Knowledge Engineering and Knowledge Representation are one of the fundamental parts of an Artificial Intelligence system. In this article, we will go through these concepts and understand how it is used in the field of AI.
Knowledge Engineering
First, let us understand what is Knowledge Engineering.
"Knowledge Engineering refers to all technical, scientific and social aspects involved in building, maintaining and using the knowledge-based system"
Knowledge-based systems or KBS is a computer program that reasons and uses the knowledge base to solve complex problems.
The subsystem of a KBS include:
1. Knowledge base: It is a computer program which stores complex structured and unstructured information that a computer system uses. This exhibits opposite properties of relational database systems. Knowledge base stores structured data but not as a table with numbers and strings, but as pointers to other objects that in turn has additional pointers. It is represented as an object model which has class, subclass, and instances. This is also referred to as Ontologies.
2. User- Interface: Human-Computer Interaction.
3. Inference Engine: Applies logical rule to the knowledge base to deduce new information.
KBS was primarily known as expert systems which were first developed by AI researchers.
What does this expert systems do??
1. It performs certain complex tasks
2. Aide human in some specific tasks
3. Replace human for some tasks
The earlier KBS were mostly expert systems. We can say that all expert systems are KBS but not all KBS are expert systems. The first system that was developed was the rule-based expert system. These systems used knowledge explicitly rather than a procedural code.
Advantages of rule-based expert systems:
1. Acquisition and maintenance are easy since there was no need for programmers. The domain experts can define and maintain the ruled themselves.
2. The knowledge here is used explicitly which allowed the systems to reason regarding how they got the results and explain them to the user.
3. These systems develop conclusion from data with strong reasoning
As the KBS became more complex, the representation also became more sophisticated. The knowledge base became more and more structured. The reasoning is based on both independent rules as well as interaction with the knowledge base. The recent advancement has been to adopt technologies to develop systems which use the internet. The internet has a large amount of complex and unstructured data which is difficult to fit in any data model. The KBS technique of classifying data and getting it into a proper format can be used here. These systems are called Semantic Web.
Knowledge Representation
We have gone through the KBS systems and now let us see what is Knowledge Representation (KR). KR defines how the information is represented so that the computer system can understand and utilize it solve complex problems. It is also known as Knowledge Representation and Reasoning.
KR justifies that the conventional procedural code is not enough to solve complex problems. KR makes a complex software easy to define and maintain procedural code. KR is connected to the automated reasoning engine which enables them to reason, make inferences, add new knowledge etc. All KBS have reasoning or inference engine as part of the system.
Knowledge Representation Framework:
1. Reasoning without taking actions
2. Answer to the questions (Ontological Commitment)
3. Intelligent reasoning, sanctioning inferences and recommending inferences.
4. A computational environment where thinking is accomplished
5. Language
Some of the examples of KR include semantic nets, system architecture, frames, rule, and ontologies.
Characteristics of Knowledge Representation & Reasoning:
1. The primitive framework used to represent knowledge like semantics, frames, rules etc.
2. Capability to have information of its own state which is also known as meta-representation
3. Non- monotonic reasoning which allows various kinds of hypothetical reasoning.
4. Reasoning efficiency
This is the brief explanation of Knowledge Engineering and Knowledge Representation. Both in itself is a very vast topic. There are so many terminologies involved and I will explain all of them in my next article.
Comments
Post a Comment