Chapter 10. Knowledge Representation (Category and objects (There are two…
Chapter 10. Knowledge Representation
Create representations of abstract concepts, focusing on general concepts (such as actions, time, physical objects and beliefs) that occur in very different domains.
General framework for the concepts
General convention to represent in the graphs the general concepts in the upper part and the more specific concepts below them
Category and objects
The organization of objects in categories is a vital part of the representation of knowledge.
The categories also serve to make predictions about the objects once they are classified.
There are two options to represent categories in first-order logic:
The categories serve to organize and simplify the base knowledge, through the Heritage.
Subclassification relationships organize categories into:
Taxonomies or relationships
Taxonomies have been used for centuries in technical fields.
Taxonomies are an important aspect in the general knowledge of common sense.
In general, the
relationship is used to say that something is part of something else.
Both in the scientific theories of the world and in those that appeal to common sense,
the objects have weight, mass, cost.
Specific purpose ontology
There are two main characteristics in general purpose ontologies that
distinguish from the collection of specific purpose ontologies:
It must be applicable to a greater or lesser extent to any domain of specific purpose
In a disparate domain, the different areas of knowledge must be unified,since the reasoning and the resolution of problems could involve several areas simultaneously.
Substances and objects
Separation in different objects
This portion is known generically as matter or substance.
Actions, situations and events
The ontology of the calculation of situations
Calculation of situations
Use the following ontologies:
They are functions and predicates that vary from one situation to the next
They are logical terms that consist of an initial situation and all the situations that are generated by the application of an action to a situation.
The functions or predicates timeless or eternal
These representations enable an agent to construct plans by logical inference.
A situation calculating agent should be able to deduce the result of a given sequence of actions.
This is the task of
. With an adequate constructive inference algorithm, you should be able to find a sequence that achieves the desired effect.
This is what is called
the planning task
Description of actions in the calculation of situations
Each action is described by two axioms:
Determines what happens when a possible action is executed
Specifies when it is possible to execute an action
Represent all things that remain unchanged
An efficient solution to the problem of the framework must be found because in the real world, almost everything remains unchanged all the time.
One approach is to write explicit
, that what they do is specify what remains unchanged.
It is specified that the truth value of each flow in the next state is a function of the action and the truth value in the current state.
Mental events and mental objects
They describe an attitude that an agent can take toward a proposition.
Reasoning systems for categories
They provide a graphical help to visualize a knowledge base, as well as efficient algorithms to infer properties of an object based on their belonging to a category.
One of the most important aspects of semantic networks is their ability to
represent default values for the categories.
Provides a formal language for constructing and combining category definitions, as well as efficient algorithms for deciding the relationships of subset and superset between categories.
It is an important form of inference, allowing to deduce the properties of objects based on their belonging to categories.
Reasoning with default information
Assumption of the closed world
Databases (and people) assume that the information provided is complete, so that atomic sentences for which there is no assertion that says they are true, are considered false
Manage knowledge updates and reviews efficient
Assumption of unique names
It is generally assumed that different names refer to different objects
Real maintenance systems
Review of the belief
Some of the inferred facts will be erroneous and will have to be retracted due to the appearance of new information.
True maintenance systems
Manage knowledge updates and reviews efficiently.