Personal tools

Knowledge Representation, Reasoning, and Logic

Old Nassau_Princeton University_110821A
[Old Nassau, Princeton University - Office of Communication]


- Overview 

Knowledge representation and reasoning (KRR) is a field of artificial intelligence (AI) that studies how to represent information about the world in a way that computers can use to solve problems. 

KRR is an important area of AI research. KRR involves: 

  • Representing knowledge in a formal language that a computer program can interpret
  • Using reasoning algorithms to solve problems


KRR can help computers solve complex tasks, such as: 

  • Diagnosing medical conditions
  • Having a natural language dialogue
  • Problem solving
  • Knowledge reasoning


Please refer to the following for more information:


- Knowledge

Knowledge is the most important resource of modern times, but there is still a lot we don't know about it. Knowledge is the amalgam of experience, values, content information, and expert insights and data that form the framework for integrating new information. 

The different kinds of knowledge that need to be represented in AI include: objects, events, Performance, facts, meta-Knowledge, and knowledge-base.

Knowledge representation plays a crucial role in AI. It is related to the "thinking" of an AI system and contributes to its intelligent behavior. Knowledge representation is a radical new approach in AI that is changing the world.

Humans are best at understanding, reasoning and interpreting knowledge. Humans know things, that's knowledge, and based on their knowledge, they perform various operations in the real world. But how machines do all these things comes under knowledge representation and reasoning. 

The purpose of knowledge representation is to express knowledge in a form that can be easily processed by a computer so that it can be used to make our AI agents perform well. 

In short, knowledge representation enables machines to behave like humans by empowering AI machines to learn from available information, experience, or experts. 


- Knowledge Representation Formalisms and Methods

Knowledge representation in AI is the process of organizing information so that AI systems can comprehend and use it. Knowledge representation is a key part of expert systems, and helps AI agents solve complex tasks.

Knowledge representation formalisms and methods are used to represent and process knowledge. 

Knowledge representation incorporates findings from psychology about how humans solve problems, and represent knowledge in order to design formalisms that will make complex systems easier to design and build. Knowledge representation and reasoning also incorporates findings from logic to automate various kinds of reasoning. 

Examples of knowledge representation formalism include semantic nets, frames, rules, logic programs and ontologies. Examples of automated reasoning engines include inference engines, theorem provers, model generators and classifiers. 

Here are some examples of knowledge representation formalisms:

  • Semantic networks: A graphic network that stores knowledge as nodes and arcs. It can represent physical objects, concepts, or situations. Semantic networks can abstract from natural language and represent knowledge captured in text.
  • Logical representation: A representation that draws conclusions based on conditions. It has a precise syntax and semantics that support sound inference.
  • Frames: An object-based knowledge representation language.
  • Description Logics: A language based on predicates expressed in logic.


Some examples of knowledge representation methods include: Symbolic representation, Connectionist representation, Hybrid representations, Declarative, Procedural, Metaknowledge, Heuristic, Structural. 

Some other knowledge representation methods include: Logical representation, Attribute-value representation, Production rules, Lists, Scripts. 


- Knowledge-based Systems

Knowledge-based systems (KBSes) are a type of artificial intelligence (AI) that use a centralized repository of data to help with problem-solving. KBSes are also known as expert systems, intelligent agent systems, or knowledge systems. 

KBSes are designed to capture the knowledge of human experts to support decision-making. They represent knowledge explicitly, rather than as procedural code. 

Here are some other aspects of knowledge processing systems in AI: 

  • AI knowledge base: A dynamic, centralized repository for information integrated with AI technologies.
  • AI knowledge management system: A tool that uses AI to analyze, interpret, and extract data from large data sets. It then organizes and distributes that knowledge within a business.
  • Meta knowledge: Predefined knowledge about things that we are already aware of. This knowledge typically includes the study of tagging, planning, learning, etc..
  • Knowledge representation: There are four main approaches to knowledge representation in AI: Simple relational knowledge, Inheritable knowledge, Inferential knowledge, Procedural knowledge


[Vienna, Austria]

- Knowledge Processing Systems

In the field of artificial intelligence (AI), whether in the field of machine learning (ML) or deep learning (DL), many complex tasks need to be evaluated. There is indeed a need to automate knowledge processing systems in such systems. 

Knowledge representation is such a process that depends on logical situations and enables policies to make decisions when knowledge is acquired. Human beings acquire various types and levels of knowledge in daily life, but it is difficult for machines to explain all types of knowledge. For this case, a knowledge representation is used. 

In knowledge representation algorithms, AI agents tend to think and participate in decision-making. With this complex thinking, they are able to solve complex problems in real-world scenarios that are difficult and time-consuming for humans.


- Semantic Knowledge Graphs

A semantic graph is a graph model and knowledge representation that uses nodes and edges to represent knowledge. It also structures the meaning of knowledge in a way that can be used for reasoning and inference. 

A semantic network, also known as a knowledge graph, is a network of real-world entities, such as events, objects, situations, or concepts. It illustrates the relationship between these entities. 

In AI, a semantic network is a knowledge representation technique for organizing and storing knowledge. It is a type of graphical model that shows the relationships between concepts, ideas, and objects in a way that is easy for humans to understand. 

Semantic knowledge graphs can help with: 

  • Structuring and analyzing large amounts of data across different sources
  • Data query and analysis
  • Inference from data
  • Generating insights

Anzo is an example of knowledge graphs and AI at the intersection. It uses semantics and graph to enable and use AI techniques. 


- The Goal of Reasoning in AI

AI requires reasoning to reason and perform similarly to a human brain. The goal of reasoning in AI is to create machines that can reason like humans. Reasoning uses logical rules and principles to derive new information from existing information. AI systems use reasoning to: Make inferences, Draw conclusions, Solve problems. 

Reasoning is the main weakness of AI machines compared to human intelligence. These machines are not only able to provide feedback and answer questions; however, they cannot explain how they arrived at their conclusions.

In AI and cognitive science, probabilistic approaches are crucial for generation, reasoning, and decision-making (simple and complex).

In AI, reasoning can be divided into the following categories:

  • Deductive reasoning
  • Inductive reasoning
  • Abductive reasoning
  • Common Sense Reasoning
  • Monotonic Reasoning
  • Non-monotonic Reasoning


- Steps for Building a Knowledge Base

Knowledge representation and reasoning (KRR) is the field of AI that represents information about the world in a form that a computer system can use. For example, an AI agent can solve a chessboard puzzle only if it has gained sufficient knowledge on how to solve the puzzle and win the game. 

Some techniques related to knowledge representation in AI include: 

  • Inferential adequacy: Making AI machines with the flexibility enough to deal with the present knowledge for newly possessed knowledge
  • Inferential efficiency: The ability to direct the inferential knowledge mechanism into the most productive directions by storing appropriate guides


Some steps for building a knowledge base include: 

  • Identifying the purpose of the knowledge base
  • Listing down the content types for the knowledge base
  • Adding visuals to boost the content readability
  • Interlinking different content modules in the knowledge base
  • Publishing the knowledge base across relevant channels


An expert system is an example of a knowledge-based system because it relies on human expertise.


- Knowledge Representation and Reasoning in AI

We can describe knowledge representation as follows: Knowledge representation and reasoning (KR, KRR) is the part of AI that involves the thinking of the AI ​​agent and how thinking leads to the intelligent behavior of the agent. 

It is responsible for representing information about the real world so that computers can understand and use this knowledge to solve complex real-world problems, such as diagnosing medical conditions or communicating with humans in natural language. It's also a way of describing how we represent knowledge in AI. 

Knowledge representation is not just about storing data into some database, it enables intelligent machines to learn from this knowledge and experience so that they can act intelligently like humans. 

Some types of knowledge representation include:

  • Declarative: Explains facts
  • Procedural: Explains behavior
  • Metaknowledge: Knowledge about other topics of knowledge
  • Heuristic: Knowledge of specific fields and domains
  • Structural: Knowledge for seeing the relations between different objects 


- Knowledge Representation Makes Complex Software Easier

Knowledge representation is important in AI because it allows computers to understand, store, and manipulate human knowledge, enabling them to solve complex problems, make decisions, and perform tasks that require intelligence. 

The basic principle of knowledge representation is that traditional programming is not the best form for solving complex problems. Knowledge representation makes complex software easier to define and maintain than procedural code, and can be used in expert systems. 

For example, talking to experts about business rules rather than code can reduce the semantic gap between users and developers, making the development of complex systems more practical. 

Knowledge representation is closely related to automatic reasoning, since one of the main purposes of explicitly representing knowledge is to be able to reason about that knowledge, make inferences, assert new knowledge, etc. Almost all knowledge representation languages ​​have reasoning or reasoning engines as part of the system.


- Logic AI

Logic is the basic concept of artificial intelligence (AI). It allows AI systems to draw conclusions and inferences from data.

Logical AI involves the use of logical statements to represent an agent's knowledge, goals, and current situation. The agent then decides what to do by inferring that a certain action or course of action is appropriate to achieve the goal.

Logical thinking AI goes beyond pattern recognition and statistical learning. It aims to imitate human cognitive processes and decision-making. Logical thinking AI focuses on building the ability to:

  • Draw logical conclusions
  • Detect inconsistencies
  • Reasoning through complex problems


Formal logic in AI is important so that agents or systems can think and act like humans. It ensures that information is shared with minimal errors and that AI conclusions are either correct or incorrect.  

Logical AI involves representing an agent's knowledge of the world, its goals, and its current situation through logical sentences. An agent decides what to do by inferring that a certain action or course of action is appropriate to achieve a goal. 

Human-level AI requires programs that can handle common sense information situations. Human-level logical AI requires expanding the way logic is used in the formalization of branches of mathematics and physical science. It also seems to require extensions to logic itself, both in the formalism used to express knowledge and in the reasoning used to arrive at conclusions.

The five logical symbols in AI are: negation, conjunction, disjunction, implication, and biconditional.


- Logic in Knowledge Representation

In knowledge representation, logic is used to construct models for given domains. Logic can give a processable form to information that can be precisely expressed in any other language. 

Logic-based representation uses formal logic to represent knowledge as a set of rules and axioms that can be used to reason about the world. 

Logical representation is a language with some definite rules which deal with propositions and has no ambiguity in representation. It represents a conclusion based on various conditions and lays down some important communication rules. 

For a knowledge-based intelligent agent, logic for knowledge representation and reasoning needs:

  • To represent knowledge about the world in a formal language
  • To reason about the world using inferences in the language
  • To decide what action to take by inferring that the selected action is good 


Statements in propositional logic can be either true or false. Predicate logic is able to understand categories and variables.



[More to come ...]

Document Actions