18-04-2013, 02:28 PM
Knowledge Representation & Reasoning
Knowledge Representation.doc (Size: 138.5 KB / Downloads: 27)
Introduction
The existing of data in AI application and the need for reasoning and gaining new knowledge from existing one has grown enough to branch a new area of AI field that is oriented to the research of the best suitable way to represent data and hence to gain knowledge in a fast way. It could be seen that every domain of knowledge has its properties and specifications so the knowledge representation is oriented to get the best general way to represent all kind of domains in a standard way. By time some languages started to form a standard way of representation like the logical representation or the ontological representation or by forming a semantic networks or leverage it to contain frames of knowledge. Semantic networks. In specific gives links between structure the attention needed, as the first order logic (or higher) were not enough to describe these links. The attention to linking structures increased to give us a new ways of linking that made great experience to all of use every day like text hyperlinks and resource description framework (RDF)
Artificial intelligence (AI)
Intelligence is a wide term that describes the min properties like capacity, abstract thoughts, reasoning, understanding, learning, planning and problem solving . Intelligence of human brain is the study case although some other organics has intelligence like some plants and mammal animals, but it seems that humans have the most complete intelligence properties. Artificial intelligence is a field of computer science that tries to add intelligence to computer machines in the domain of acting and/or thinking .
Knowledge
There is no single definition for knowledge but we can use these definitions to get a general view of knowledge:
1. Expertise of skill acquired by person that is gained by experience or education.
2. Facts and Information known in particular field or in general.
3. Awareness with facts or situations and how to deal with it
Representation languages and notation
Some opinions say that we should simulate human mind in storing and manipulating data and also human natural language, but unfortunately human brain capabilities in processing natural language are not fully revealed so, that opinion is full of imperfections that are the source of error. So there is several Artificial intelligence symbolic languages and notations that is structured to define knowledge, some of them has been made for specific domain and others for general domain
Need of knowledge representation
In order to process the information and inference we need the data to be in a form that a machine could parse easily and to apply inference methods to gain new knowledge, a better representation is better performance of inference and reasoning, so KR is an important area of Artificial intelligence field
Frame representation
In order to expand the capabilities of semantic networks an object in semantic networks is expanded to be a frame, every frame has some properties, the frame representation follow the slot/filler approach. Inheritance could be applied to frames where frame inherit properties of another frame. When agent comes to a new situation a frame slots could be filled with values to generate new object .
Note: the templates of CLIPS are just what are we talking about if represented in a graph.
Conclusion
Knowledge representation is an important area of artificial intelligence field, and it has been proven that research in this area was a force power in pushing the Web, and it is much related to mathematics and theorems proving. The problem of AI is to describe and build agents that receive precepts from the environment and perform actions, and each such agent is implemented by a function that maps precepts to actions. It explains the role of learning as extending the reach of the designer into unknown environments, and shows how it constrains agent design, favouring explicit knowledge representation and reasoning . It analyzes basic techniques for addressing complexity . The idea is to Integrate state-of-the art AI techniques into intelligent agent designs, using examples from twelve agents to full knowledge-based agents with natural language capabilities and so on . This leads to the study of Multi-Agent systems and its applications