|
1.0前言
我经常会借助雅虎网站来指导我的研究任务。雅虎是一个著名的门户网站，需要处理大量的服务需求。它的服务包括最新的全球和本地新闻，娱乐，体育资讯，天气预报，市场以及电子邮件服务。虽然大部份的服务和产品都是免费提供的，消费者也可以支付一定的协议费用来获取更多的内容和服务。大部份产品支持全球20多种语言。 雅虎拥有大量的电子交易市场，例如像雅虎购物，雅虎房产以及雅虎旅游这样的服务。这些服务允许用户收集重要的信息，同时也允许他们进行在线的交易和采购。雅虎的收入主要来源于它的市场服务，而其中最大的部分则是搜索广告。广告商通过购买检索关键词，把他们的广告展现在搜索结果中。除此之外，还有一大笔收入是来源于展示性广告和上下文关联广告。 雅虎网站同时也拥有一个综合性很强的搜索引擎。用户可以通过这一搜索引擎进行信息，图片，视频，新闻，指南以及购物等网站搜索。雅虎的搜索引擎拥有很强大的且独一无二的数据库。这个数据库可以缓存网页副本并连接到雅虎目录。不仅如此，雅虎搜索引擎还可以全布尔搜索以及支持短语通配符。 1.0 INTRODUCTION The organization that I will be conducting my research and assignment will be based on Yahoo.com. Yahoo is a very famous portal that handles massive traffic and requests for many of their services. Among the services provided are latest world and local news, entertainment, sports information, weather forecasting, marketplace and also email services. While most of the services and products can be accessed for free, customers can get premium contents or services for an agreed fee. The majority of the products offered are available worldwide in more than 20 languages. Yahoo has a huge number of electronic commerce transactions from services such as Yahoo Shopping, Yahoo Real Estate and Yahoo Travel. This service allows users to collect important information and also allows them to make commercial transactions and purchases online. Yahoo’s revenue model is mostly contributed by their marketing services. The largest chunk of it comes from search advertising, where advertisers bid for search terms to display their advertisements on the search results. Apart from that, a big sum of revenue comes from the display and contextual advertising. Yahoo’s website also has a very comprehensive search engine. It allows users to search the web for information, images and videos, news, directories and shopping. Yahoo’s search engine has a large, unique search engine database. It also includes cached copies of pages and links to the Yahoo directory. Not only that, Yahoo’s search engine supports full Boolean searching and Wild Card Word in phrase. Looking at the how enormous the website is and the wide range of features and services offered, it is for sure that Yahoo is assisted by many types of intelligent agents to carry out the numerous tasks and procedures. The search engine alone needs a few potent intelligent agents to make sure it returns expected results. Further on in this assignment, I will identify the various types of intelligent agents that is implemented in Yahoo’s website and discuss them in depth. 2.0用户代理系统—— INTERFACE AGENT 2.1 Overview One of the intelligent agents identified in Yahoo is the interface agent. In short, an interface agent can be defined as a character performed by a computer that interacts with the user, in a socially engaging approach. Interface agents also help users in carrying out computer-based tasks. The important characteristics of an interface agent include responsiveness, competence and accessibility. Interface agents can also act autonomously to carry out processes without precise directions from the user. Apart from that, they can also collaborate with other types of intelligent agents. Not only do they collaborate with other intelligent agents, there is also collaboration between the interface agent and the user in a same work environment. However, an interface agent may require specific agent communication language when cooperating with other agents but may not be the same when collaborating with a user. To be able to assist users, these interface agents must be presented with some form of knowledge of its domain. The intelligent agent is provided with some background knowledge and learns the suitable behavior either from the user or from other agents. Basically, interface agents offer help and assistance to users who are using an application or system. The interface agent observes and monitors the actions carried out by the user in the interface, learns short cuts, and recommends other methods of performing a task better. Interface agents usually learn to better assist its user in 4 ways (Maes, 1994). Firstly, it learns by observing and imitating the user. Secondly, it learns by receiving instructions from the user. Thirdly, from positive and negative feedback that the user provides, the interface agent learns. Lastly, it can also learn from its peers, which are other agents. Interface agents have been applied in various situations of application situations, in which the one measured here is software interfaces, for instance, the software that sits between the main application codes and the user. From the user’s point of view, this is the software system, so this component is very vital. Even a system which is developed very well will be useless if the user is unable to use the system due to poor interface. Hence, an interface agent is accountable for providing this interface. Normally, software interfaces will be reactive instead of being proactive. Consequently, the typical user will perform actions, send out commands or just interact with the interface which will pass this on straight ahead to the underlying application. Therefore, the interface is a straightforward matter, which will provide a handy means for accessing the aspects of the functionality of the system of interest to the user. Figure 1.1 Structure of an interface agent application 2.2 Application of Interface Agent in Yahoo 2.2.1 User Registration The application of interface agent that can be seen in Yahoo is in the page where new user’s sign up for an account under them. There are many types of online account registration. They can vary from one that only requires one’s email and choice of password to others that require one’s personal details, preferences and details on interests. Those that require users to fill in lots of information have to have some form of HCI usability to guide users to fill in the right kind of information. For instance, if someone enters alphabets into a field that requires only numbers, the form has to notify the user to change it. In Yahoo’s user registration page, the interface agent interacts with the user to make sure the user enters the proper type of information. For instance as seen in Figure 1.2, when gender is not selected, the interface agent displays an error message requiring that the user chooses a gender. Also, when the date “February, 30, 2010” is entered into the birthday field, the interface agent once again notifies the user with an error message, “Are you really from the future?”. Figure 1.2 Yahoo Registration (Personal Info) Figure 1.3 Yahoo Registration (ID) From Figure 1.3, we can see that the interface agent immediately informs the user that the ID must be between 4 and 32 characters long. Apart from that, there is another message below it to let the users know what they must and must not enter when choosing an ID. Figure 1.4 Yahoo Registration (ID 2) As seen in Figure 1.4, when an appropriate ID has been entered, another conflict arises and yet again, the interface agent notifies the user. This time, the problem is about the unavailability of the ID. So, the interface agent scans the database and comes out with a few suggestions of ID that has not been taken. Figure 1.5 Yahoo Registration (Password) From Figure 1.5, we can see that after choosing the available ID, the user is required to enter the choice of password. The interface agent then lets the user know how to make the password more secure and they are required to enter at least 6-32 characters with no spaces. Figure 1.6 Yahoo Registration (Password 2) In Figure 1.6, when the password which is “abc123” is entered, the password strength meter on the right of it went up to 3 bars. The interface agent here interacts with the user, letting them know the strength of the password as they enter it. 2.2.2 Music Recommendation Another application of interface agent in Yahoo is in the music section. In this section, there are thousands of songs from many various genres available for users to listen to. Users are also able to view information regarding the band, artist, composer or the pictures of the album cover. Instead of displaying all the music selections that they have, Yahoo’s music recommendation feature implements interface agents to learn the user’s preference in music so that the system could just provide the user with a small portion of the overall database of music which it thinks the user would like. However, the system also provides easy access to the whole database of music as the user’s preference may differ as time pass or the user may just choose to look at different types of music.#p#分页标题#e# The interface agents make better recommendations as the user’s actions increases. For instance, the agents can narrow down and identify the user’s preferences by observing the music that the user has listened to, the searches he/she has made and the music rated (so that the agent can know whether the user likes that particular music or not). From these actions, the user will know what type of music the user likes and proceed to recommend other songs to the user in the future. 3.0 信息代理系统——INFORMATION AGENT 3.1 Overview Information agents or also known as internet agents, have been implemented because tools which assist us in managing the ever expanding growth of information are steeply required. This trend will only continue to increase in terms of demand. The roles carried out by information agents include influencing, supervising or assembling information from various dispersed sources. Given the fact that the Internet has grown vastly beyond what it is originally created for, it is now almost impossible for a web user to find all of his or her information interests through surfing. Furthermore, it is extremely hard for typical users to know where and how to find the right information with overflowing rich knowledge sources on the Internet. This is where information agents come in handy. Information agents are a bit similar to interface agents in terms of how they perform and also the goals they aim to achieve. The difference here between information and interface agents is, information agents are recognized by what they do and interface agents on the other hand are recognized by what they are. A distinct characteristic of information agents is they can be static or mobile. Information agents have no customary approach to their operation. They could or could not learn and they can be either social or non-cooperative. In the case of information agents which are mobile, these agents are capable of navigating through the World Wide Web to collect information and report what they are able to gather to a home site. On the other hand, a static information agent generally uses various internet management tools such to congregate information. The information agent is possibly linked to some particular indexer, for instance, a Spider. Typically, an information agent which has been instructed to gather information on a particular matter will send out a variety of search requests to either one or even several URL search engines. When the information is gathered from its sources, it will be assembled and sent back to the user. 3.2 Application of Information Agents in Yahoo 3.2.1 Yahoo Search Engine The Yahoo Search Engine can be found on its main page and is a very competent search engine besides Google. The search engine implements information agent to help search for relevant information on a subject and bring back the information to accommodate the user’s request. When a user searches a particular topic or subject, the results returned are the most relevant web pages found by Yahoo in response to the user’s search terms. The web results obtained are generated and produced by the information agents from the billions of web pages discovered, crawled, and indexed. Figure 1.7 Yahoo Search From Figure 1.7, when “inti college” is entered into the search field, the information agents in Yahoo work to gather information about “inti college” from the World Wide Web and return it as web results to the user. From what can be seen from the results, INTI University College Malaysia which is the main branch is listed first followed by the Subang Jaya campus. Also, the information engine recommends the user to try a few other search terms to narrow down the results. 3.2.2 Car Finder In the Yahoo Autos section, the car finder section also implements the information agent to assist users in information filtering. With car finder, users can browse various cars available but in a narrowed down scope so they can view the cars based on a few criterions. The results displayed can be narrowed down according to criterion such as price, make, body style, year, driving and performance. The information agent takes the query that is submitted by the user which specifies the filtering and it converts the query and forwards it to the indexers to help search for the desired car and subsequently the information agent collates the results and returns them to the user. Figure 1.8 Car Finder Results In Figure 1.8, we can see the criterion selected in the first box, which is $35,000-$45,000 (price) + Sedan (body type) + Automatic (transmission). From that criterion, 24 results are returned. 3.2.3 Yahoo Shopping Yahoo Shopping helps users find products and compare different prices from different merchants who are selling the same item. The information agent is also present in this environment to help search through millions of products offered by third party merchants. The results returned by the information agents will be based on factors such as algorithmic ranking and relevance. Payment by participating merchants can also affect the search results in which the merchant will be able to obtain a higher search ranking. The information agents collate information gather from web pages which are crawled and indexed by the Yahoo! Shopping web crawl process. 4.0反应系统—— REACTIVE AGENT 4.1 Overview This agent works and acts in a stimulus-response behavior to the current state of the environment in which they are placed. A precise action can be programmed to be carried out for certain sensor information, which can be executed using simple if-then rules. What is to be achieved or the aimed results of the agent are wholly represented by rules and this makes it difficult to make certain the desired behavior. This is because every single situation or possibility must be considered by the programmer in advance. For instance, in a situation where a robot lawn-mower is responsible for cutting the grass in a garden (assuming the garden is rectangular in shape). The robot will only not cut the grass if there’s a rule like this: IF (area == no grass) THEN move_forward ENDIF In this situation, if the programmer fails to foresee all possible events, he may forget an additional rule which may cause the robot to keep cutting the grass even when at that particular area has no grass. It is relatively more expensive to develop these kinds of systems and monitor their behavior because reactive agents in more complex environments usually contain hundreds of rules. Reactive agents are reasonably simple and they cooperate with other agents in the most basic ways. However, complex patterns of behavior surface from these interactions when a group of these agents is view globally. There are 3 key main points which highlight reactive agents (Maes, 1991). The first one is emergent functionality. For example, the dynamics of the communication leads to surfacing complication. The second main point is task decomposition; A reactive agent is seen as a set of components which function autonomously and accountable for explicit tasks such as computations, sensing and motor controls. The third main point highlights the tendency of reactive agents to work on representations which are near to raw sensor information which makes it different from other types of agents which have plentiful of high-level symbolic representations. 4.2 Applications of Reactive Agents in Yahoo 4.2.1 Games The Yahoo Games is a section of Yahoo where hundreds of various games can be found. These games can range from simple to complex games and from single to multiplayer games. The application of reactive agents is apparent in the more complex games. As an example, the reactive agent is used in the Yahoo Pool game. This is a game where there are there are many angles and speed at where the ball can travel. The reactive agent calculates every input the player makes and computes where the end state of the balls should be. Besides that, the reactive agent can also determine whether a particular action of the player can make him/her score the ball. Another example of the usage of reactive agents is in the Chess game. Here, the player can play against the computer which is made out of reactive agents. These agents receive the input (player’s move) and calculate every possible good move to play against the player. There are hundreds of possible move that are already programmed into the knowledge based. The reactive agent also learns from the player’s move and will pinpoint the player’s strategy in attempt to defeat the player. These agents react to every different move of the player, meaning that each and every game will have different possible moves and outcomes. 4.2.2 Email Another area where the reactive agent is present is in the emailing system of Yahoo. Yahoo Mail provides users with free email accounts where they can compose, send, receive and manage email messages. The reactive agent is implemented here to help filter out spam emails so that they do not flood the inbox of the user who only intends to receive what he or she is supposed to receive. The reactive agent works by looking at the email’s routing information such as IP addresses and compare legitimate & standardize protocols of creation & transmission. Then, the reactive agent also scans the email. The first thing that is considered is the title. Non-compliant titles will straightaway cause the email to be marked as spam. Apart from that, the reactive agent also compares the sender with a list of known domain names and email addresses previously reported as spam senders. Subsequently, the reactive agent will search the body of the email for terms and phrases that are commonly used in spam and scam mails. The agent can also learn to classify spam mails better from the user input. The user can provide feedback by marking a mail that has passed through the spam filtering as spam. So, in the future, those similar types of mails that have been identified not relevant to the user will be filtered out by the reactive agent.#p#分页标题#e# 5.0 移动代理系统——MOBILE AGENT 5.1 Overview A mobile agent has the ability to travel from one host to another to accomplish its goals in a heterogeneous network environment. In other words, a mobile agent is capable of roaming Wide Area Networks such as the Internet, it is able to interact with foreign host and also able to collect data on the user's behalf and coming back home having performed the tasks set by the user. The mobile agent has the capabilities to transport itself and its state in which when it reaches the new host, it will be able to perform accordingly in the new environment. However, just depending on mobility itself is not a very good idea because mobility itself does not make it an agent. They are classified as agents because mobile agents are autonomous and they work and interact together. For instance, the position of several of its internal objects and methods are broadcasted, thanks to the cooperation between the mobile agents. In this process, an agent is participating in some form of information exchange without giving all of its information away. A good way to fully utilize mobile agents is to put them in unreliable networks to process data. In environments like this, the low-reliability network can be used to transfer agents, instead of blocks of data from one place to another. Thanks to mobile agents, many systems today are becoming more increasingly flexible. One can update his or her Windows operating system just by connecting to Microsoft’s web servers. Apart from that, one can also find out which updates is missed. Mobile agents offer us this degree of flexibility in many distributed applications. For example, we can obtain much greater level of flexibility because these agents are not limited just to the specific machines. They are able to move around enabling us to update whenever we want. When compared to client-server architecture, mobile agents offer a more standardized method in handling code and data in a distributed system. The client-server architecture is rather static and the systems are hard to reconfigure and update. The mobile agent system on the other hand, is naturally much more flexible. One machine can be a server and another machine can dynamically take over that role by simply having the application move to that other machine. 5.2 Application of Mobile Agents in Yahoo 5.2.1 Travel The usage of mobile agents can be seen in the travel section in Yahoo. Yahoo Travel is where users can research destination guides, get world travel guide recommendations, see photos, videos, trip plans and also book their vacation package. The mobile agents are responsible in this environment to help users find the cheapest or the most affordable deals for their holiday. Apart from that, the mobile agents would save the user a considerable amount of time because the mobile agent can find all the prices from several airlines’ website and display it, enabling the user to pick the most desirable one. For example, if a user wants to depart from Kuala Lumpur to Phuket, from 18th June 2009 to 23rd June 2009, the mobile agents would then search and gather all the relevant information from a number of airlines’ website available and find all the flights available for the date and destination specified. The user can then select from a list that displays all the flight information including the price. The mobile agents are useful here because users do not need to visit various airlines website to research on the pricing. 5.2.2 Email The protocol used by Yahoo’s email system is POP3 (Post Office Protocol 3). When sending an email to another email server such as Hotmail which is using the HTTP Protocol, the mobile agents used here will communicate using the SMTP (Simple Mail Transfer Protocol). In detail, when a user sends an email, the mobile agents first send it from the user’s computer to the mail server which from there, other mobile agents will send it over to the receiver’s email server. These mobile agents are able to travel to different email servers using different type of protocols. For example, a user with a Yahoo account which uses the POP3 protocol is able to send an email to another user with a Gmail account which uses the IMAP (Internet Message Access Protocol). The IMAP protocol is a standard protocol for accessing e-mail from your local server. The POP3 protocol on the other hand provides a simple, standardized way for users to access mailboxes and download messages to their computers. As for the HTTP protocol, it is not a protocol dedicated for email communications, but it can be used for accessing the mailbox. 6.0协作代理系统—— COLLABORATIVE AGENT 6.1 Overview To carry out tasks for their owners, collaborative agents put emphasis on autonomy and cooperation with other agents. Though they can learn, learning is not the main strength of their operation. Negotiation to reach commonly acceptable agreements on some issues is needed to have an organized collection of collaborative agents. The main characteristics of these agents include responsiveness, social ability, autonomy and proactiveness. Therefore, these agents have the ability to act reasonably and autonomously in environments. Collaborative agents are more likely to be static, generous, rational, straightforward and some combinations of these or even neither. As the number of computer communication networks increases, there is a huge step towards developing virtual societies. Collaboration between individuals in these virtual societies requires that communication links be established and used effectively. This is where collaborative agents come in. The goal of collaborative agents is to interconnect separately developed collaborative agents, consequently allowing the collection to function beyond the capabilities of any of its members. Implementing efficient ways of cooperation among agents is actually one of the central issues for Multi-Agent Systems Development. Collaborative agents are able to solve large problems with static distributed computing. Apart from that, various existing legacy systems can be interconnected and interoperated. However, there is still a ongoing issue on inter-agent coordination. Collaborative agent systems usually have agent mechanisms distributed throughout network nodes. 6.2 Application of Collaborative Agents in Yahoo 6.2.1 Yahoo Personals An example of where collaborative agents can be seen in Yahoo is in the Personals section. This is sort of like a matchmaking service or even for single people to meet. Users can even search singles from the opposite sex based on criterions such as location area, age range, ethnicity, height, etc. Users can also send messages to particular other users. It supports a distributed registry of user interests while preserving privacy. In such environment, collaborative agents work by keeping track of other agents it encounter and advances itself by asking for referrals to other agents in order to find other users that may match the specific interest of a given user. In this situation, the characteristics of collaborative agents can be seen. For instance, they work together (collaborate), they negotiate and cooperate while still able to work autonomously. In this service, there are task-specific collaborative agents which have the knowledge on how to carry out a command, knowledge on how to collect information for the task given, knowledge of other information and task agents it must work together with in order to meet the goals, protocols that enable synchronization with other agents and strategies for conflict solving and information union. The collaborative agents here have information of the databases it is connected to, in addition to other particulars such as the database's size, average time taken to answer a query and costs for query processing. Apart from that, the collaboration agents have the knowledge of accessing database, knowledge of resolving conflicts and data union strategies and protocols for working together with other related agents. These collaborative agents are also smart enough to cache the answers of frequently asked queries and can also encourage database regularities which are used by the collaborative agents during inter-agent communication. 7.0代理优势—— AGENT ADVANTAGES 7.1 Interface Agent Advantage With interface agents, end users and application developers are able to accomplish more even when they do less work. Apart from that, over time, interface agents can adapt to its user’s preferences and habits. This can result in efficiency and time saving. Also, knowledge or expertise among the diverse users in the community may be shared, for example, when agents learn from their peers. Apart from being able to save user’s time, interface agents also helps in preventing common errors from happening because they can help users carry out repetitive task by learning. 7.2 Information Agent Advantage Information agents help find the required information from the vast amount of information available on the internet. They are useful on the web where they can help us with routine and dull tasks. For example, searching for information on the Internet consumes a lot of time. The information agents can carry out such tasks for us and later on present us with the results. Information agents also bring big financial benefits. Take for example Netscape. They went being unknown to being a multi billion dollar company in a short amount of time. Their Netscape information manager with browsing capabilities helped them generate their revenue.#p#分页标题#e# 7.3 Reactive Agent Advantage The good thing about reactive agents is their ability to respond fast. However, their reactive nature may cause them to lose longer-term reasoning. Reactive agents are uncomplicated and easy to understand and their cognitive economy is very low (Ferber, 1994). This is due to their nature of having only very little to remember because planning ahead and the revision of world models is not done. Their actions only depend on what is happening at that very moment. Apart from that, reactive agents are also more robust and fault tolerant than other agents. For example, a reactive agent may be lost but it would not suffer any other terrible effects. Reactive agents are also flexible and adaptive unlike classical AI systems which are inflexible and demonstrate fragility and slow response times. 7.4 Mobile Agent Advantage Mobile agents are very efficient because they consume fewer network resources since they move the computation to the data instead of the data to the computation. Mobile agents also consumed considerably less bandwidth because bandwidth is consumed only when they travel. Apart from that, mobile agents have the ability to react dynamically to situations and this makes it easier to develop fault tolerance behavior in complex distributed systems. Mobile agents have a flexible distributed computing architecture – they provide a unique distributed computing architecture which works differently from the static set-ups. This will bring in an innovative way of executing distribute computation. Mobile agents also have support for e-commerce. They are used to develop e-markets since they represent the intentions, desires and resources of the participants in the market. Not only that, with mobile agents, there is prospect for fundamental and striking rethinking of the design process in general. Innovative and creative products would also materialize out of mobile agent technology. 7.5 Collaborative Agent Advantage With collaborative agents, control of consumer electronics and decentralized management can be executed. Being able to do that, collaborative agents are can solve problems which are too complex for a centralized single agent to handle due to resource constraints. Apart from that, with collaborative agents, multiple existing legacy systems such as decision support systems can be interconnected and interoperated. Collaborative agents also offer solutions which come from distributed information sources. 7.6 Benefits of Agents On the whole, all of the agents mentioned in this assignment have a few benefits. The first benefit is Automation. In detail, agents help in automating repetitive behavior of a single user, similar behavior of a group of users, and also repetitive sequential behavior of a number of users in a workflow thread. The next benefit is Customization. Agents can give customization advantages by relaying information that matches a user's personal data, a user's profile or even a user's interaction style preference. The third benefit is Notification. Agents can help a user by providing notification services which can help save the user's time and workload because he or she does not need to monitor events of personal importance. For example, such an agent can monitor websites of interest for any changes and then report them to the user. The fourth benefit is Tutoring. With the ability to monitor events and carry out inferencing, agents with tutoring ability can guide and train a user. With this, training requirements and modules can be reduced. A good example can be seen is the application wizards in the Windows OS environment. The last benefit is Messaging. Users are able to complete tasks at remote sites with the help of agents with messaging capabilities. An example of messaging agents is mobile agents which can transport themselves from one place to another to interact with other agents to carry out tasks on behalf of its owner. 8.0 总结——SUMMARY From this assignment, we can come up with a few key points. First and foremost, software agents will enable an e-business to offer round the clock self-service interaction which will be convenient for employees, customers as well as suppliers. Cost can be saved because less staff is needed to support customers, with software agents running in the background for self-service operation. Software agents relieve an organization's staffs by performing all the repetitive tasks so that the staffs can focus on other strategic workload. Apart from that, these agents help an e-business in creating a personal one-to-one relationship between the organization and the customers. This is because the agents can help in personalizing self-service operation by treating customers as individuals. Such personalization includes the agent remembering and analyzing the customer's previous interactions with the business, and narrow down the presentation of the business to the dynamic personal interests of the customer. In the end, the benefits of this are, existing customers are retained and loyal customer base can be developed. These software agents mentioned will help an e-business to form virtual communities among its customers and also to provide product and service recommendations by making use of the shared experiences of similar groups in these communities. A collaboration like this needs agents to analyze customer interaction data and search for groups made up of individuals with similar preferences. The advantage of doing this is the organization is able to target its products or services to more narrow segments. Subsequently, software agents help in allowing a business to organize a proactive dialogue with the customers. These agents can commence dialogues with customers to inform them of information regarding new as well as existing products and services and the status of pending customer-specified tasks. Thus, costs can be reduced and customer service will be improved with such proactivity. In fact, this excellent service will enable the business to increase customer loyalty. Other than that, software agents will significantly increase the speed of the interaction between the business and the customers. Collecting and analyzing the interaction data of customers in order to develop tactical corrections to the business operations can be done in real-time. 9.0结论—— CONCLUSION Intelligent agents are just programs executing what they were programmed to do, nothing more and nothing less. However, it is possible to start to apply existing technologies in new ways for the realization of end-user applications and the result of this innovation can be very compelling new set of capabilities. Compilers, database and Java systems are all examples of innovations that were based on existing technologies, but which created dramatic progress in computing through changing the user's model of how to use the computer. It is believed that while there are many ways to define intelligent agents, at their core, agents represent only a new application of existing technologies. The new application, however, provides a very compelling set of benefits to the end-user or the enterprise. 10.0参考文献—— REFERENCES 10.1 BOOK REFERENCES MULLER J.P., WOOLDRIDGE M., JENNINGS N. (1996). Intelligent Agents III: Agent Theories, Architectures and Languages. Springer, Germany. FERBER, J. (1999). Multi-Agent An Introduction to Distributed Artificial Intelligence. Addison-Wesley, Great Britain. CAGLAYAN A. & HARRISON C. (1997). AGENT Sourcebook: A Complete Guide to Desktop, Internet, and Intranet Agents. Wiley Computer Publishing, USA. 10.2 INTERNET REFERENCES LIEBERMAN H., Autonomous Interface Agents, [Online], 3rd June 2009. |
 |
|||
| 网站地图 |
