How would you describe all the knowledge in the world, and what would you do with it? That, in a nutshell, is the concern of knowledge representation (KR), a subfield of study within artificial intelligence (AI). It’s a process that takes all the concepts in a domain, establishes how these concepts are related, and defines rules that control how they behave.
To illustrate, think about the real world. It’s large and infinitely complex. So we can reduce it to an abstract model like a map that captures only aspects of the world that are relevant to us, such as its geography. But a computer can’t understand a map, so we reduce this even further into a set of rules and statements that represents the map. In other words, KR represents information in a way computers can understand.
Other interesting terms…
Read More about Knowledge Representation in Artificial Intelligence
KR is a necessary component of AI. It allows programmers to convert real-world data into a language that a computer can use to develop an information system (IS) that is critical in accomplishing tasks.
An example would be when a company needs to write a contract. It can use KR to state the intentions of the parties involved to ensure the agreement reflects the best interests of all stakeholders. How can knowledge engineers achieve this goal? They do so by creating a sound KR system.
What Properties Does a Good Knowledge Representation System Have?
To be considered good, a KR system must have the following features:
1. Representational Adequacy
It should be able to represent the different kinds of knowledge required.
2. Inferential Adequacy
The KR system should be able to come up with new structures or knowledge that it can infer from the original or existing structures.
3. Inferential Efficiency
It should be able to integrate additional mechanisms to existing knowledge structures to direct them toward promising directions.
4. Acquisitional Efficiency
The KR system should be able to gain new knowledge through automated methods instead of relying on human intervention. However, it should also allow for the injection of information by a knowledge engineer.
To date, no single KR system has all of these properties.
What Is the Cycle of Knowledge Representation?
AI follows the so-called “cycle of knowledge representation” in order for systems to work properly. The cycle comprises five stages—perception, learning, KR, reasoning, planning, and execution. Let’s tackle them one by one.
Stage #1: Perception
Perception involves acquiring information from the environment through sensors. The data acquired is then processed and interpreted to generate knowledge.
Stage #2: Learning
Learning entails acquiring new knowledge from experience. It can be achieved through supervised learning, unsupervised learning, or reinforcement learning.
Stage #3: Knowledge Representation and Reasoning
KR and reasoning involves transforming acquired knowledge into a form that machines can process. It involves choosing an appropriate KR technique and representing knowledge using that technique. Reasoning means using the knowledge represented to draw inferences and make decisions.
Stage #4: Planning
Planning is where a system uses the knowledge acquired and reasoning to generate actions to achieve a particular goal. It involves choosing the most appropriate actions to achieve the goal, taking into account constraints or limitations.
Stage #5: Execution
Execution is where a system performs planned actions. Its success depends on the accuracy and completeness of the KR, reasoning, and planning.
What Kinds of Knowledge Are Represented in Artificial Intelligence?
Various knowledge types need to be represented in AI, namely:
- Objects: Real-world things with physical properties. They can be observed, touched, or manipulated.
- Events: Real-world actions or occurrences like driving a car, writing, or watching TV.
- Performance: Behavior of agents or systems that perform tasks, including their goals and objectives and the criteria used to evaluate performance.
- Facts: Propositions that are either true or false. They are statements that can be verified using evidence or logical deduction.
- Meta-knowledge: Information about the structure and organization of knowledge, its sources, and its reliability and validity. It helps machines reason about the quality and validity of the knowledge they are using.
- Knowledge base: Collection of knowledge organized and stored in a way that machines can access and use. It includes facts, rules, procedures, and other knowledge relevant to a particular domain.
What Factors Should You Consider When Choosing a Knowledge Representation System?
While finding the perfect KR system is not feasible for now, an effective one should have these characteristics:
Comprehensive
As much as possible, KR systems must be extensive. All aspects that a KR system claims to consider must be well-represented and easily decipherable.
Computable
KR systems must cover a wide range of standard computing procedures to ensure widespread application.
Accessible
KR systems must be easily accessible. Each domain or system must have the means to identify events and decipher how different components react.
Relevant
KR systems must not contain unnecessary details that may only complicate processes.
Transparent
All stakeholders must get an overview of all processes and a clear understanding of the events that surround a KR system’s implementation.
Concise
All KR system outputs must be timely and accurate.
Key Takeaways
- KR, a subfield of AI, is a process that takes all the concepts in a domain, establishes how these concepts relate to each other, and defines the rules that control how they behave.
- An effective KR system has representational adequacy, inferential adequacy, inferential efficiency, and acquisitional efficiency.
- A good KR system candidate is comprehensive, computable, accessible, relevant, transparent, and concise.
