers. If we utterly confuse Wijnand van de Calseyde on the consumers about what Software Agents Mailing List agent technology is (as is the case today) then we'll On the other hand, the description of have a hard time fully agent capabilities should not be too developing the market rose-coloured either. potential." Not everybody is that thrilled about J. Williams on the Software Agents agents. Especially from the field of Mailing List13 computer science, a point of criticism often heard about agents is that they are Because of the fact that currently the not a new technique really, and that term "agent" is used by many parties in anything that can be done with agents many different ways, it has become "can just as well be done in C".14 difficult for users to make a good According to these critics, agents are estimation of what the possibilities of nothing but the latest hype. the agent technology are. At this The main points of criticism can be moment, there is every appearance that summarised as follows: there are more definitions than there are • Mainstream AI research (expert working examples of systems that could systems, neural networks) is not as be called agent-based. successful as many people had hoped 12 For general information about AI, see this WWW page: http://wombat.doc.ic.ac.uk/?AI 14 C is a structured programming language 13 This is a discussion list (using e-mail as the developed by Dennis Ritchie at Bell means of communication) about the subject of Laboratories in 1972. C is a compiled language Software Agents. The list is used and read by that contains a small set of built-in functions both users as well as developers of such agents. that are machine dependent. The rest of the C For more information see functions are machine independent and are http://www.ee.mcgill.ca/~belmarc/agent_faq.ht contained in libraries that can be accessed from ml. C programs.
  13. Intelligent Software Agents Theory and the new paradigm of agents is the where independent agents are way to escape; working together to accomplish all • Everything that has the label "agent" kinds of tasks) is used; sells (this also counts in research). • Agents make it possible to eradicate Like the words 'plus', 'super' and the differences between the different 'turbo', the term 'agent' sounds very kinds of networks (WAN, LAN17 , attractive, even when most people do Internet) and to make the borders not know the exact meaning of 'plus', between them 'disappear'. Some 'super', 'turbo' or 'agent'. Agents are researchers like to take this one step nothing more but old wine in new further by playing with the notion of bottles; agents that supersede AI. 18 • Because of the fact that in most cases The response of (particularly) these current software agents have neither a researchers to the pronouncement very sophisticated nor a very quoted earlier, that what agents can do complicated architecture, some "can just as well be done in C", can be wonder what qualifies them as summarised in the following points: "intelligent". 15 ∗ It does not matter what the underlying technique of agents is. Particularly by researchers in the field Whether that is a C program, a Perl of AI, these points of criticism are script, or a LISP program: what it all refuted with the following arguments: boils down to is what the agent is and • What distinguishes multi-agent is not capable of doing. Or to be architectures from other architectures more precise: whether the agent is is that they provide acceptable capable of displaying intelligent solutions to certain problems at an behaviour. And whether the basis for affordable price. These are the kind that behaviour is a C program, or of problems that cannot not be solved whatever other programming with available resources in reasonable language or technique, does not time with monolithic knowledge really matter; based systems.16 ∗ It does not always apply that An example of this can be found in the everything that can be done by field of integrated decision making, multiple co-operative agents may where systems are built where a "just as well be done in C" (not even single final diagnose is based on the diagnoses of individual worker 17 LAN stands for Local Area Network (as agents. opposed to a WAN: a Wide Area Network). A Moreover, there are some problems in LAN is a group of computers and other devices the field of AI that cannot be solved dispersed over a relatively limited area and connected by a communications link that satisfactorily unless a multi-agent enables any device to interact with any other on architecture (i.e. an architecture the network. LANs commonly include microcomputers and shared (often expensive) 15 Unfortunately that question opens up the old resources such as laser printers and large hard AI can-of-worms about definitions of disks. Most (modern) LANs can support a wide intelligence. E.g., does an intelligent entity variety of computers and other devices. necessarily have to possess emotions, self- 18 These researchers see a paradigm shift from awareness, etcetera, or is it sufficient that it those who build intelligent systems and performs tasks for which we currently do not consequently grapple with problems of possess algorithmic solutions? knowledge representation and acquisition, to 16 The 'opposite' can be said as well: in many those who build distributed, not particularly, cases the individual agents of a system aren't intelligent systems, and hope that intelligence that intelligent at all, but the combination and will emerge in some sort of Gestalt fashion. co-operation of them leads to the intelligence The knowledge acquisition problem gets solved and smartness of an agent system. by being declared to be a 'non-problem'.
  14. Intelligent Software Agents Theory in the object oriented variant of that however, a possible informal definition programming language). There are of an intelligent software agent was several tasks and problems for which given: there is scientific proof that they cannot be accomplished or solved by "A piece of software which one single program or person. These performs a given task using kind of problems call for a information gleaned from its distribution of the task or problem environment to act in a suitable over multiple entities (i.e. a multi- manner so as to complete the agent architecture) because this will task successfully. The lead to a solution in a much shorter software should be able to time, and quite often to a solution of adapt itself based on changes a higher quality because it is the occurring in its environment, so result of a subtle combination of the that a change in circumstances partial results of each individual will still yield the intended entity. result." (with thanks to G.W. Lecky-Thompson The 'pros' and 'cons' with regards to for this definition) agents as they are mentioned here, are by no means complete, and should be seen as merely an illustration of the Instead of the formal definition, a list of general discussions about agents. What general characteristics of agents will be it does show is why it is necessary (in given. Together these characteristics several respects) to have a definition of give a global impression of what an the concept "intelligent software agent" agent "is".19 that is as clear and as precise as possible. It also shows that there is The first group of characteristics, which probably a long way to go before we will be presented in section 2.2.1, are arrive at such a definition - if we can connected to the weak notion of the come to such a definition at all. concept "agent". The fact that an agent should possess most, if not all of these characteristics, is something that most 2.2 Definition scientists have agreed upon at this moment. This is not the case, however, with the "An agent is a software thing second group of characteristics, which that knows how to do things are connected to the strong notion of the that you could probably do concept "agent". The characteristics that yourself if you had the time." are presented in section 2.2.2 are not things that go without saying for Ted Selker of the IBM Almaden everybody. Research Centre (quote taken from What "intelligence" is, and what the [JANC95]) related term "agency" means, is explained in section 2.2.3. In this section we will not come to a rock-solid formal definition of the concept "agent". Given the multiplicity of roles agents can play, this is quite 19 See [WOOL95] for a more elaborated impossible and even very impractical. overview of the theoretical and practical aspects On the Software Agents Mailing List, of agents.
  15. Intelligent Software Agents Theory 2.2.1 The weak notion of the are able to exhibit goal-directed concept "agent" behaviour by taking the initiative; Perhaps the most general way in which • temporal continuity: agents are the term agent is used, is to denote a continuously running processes hardware or (more usually) software- (either running active in the based computer system that enjoys the foreground or sleeping/passive in the following properties: background), not once-only • autonomy: agents operate without the computations or scripts that map a direct intervention of humans or single input to a single output and others, and have some kind of control then terminate; over their actions and internal state; 20 • goal orientedness: an agent is capable • social ability: agents interact with of handling complex, high-level other agents and (possibly) humans tasks. The decision how such a task is via some kind of agent best split up in smaller sub-tasks, and communication language;21 in which order and in which way • reactivity: agents perceive their these sub-tasks should be best environment (which may be the performed, should be made by the physical world, a user via a graphical agent itself. user interface, a collection of other agents, the Internet, or perhaps all of Thus, a simple way of conceptualising these combined), and respond in a an agent is as a kind of UNIX-like timely fashion to changes that occur software process24 , that exhibits the in it 22 . This may entail that an agent properties listed above. A clear example spends most of its time in a kind of of an agent that meets the weak notion sleep state23 from which it will awake of an agent is the so-called softbot if certain changes in its environment (‘software robot’). This is an agent that (like the arrival of new e-mail) give is active in a software environment (for rise to it; instance the previously mentioned • proactivity: agents do not simply act UNIX operating system). in response to their environment, they 20 See: Casterfranchi, C (1995). Guarantees for 2.2.2 The strong(er) notion of the autonomy in cognitive agent architecture. In concept "agent" Woolridge, M. and Jennings, N. R., ed., For some researchers - particularly those Intelligent Agents: Theories, Architectures, and working in the field of AI - the term Languages (LNAI Volume 890), page 56-70. agent has a stronger and more specific Springer-Verlag: Heidelberg, Germany. 21 See: Genesereth, M. R. and Ketchpel, S. P. meaning than that sketched out in the previous section. These researchers (1994). Software Agents. Communications of the ACM, 37(7): page 48-53. generally mean an agent to be a 22 Note that the kind of reactivity that is computer system that, in addition to displayed by agents, is beyond that of so-called having the properties as they were (UNIX) daemons. Daemons are system previously identified, is either processes that continuously monitor system conceptualised or implemented using resources and activities, and become active concepts that are more usually applied once certain conditions (e.g. thresholds) are met. As opposed to agents, daemons react in a very straight-forward way, and they do not get 24UNIX is an operating system that is mostly better in reacting to certain conditions. used on large computer systems and 23 Analogous to the "sleep" state in a UNIX workstations. The concept of process is the system (see the next footnote): a process that basic idea behind UNIX (a program running has no further tasks to be done, or has to wait under UNIX consists of one or more for another process to finish, goes into a sleep independent processes which usually are state until another process wakes it up again. operating in parallel).
  16. Intelligent Software Agents Theory to humans. For example, it is quite • rationality: is (crudely) the common in AI to characterise an agent assumption that an agent will act in using mentalistic notions, such as order to achieve its goals and will not knowledge, belief, intention, and act in such a way as to prevent its obligation 25 . Some AI researchers have goals being achieved - at least insofar gone further, and considered emotional as its beliefs permit; 32 agents26 . • adaptivity: an agent should be able to Another way of giving agents human- adjust itself to the habits, working like attributes is to represent them methods and preferences of its user; visually by using techniques such as a • collaboration: an agent should not cartoon-like graphical icon or an unthinkingly accept (and execute) animated face27 . Research into this instructions, but should take into matter 28 has shown that, although agents account that the human user makes are pieces of software code, people like mistakes (e.g. give an order that to deal with them as if they were dealing contains conflicting goals), omits with other people (regardless of the type important information and/or of agent interface that is being used). provides ambiguous information. For instance, an agent should check Agents that fit the stronger notion of things by asking questions to the agent usually have one or more of the user, or use a built-up user model to following characteristics29 : solve problems like these. An agent • mobility: the ability of an agent to should even be allowed to refuse to move around an electronic network; 30 execute certain tasks, because (for • benevolence: is the assumption that instance) they would put an agents do not have conflicting goals, unacceptable high load on the and that every agent will therefore network resources or because it always try to do what is asked of it; 31 would cause damage to other users.33 25 See: Shoham, Y. Agent-oriented Although no single agent possesses all programming. Artificial Intelligence, 60(1): these abilities, there are several page 51-92, 1993. 26 See, for instance, Bates, J. prototype agents that posses quite a lot The role of emotion in believable agents. Communications of them (see section 3.2.2 for some of the ACM, 37(7): page 122-125, 1994. examples). At this moment no consensus 27 See: Maes, P. Agents that reduce work and has yet been reached about the relative information overload. Communications of the importance (weight) of each of these ACM, 37(7): page 31-40, 1994. characteristics in the agent as a whole. 28 See, for instance, Norman, D. How Might What most scientists have come to a People Interact with Agents. In consensus about, is that it are these Communications of the ACM, 1994 issue, Juli 1994. kinds of characteristics that distinguish 29 This list is far from complete. There are many agents from ordinary programs. other characteristics of agents that could have been added to this list. The characteristics that are mentioned here are there for illustrative purposes and should not be interpreted as an ultimate enumeration. 32See: Galliers, J. R. A Theoretical Framework 30 See: White, J. E. Telescript technology: The for Computer Models of Cooperative Dialogue, foundation for the electronic marketplace. Acknowledging Multi-Agent Conflict. PhD White paper, General Magic Inc., 1994. thesis, page 49-54, Open University, Great 31 See: Rosenschein, J. S. and Genesereth, M. R. Britain, 1994. Deals among rational agents. In Proceedings of 33 See: Eichmann, D. Ethical Web Agents. the Ninth International Joint Conference on Proceedings of the Second International World- Artificial Intelligence (IJCAI-85), page 91-99, Wide Web Conference. Chicago, United States, Los Angeles, United States, 1994. October 1994.
  17. Intelligent Software Agents Theory 2.2.3 "Agency" and "Intelligence" resources available to the The degree of autonomy and authority agent. Such a system might, vested in the agent, is called its agency. like a human assistant, It can be measured at least qualitatively discover new relationships, by the nature of the interaction between connections, or concepts the agent and other entities in the independently from the human system in which it operates. user, and exploit these in At a minimum, an agent must run a- anticipating and satisfying user synchronously. The degree of agency is needs." enhanced if an agent represents a user in some way. This is one of the key values of agents. A more advanced agent can 2.3 The User's "definition" of interact with other entities such as data, agents applications, or services. Further advanced agents collaborate and "User knowledge, rather than negotiate with other agents. product capability, is the principal determinant of agent- What exactly makes an agent enabled application usage "intelligent" is something that is hard to today. define. It has been the subject of many [...] User need is the principal discussions in the field of Artificial consideration in Intelligence, and a clear answer has yet developing/executing business to be found. strategies for agent-enabled Yet, a workable definition of what products." makes an agent intelligent is given in [IBM95]: from [JANC95] "Intelligence is the degree of Just like in the oncoming information reasoning and learned society, the success and development of behaviour: the agent's ability to agents and the agent technique are accept the user's statement of driven by users really, instead of by goals and carry out the task producers or researchers. 34 So, when delegated to it. considering just exactly what an agent At a minimum, there can be is, and which aspects of it are very some statement of important and which are less important, preferences, perhaps in the the ever important user factor should not form of rules, with an inference be overlooked. engine or some other Users will not start to use agents reasoning mechanism to act on because of their benevolence, these preferences. proactivity or adaptivity, but because Higher levels of intelligence they like the way agents help and include a user model or some support them in doing all kinds of tasks; other form of understanding soon users will use all sorts of and reasoning about what a convenient (i.e. "intelligent) user wants done, and planning applications, without them realising they the means to achieve this goal. are using agents by doing so. Further out on the intelligence scale are systems that learn and adapt to their environment, both in terms of the user's 34Users will not play that much of a very active objectives, and in terms of the steering-role, but user acceptance and adoption will be the ultimate test of agent's success.
  18. Intelligent Software Agents Theory As was pointed out at the beginning of quote taken from [JANC95] this chapter, there is one good reason why a fairly concise definition of an More about (professional) user's views agent that can meet with general on agents, will follow in chapter five approval, should be drawn up as soon as and six. possible: clarity towards the user. By all means it should be prevented that "agent" becomes a vague, meaningless 2.4 Summary and empty term, in the way a term such Today, agents are a popular research as "multi-media" has lost its meaning in object in many scientific fields. An the course of time. Agents will be exact definition and exact set of perceived as nothing but the latest characteristics (and their relative marketing hype: weight) are yet to be stated and chosen. "Just take your old program, Ultimately, users of agents and agent- and add an agent to the end of enabled programs will be the principal your product name. Voila! You determinant of how agents will look, have an Object Agent, Test what they will be, and what things they Agent. [...]" should and should not be able to do.
  19. Intelligent Software Agents in Practise 3 Intelligent Software Agents this does not mean that there will be no in Practise or little interesting developments and opportunities in the area of smaller, 3.1 Applications of Intelligent more low-level agent applications.) Agents In chapter four a model will be presented which supports this trend The current applications of agents are of towards more complex, integrated a rather experimental and ad hoc nature. systems. In this model basic agents can Besides universities and research centres easily be combined to create complex a considerable number of companies, structures which are able to perform like IBM and Microsoft, are doing high-level tasks for users, suppliers and research in the area of agents. To make intermediaries. The interface to this sure their research projects will receive system (i.e. model) is through a single further financing, many researchers & agent which delegates sub-tasks and developers of such companies (but this queries to other agents. is also applicable on other parties, even non-commercial ones) are nowadays In [IBM95] eight application areas are focusing on rather basic agent identified where now (or in the near- applications, as these lead to future) agent technology is (or will be) demonstrable results within a definite used. time. These areas are: Examples of this kind of agent 1. Systems and Network applications are: Management: • Agents who partially or fully handle Systems and network management someone's e-mail; is one of the earliest application • Agents who filter and/or search areas to be enhanced using through (Usenet) news articles intelligent agent technology. The looking for information that may be movement to client/server interesting for a user; computing has intensified the • Agents that make arrangements for complexity of systems being gatherings such as a meeting, for managed, especially in the area of instance by means of lists provided LANs, and as network centric by the persons attending or based on computing becomes more the information (appointments) in the prevalent, this complexity further electronic agenda of every single escalates. Users in this area participant. (primarily operators and system administrators) need greatly The current trend in agent developments simplified management, in the face is to develop modest, low-level of rising complexity. applications. Yet, more advanced and Agent architectures have existed in complicated applications are more and the systems and network more being developed as well. management area for some time, At this moment research is being done but these agents are generally into separate agents, such as mail "fixed function" rather than agents, news agents and search agents. intelligent agents. However, These are the first step towards more intelligent agents can be used to integrated applications, where these enhance systems management single, basic agents are used as the software. For example, they can building blocks. Expectations are that help filter and take automatic this will become the trend in the next actions at a higher level of two or three years to come. (Note that abstraction, and can even be used
  20. Intelligent Software Agents in Practise to detect and react to patterns in mail by organisational function system behaviour. Further, they rather than by person. can be used to manage large Intelligent agents can facilitate all configurations dynamically; these functions by allowing mail handling rules to be specified 2. Mobile Access / Management: ahead of time, and letting As computing becomes more intelligent agents operate on behalf pervasive and network centric of the user according to those computing shifts the focus from rules. Usually it is also possible the desktop to the network, users (or at least it will be) to have want to be more mobile. Not only agents deduce these rules by do they want to access network observing a user's behaviour and resources from any location, they trying to find patterns in it; want to access those resources despite bandwidth limitations 35 of 4. Information Access and mobile technology such as Management: wireless communication, and Information access and despite network volatility. management is an area of great Intelligent agents which (in this activity, given the rise in case) reside in the network rather popularity of the Internet and the than on the users' personal explosion of data available to computers, can address these needs users. It is the application area that by persistently carrying out user this thesis will mainly focus on. requests despite network Here, intelligent agents are helping disturbances. In addition, agents users not only with search and can process data at its source and filtering, but also with ship only compressed answers to categorisation, prioritisation, the user, rather than overwhelming selective dissemination, the network with large amounts of annotation, and (collaborative) unprocessed data; sharing of information and documents; 3. Mail and Messaging: Messaging software (such a 5. Collaboration: software for e-mail) has existed Collaboration is a fast-growing for some time, and is also an area area in which users work together where intelligent agent function is on shared documents, using currently being used. Users today personal video-conferencing, or want the ability to automatically sharing additional resources prioritise and organise their e- through the network. One common mail, and in the future, they would denominator is shared resources; like to do even more another is teamwork. Both of these automatically, such as addressing are driven and supported by the move to network centric computing. 35 Bandwidth is - in technical terms - the Not only do users in this area need measure of information-carrying capability of a communication medium (such as optical fibre). an infrastructure that will allow An Internet service such as the World Wide robust, scaleable sharing of data Web, which makes use of graphical (and and computing resources, they also sometimes even audio or video) data, needs need other functions to help them considerable amounts of bandwidth, whereas an actually build and manage Internet service such as e-mail needs only very small amounts.
Đồng bộ tài khoản