One immediate problem for building agent-oriented systems is finding an appropriate programming language platform. Currently, comparatively few programming languages have built - in support for agented-oriented programming and no completely agent - oriented mainstream programming language exists.Various agent - oriented programming languages have been proposed, such as:
This list is by no means exhaustive. Its purpose is to illustrate the variety of solutions that have been devised. Figure below provides a diagrammatic classification of a selection of the agent-based modeling platforms and languages.
Figure A selection of agent-oriented programming languages (based on Zhang, Lewis and Sierhuis).
The classification in the diagram splits agent - based modelling into three categories – agent-directed simulation, agent - oriented languages, and cognitive modelling. The first category covers languages that combine agent modelling with simulation. In the remaining sections of this chapter, we will look at one particular example in this category – NetLogo. According to Zhang, Lewis and Sierhuus
Netlogo’s advantages are that it has a small learning curve for the beginning programmer, it is easy to debug, and provides easy access to basic visualization, as it was “designed with the inexperienced programmer in mind”. Another advantage is that NetLogo programs are very compact and easy - to read, and therefore are much easier to convert to other programming languages if required. Hence, it is an ideal tool for pedagogical purposes.
An obvious feature of Figure above is the predominance of instances in the second category, with very few in the third category. As stated above, no agent - oriented programming language has yet to become mainstream in the way that C, Java and Python have become mainstream. Perhaps their common failing is the same failing that Prolog has – too much emphasis on a logic - based paradigm – which can be difficult to understand and debug for novice programmers. For example, Huget has listed desiderata for an agent-oriented programming language that includes features for handling logic (such as BDI for rational agents); but he has also listed other desirable features such as: knowledge and communication; definitions of organisations and environments; abstraction in order to avoid complexity in design; conformation to standards; ability to mix formalisms such as automata and Petri nets; and an event driven architecture (as used by reactive agents who react to events as they occur).
Although logic - based languages are interesting and powerful in their own right, they do not cater well to what programmers like doing best – getting useful things to work well with a minimum of fuss. Rather than develop a whole new agent - oriented programming language, an alternative approach is to develop a hybrid system on a non agent - oriented programming language. As stated, most programming languages do not have support for agent-oriented programming. The usual solution is to develop an agent framework in an object-oriented programming language since that is the current predominant programming paradigm in vogue.
There have been numerous agent frameworks developed, and most concentrate on issues to do with reasoning using a logic-based solution (such as BDI) as well as other capabilities such as agent - to - agent communication, usually using the standard KQML communication language.
What would an object-oriented agent framework look like? Objects can be used to represent agents in the system or application (Knapik and Johnson) such as scheduling agents, human interface agents, search agents, and so on.Object-oriented agents are defined with a class (‘AgentClass’ or ‘Root AgentClass’ say), and all agents have various things in common such as a name or ID, other global attributes, and a basic set of communication and error - handling protocols. Subclasses are used to define more specific types of agents – for example, ‘Text SearchAgent’ – that will have specific attributes and protocols. Each object-oriented agent has attributes that define the agent’s state and operations / methods that define the agent’s behaviour.
What are the benefits of developing agents using object-oriented technologies? Knapik and Johnson list the same benefits that accrue to the object-oriented programming paradigm: re-usable code; reduced agent development costs; flexible structuring of agent design; maintainability; extensibility; understandability; support for interconnected hierarchies of agents and domains; system knowledge is intrinsic; and a readily available development environment for modeling and simulation.
An increasing number of object-oriented agent frameworks have been developed using Java. (NetLogo is also implemented in Java). Java offers an object-oriented solution that is already integrated into the universal client – Web browsers – and supports packages such as java .net that can be used by agents to access and extract information from Web pages. Java has tremendous potential as an agent development language due to its widespread use on the Web. Similarly, Python is another language that is increasingly being used on the Web, and has support for the object-oriented programming paradigm, but to date, there has been less agent frameworks developed for it than Java. We will now have a brief look at a few other agent frameworks. Again, the purpose is not to be exhaustive and document the frameworks that are available as there are too many. The purpose is mainly to illustrate the variety of solutions that are possible, and to emphasize again that no single solution has captured the imaginations and interests of developers world-wide to become mainstream.
In Java, applets and servelets can be considered as somewhat akin to agents (Knapik and Johnson). Applets migrate from the server to the client, so execute on the client and thereby freeing up the server, so have some degree of mobility, an important trait of agents. Java servelets also let a user upload an executable program to the network, so a client user or application could launch a servelet based agent to search the network for information or respond automatically or give periodic updates. IBM’s aglets (for agile agents) is a Java mobile agent platform that adds further functionality to Java that is specifically focused on agent tasks (Aglets). Originally developed at the IBM Tokio Research Laboratory, the Aglets technology is now hosted at sourceforge.net. An aglet is a Java agent able to autonomously and spontaneously move from one host to another roaming the Internet.
A complete Java mobile agent platform is available, along with a stand-alone server called Tahiti, and there is a library that allows developer to build mobile agents or to have aglet code embedded in an application.Aglets can be halted, be packaged with their current state in another object dispatched to another environment, and have their execution resumed. The Java API has various classes such as: the class Aglet, an abstract class which is used to define the aglets; the Aglet Context which is an interface an aglet uses to gain knowledge about its environment; the Aglet Proxy that is a class that encapsulates the real aglet, protecting against direct access to the aglet's public interface; the Aglet Identifier that is a class that sets up a unique identifier for an aglet; the Itinerary class represents the routing to travel plans for an aglet; and the Message class that enables communication between agents.
JADE (for Java Agent Development Framework) is another framework fully implemented in Java (JADE) and is free software distributed by Telecom Italia. It supports the development of multiagent systems that comply to the FIPA specifications.There are graphical tools that aid the debugging and deployment phases, and remote GUI can be used to control the agent configuration, which can be distributed across machines, and which can be altered at run-time by moving agents from one machine to another.
LEAP (for Light Extensible Agent Platform) is an extension of JADE that allows a FIPA - compliant platform with reduced footprint compatible with mobile Java environments to run on wireless devices and PDA's such as cell phones and Palm computers. WADE (Workflows and Agents Development Environment) is an extension to JADE that allows agents to execute tasks defined according to the workflow metaphor. A WADE workflow is a Java class with a well defined structure that allows the developer to define a process in terms of the activities to be executed, their activation and termination criteria, and the relationships between them. Further information can be specified such as the participants in the workflow, the software tools to be invoked, the required inputs and expected outputs and internal data that will be manipulated during the execution.
The advantage of the workflow approach is that the execution steps as well as their sequencing are all made explicit. WADE comes with a development platform called WOLF that is an Eclipse plug - in. As stated above, mainly for pedagogical reasons, an agent directed simulation approach will be adopted for these books rather than the alternative approaches of using an agent programming language or using cognitive modelling.
Artificial Intelligence Related Interview Questions
|Cloud Computing Interview Questions||IBM Cloud Computing Infrastructure Architect V1 Interview Questions|
|Machine learning Interview Questions||Amazon Cloud Computing Interview Questions|
|Web analytics Interview Questions||SAS DI Interview Questions|
|Base Sas Interview Questions||SAS Macro Interview Questions|
Artificial Intelligence Related Practice Tests
|Cloud Computing Practice Tests||IBM Cloud Computing Infrastructure Architect V1 Practice Tests|
|Web analytics Practice Tests||SAS DI Practice Tests|
Artificial Intelligence Tutorial
Agents And Environments
Frameworks For Agents And Environments
All rights reserved © 2018 Wisdom IT Services India Pvt. Ltd
Wisdomjobs.com is one of the best job search sites in India.