750 likes | 984 Views
CS 785, Fall 2001. Object Ontologies. Gheorghe Tecuci tecuci@cs.gmu.edu http://lalab.gmu.edu/. Learning Agents Laboratory Department of Computer Science George Mason University. Overview. General organization of the knowledge base. 1. Representation of the object ontology. 2.
E N D
CS 785, Fall 2001 Object Ontologies Gheorghe Tecuci tecuci@cs.gmu.eduhttp://lalab.gmu.edu/ Learning Agents LaboratoryDepartment of Computer Science George Mason University
Overview General organization of the knowledge base 1 Representation of the object ontology 2 Reasoning with the object ontology 3 An example: the COA object ontology 4 More exercises 5 6 Required reading
1. General organization of the knowledge base The structure of the knowledge base The generality of the ontology The generality of the rules
The structure of the knowledge base Knowledge Base = Object ontology + Task reduction rules The objectontology is a hierarchical description of the objects from the domain, specifying their properties and relationships. It includes both descriptions of types of objects (called concepts) and descriptions of specific objects (called instances). The task reductionrules specify generic problem solving steps of reducing complex problem solving tasks to simpler tasks. They are described using the objects from the ontology.
The structure of the knowledge base (cont.) Knowledge Base = Object ontology + Task reduction rules A task reduction rule is an IF-THEN structure that expresses the condition C under which a task T1 can be reduced to the simpler tasks T1a, or to a set of simpler tasks T11, … , T1n. T1 T1 C1 C2 T1a T11 T12 … T1n
The generality of the ontology An object ontology is characteristic to an entire application domain, such as military or medicine. In the military domain the object ontology will include descriptions of military units and of military equipment. These descriptions are most likely needed in almost any specific military application. Because building the object ontology is a very complex task, it makes sense to reuse these descriptions when developing a knowledge base for another military application, rather than starting from scratch.
The generality of the rules The rules from the knowledge base are specific to a particular application and even to a particular SME. Consider, for instance, the agents discussed before, the agent that critiques courses of action with respect to the principles of war, and the agent that plans the repair of damaged bridges or roads. While both agents need to reason with military units and military equipment, their reasoning rules are very different, being specific not only to their particular application (critiquing vs planning), but also to the SMEs whose expertise they encode.
2. Representation of the object ontology Sample application: COA critiquing Semantic network representation of the ontology Instances, concepts and generalization Object features Definition of instances and concepts
Sample application: COA critiquing Identifies strengths and weaknesses in a military course of action based on the principles of war and tenets of army operations.
COA411 – the sketch Graphical depiction of a preliminary plan. It includes enough of the high level structure and maneuver aspects of the plan to show how the actions of each unit fit together to accomplish the overall purpose.
COA411 – the statement Explains what the units will do to accomplish the assigned mission.
Object ontology as a Semantic network The underlying idea of the semantic network representation is to represent the object ontology in the form of a graph in which the nodes represent objects and the arcs represent the relations between them.
Characterization of instances and concepts A concept is a representation of a set of instances. Represents the set of all armored units (which includes BLUE-ARMORE-BRIGADE2 and BLUE-TASK-FORCE1) ARMORED-UNIT--MILITARY-SPECIALTY INSTANCE-OF INSTANCE-OF Represents the entity calledBLUE-TASK-FORCE1 BLUE-ARMOR-BRIGADE2 BLUE-TASK-FORCE1 An instance is a representation of a particular entity in the application domain.
Intuitive definition of generalization Generalization is a fundamental relation between concepts. Intuitively, a concept P is said to be more general than (or a generalization of) another concept Q if and only if the set of instances represented by P includes the set of instances represented by Q. MODERN-MILITARY-UNIT--DEPLOYABLE MANEUVER-UNIT-MILITARY-SPECIALTY AVIATION-UNIT- -MILITARY-SPECIALTY INFANTRY-UNIT--MILITARY-SPECIALTY ARMORED-UNIT--MILITARY-SPECIALTY
A generalization latice/hierarchy ORGANIZATION SUBCLASS-OF MODERN-MILITARY-ORGANIZATION SUBCLASS-OF MODERN-MILITARY-UNIT--DEPLOYABLE SUBCLASS-OF SUBCLASS-OF MANEUVER-UNIT-MILITARY-SPECIALTY AVIATION-UNIT--MILITARY-SPECIALTY SUBCLASS-OF SUBCLASS-OF ARMORED-UNIT--MILITARY-SPECIALTY INFANTRY-UNIT--MILITARY-SPECIALTY INSTANCE-OF SUBCLASS-OF BLUE-ARMOR-BRIGADE2 MECHANIZED-INFANTRY-UNIT--MILITARY-SPECIALTY INSTANCE-OF BLUE-TASK-FORCE1 INSTANCE-OF INSTANCE-OF BLUE-MECH-BRIGADE1 BLUE-TASK-FORCE2 INSTANCE-OF BLUE-TASK-FORCE3
Object features The objects in the application domain may be described in terms of their properties and their relationships with each other. OBJECT-ACTED-ON RED-MECH-COMPANY4 PENETRATE1 FORCE-RATIO 10.6 PENETRATE1 acts on RED-MECH-COMPANY4 with a force ratio of 10.6.
Feature definition An object feature is itself characterized by several features which include: documentation, domain and range. The domain is the concept that represents the set of objects that could have that feature. The range is the set of possible values of the feature. ASSIGNMENT BLUE-TASK-FORCE2 SUPPORTING-EFFORT1 <OBJECT-FEATURE> DOCUMENTATION SUBCLASS-OF "Indicates the assignment of a unit" DOMAIN ASSIGNMENT MODERN-MILITARY-UNIT--DEPLOYABLE RANGE COA-ASSIGNMENT
Feature definition: example IS-OFFENSIVE-ACTION-FOR PENETRATE--MILITARY-TASK "military offensive operation" <OBJECT-FEATURE> DOMAIN ACTION SUBCLASS-OF IS-OFFENSIVE-ACTION-FOR RANGE {"military offensive operation”, "military offensive operation”}
Partially learned feature IS-OFFENSIVE-ACTION-FOR PENETRATE--MILITARY-TASK "military offensive operation" DOCUMENTATION "Indicates the context in which the action is considered as having an offensive nature” PLAUSIBLE UPPER BOUND: ACTION DOMAIN IS - OFFENSIVE - ACTION - FOR PLAUSIBLE LOWER BOUND: PENETRATE-MILITARY-TASK {"military offensive operation”, "military defensive operation”} PLAUSIBLE UPPER BOUND: RANGE {"military offensive operation”} PLAUSIBLE LOWER BOUND:
Feature hierarchy <OBJECT - FEATURE> DOCUMENTATION "Indicates the assignment of a unit" MODERN - MILITARY - UNIT -- DEPLOYABLE UPPER BOUND: DOMAIN MANEUVER - UNIT - MILITARY - SPECIALTY LOWER BOUND: ASSIGNMENT COA - ASSIGNMENT UPPER BOUND: RANGE MAIN - EFFORT LOWER BOUND: DOMAIN ACTION IS - ACTION - TYPE - FOR RANGE {"military offensive operation”, "military defensive operation”} IS - SURPRISE - ACTION - FOR IS - SECURITY - ACTION - FOR "Indicates the context in which the action is considered as having an offensive nature” DOCUMENTATION ACTION UPPER BOUND: DOMAIN IS - OFFENSIVE - ACTION - FOR PENETRATE - MILITARY - TASK LOWER BOUND: UPPER BOUND: {"military offensive operation”,"military defensive operation”} RANGE {"military offensive operation”} LOWER BOUND:
Definition of instances and concepts When designing a knowledge base, one has to first specify some basic concepts, as well as the features that may characterize instances and concepts. Once basic concepts and features are specified, one can define new concepts and instances as logical expressions of the known concepts.
Basic representation unit conceptk ISA concepti FEATURE1 value1 . . . FEATUREn valuen This is a necessary definition of ‘conceptk’. It defines ‘conceptk’ as being a subconcept of ‘concepti’ and having additional features. This means that if ‘concepti’ represents the set Ci of instances, then ‘conceptk’ represents a subset Ck of Ci. The elements of Ck have the features ‘FEATURE1’,..., ‘FEATUREn’ with the values ‘value1’,..., ‘valuen’, respectively.
Example: Concepts definition We can define a concept as being a sub-concept of known concepts and having additional features, as in the following example: MILITARY-MANEUVER MILITARY-ATTACK COMPLEX-MILITARY-TASK INDICATES-MISSION-TYPE "military offensive operation" SUBCLASS-OF IS-OFFENSIVE-ACTION-FOR "military offensive operation" PENETRATE--MILITARY-TASK RECOMMENDED-FORCE-RATIO 3 HAS-SURPRISE-FORCE-RATIO 6 PENETRATE--MILITARY-TASK is a complex military task, a military maneuver, and a military attack. It indicates that a COA that has a penetration mission is an offensive COA. In the context of an offensive operation, a penetration should be considered an offensive action. The doctrinal recommended force ratio for a penetration is 3.0. A force ratio of 6.0 should be considered a surprisingly high force ratio.
Example: Instance definition ARMORED-UNIT--MILITARY-SPECIALTY MECHANIZED-INFANTRY-UNIT--MILITARY-SPECIALTY INSTANCE-OF INSTANCE-OF ECHELON-OF-UNIT BATTALION--UNIT-DESIGNATION SOVEREIGN-ALLEGIANCE-OF-ORG BLUE--SIDE ASSIGNMENT MAIN-EFFORT1 BLUE-TASK-FORCE1 is a blue armored and mechanized infantry battalion assigned to be main effort1. It performs two tasks, penetrate1 and clear1. It has a regular strength and has the following units under its operational control: BLUE-MECH-COMPANY1, … TASK PENETRATE1 TASK BLUE-TASK-FORCE1 CLEAR1 TROOP-STRENGTH-OF-UNIT REGULAR-STATUS OPERATIONAL-CONTROL-MILITARY-ORG BLUE-MECH-COMPANY1 OPERATIONAL-CONTROL-MILITARY-ORG BLUE-MECH-COMPANY2 OPERATIONAL-CONTROL-MILITARY-ORG BLUE-ARMOR-COMPANY1 OPERATIONAL-CONTROL-MILITARY-ORG BLUE-ARMOR-COMPANY2
3. Reasoning with the object ontology Transitivity of INSTANCE_OF and SUBCLASS_OF Inheritance Object expressions Generalization and specialization rules Types of generalizations and specializations Ontology matching Ontology maintenance Steps in ontology development
Transitivity of INSTANCE_OF and SUBCLASS_OF ORGANIZATION ORGANIZATION ORGANIZATION SUBCLASS-OF MODERN-MILITARY-ORGANIZATION SUBCLASS-OF MODERN-MILITARY-UNIT--DEPLOYABLE SUBCLASS-OF MANEUVER-UNIT--MILITARY-SPECIALTY SUBCLASS-OF SUBCLASS-OF INFANTRY-UNIT--MILITARY-SPECIALTY INFANTRY-UNIT--MILITARY-SPECIALTY SUBCLASS-OF MECHANIZED-INFANTRY-UNIT--MILITARY-SPECIALTY INSTANCE-OF INSTANCE-OF BLUE-MECH-BRIGADE1 BLUE-MECH-BRIGADE1
Inheritance MILITARY-TASK MILITARY-MANEUVER SUBCLASS-OF MILITARY-ATTACK COMPLEX-MILITARY-TASK INDICATES-MISSION-TYPE "military offensive operation" SUBCLASS-OF IS-OFFENSIVE-ACTION-FOR "military offensive operation" PENETRATE--MILITARY-TASK RECOMMENDED-FORCE-RATIO 3 HAS-SURPRISE-FORCE-RATIO 6 INSTANCE-OF OBJECT-ACTED-ON RED-MECH-COMPANY4 FORCE-RATIO PENETRATE1 10.6 IS-TASK-OF-OPERATION ATTACK2 TASK-HAS-PURPOSE TASK (missing element) UNIT-ASSIGNED-TO-TASK ASSIGNMENT BLUE-TASK-FORCE1 MAIN-EFFORT1 SOVEREIGN-ALLEGIANCE-OF-ORG BLUE--SIDE
Object concept representation One can define more complex concepts as logical expressions involving the basic concepts from the object ontology. The following concept represents the set of all COA specifications that contain between 1 and 25 offensive actions: ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION ?N1 ?N1 IS-IN [1 .. 25] ?O1 is a generic instance of this concept. It is a COA specification that has ?N1 offensive actions, where ?N1 is between 1 and 25.
Object expressions The following, for instance, represents the set of deployable military units ?O2 of the red side that perform some intelligence collection military task: ?O2 IS MODERN-MILITARY-UNIT--DEPLOYABLE SOVEREIGN-ALLEGIANCE-OF-ORG ?O4 TASK ?O3 ?O3 IS INTELLIGENCE-COLLECTION--MILITARY-TASK ?O4 IS RED--SIDE
Generalization (and specialization) rules Turning constants into variables Climbing the generalization hierarchy Dropping condition Generalizing numbers Adding alternatives
Generalization and specialization rules A generalization rule is a rule that transforms an expression/concept into a more general one. A specialization rule is a rule that transforms an expression/concept into a less general one. The reverse of any generalization rule is a specialization rule.
Turning constants into variables Generalizes an expression by replacing a constant with a variable. The set of COAs with 5 offensive actions. ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION 5 generalization specialization 5 ?N1 ?N15 ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION ?N1 The set of COAs with any number of offensive actions.
Climbing the generalization hierarchy Generalizes an expression by replacing a concept with a more general one. MODERN MODERN - - MILITARY MILITARY - - UNIT UNIT -- -- DEPLOYABLE DEPLOYABLE SUBCLASS SUBCLASS - - OF OF SUBCLASS SUBCLASS - - OF OF MANEUVER MANEUVER - - UNIT UNIT - - MILITARY MILITARY - - SPECIALTY SPECIALTY AVIATION AVIATION - - UNIT UNIT -- -- MILITARY MILITARY - - SPECIALTY SPECIALTY SUBCLASS SUBCLASS - - OF OF SUBCLASS SUBCLASS - - OF OF ARMORED ARMORED - - UNIT UNIT -- -- MILITARY MILITARY - - SPECIALTY SPECIALTY INFANTRY INFANTRY - - UNIT UNIT -- -- MILITARY MILITARY - - SPECIALTY SPECIALTY The set of armored units at the battalion level. ?O1 IS ARMORED-UNIT--MILITARY-SPECIALTY ECHELON-OF-UNIT BATTALION--UNIT-DESIGNATION generalization specialization ARMORED-UNIT--MILITARY-SPECIALTY MANEUVER-UNIT--MILITARY-SPECIALTY MANEUVER-UNIT --MILITARY-SPECIALTY ARMORED-UNIT--MILITARY-SPECIALTY ?O1 IS MANEUVER-UNIT--MILITARY-SPECIALTY ECHELON-OF-UNIT BATTALION--UNIT-DESIGNATION The set of maneuver units at the battalion level.
Dropping condition Generalizes an expression by removing a constraint from its description. The set of armored units at the battalion level. ?O1 IS ARMORED-UNIT-MILITARY-SPECIALTY ECHELON-OF-UNIT BATTALION-UNIT-DESIGNATION generalization specialization ?O1 IS ARMORED-UNIT-MILITARY-SPECIALTY The set of armored units (at any level). Conversely, one can specialize an expression by adding a constraint.
Generalizing numbers Generalizes an expression by replacing a number with an interval, or by replacing an interval with a larger interval. The set of COAs with 5 offensive actions. ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION 5 generalization specialization 5 [1 .. 25] [1 .. 25] 5 ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION ?N1 ?N1 IS-IN [1 .. 25] generalization specialization [1 .. 25][1 .. 30] [1 .. 30] [1 .. 25] ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION ?N1 ?N1 IS-IN [1 .. 30] The set of COAs with at least one and at most 30 of offensive actions.
Adding alternatives Generalizes an expression by replacing a concept C1 with the union (C1 U C2), which is a more general concept. The set of armored units at the battalion level. ?O1 IS ARMORED-UNIT-MILITARY-SPECIALTY ECHELON-OF-UNIT BATTALION-UNIT-DESIGNATION Removing alternatives specializes an expression. generalization specialization ?O1 IS (ARMORED-UNIT-MILITARY-SPECIALTY or INFANTRY-UNIT-MILITARY-SPECIALTY) ECHELON-OF-UNIT BATTALION-UNIT-DESIGNATION The set including both armored units and infantry units at the battalion level.
Types of generalizations and specializations Operational definition of generalization/specialization Generalization/specialization of two concepts Minimally general generalization of two concepts Maximally general specialization of two concepts Least general generalization of two concepts
Operational definition of generalization Non-operational definition: A concept P is said to be more general than another concept Q if and only if the set of instances represented by P includes the set of instances represented by Q. This definition is not operational because it requires to show that each instance I from a potential infinite set Q is also in the set P. Operational definition: A concept P is said to be more general than another concept Q if and only if Q can be transformed into P by applying a sequence of generalization rules.
Generalization of two concepts Definition: The concept Cg is a generalization of the concepts C1 and C2 if and only if Cg is more general than C1 and Cg is more general than C2. MANEUVER-UNIT-MILITARY-SPECIALTY MANEUVER-UNIT-MILITARY-SPECIALTY is a generalization of ARMORED-UNIT--MILITARY-SPECIALTY and INFANTRY-UNIT--MILITARY-SPECIALTY INFANTRY-UNIT--MILITARY-SPECIALTY ARMORED-UNIT--MILITARY-SPECIALTY Operational definition: The concept Cg is a generalization of the concepts C1 and C2 if and only if both C1 and C2 can be transformed into Cg by applying generalization rules.
Generalization of two concepts: example C1: ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION 10 TYPE OFFENSIVE C2: ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION 5 Generalize 10 to [5 .. 10] Drop “?O1 TYPE OFFENSIVE” Generalize 5 to [5 .. 10] C: ?O1 IS COA-SPECIFICATION-MICROTHEORY TOTAL-NBR-OFFENSIVE-ACTIONS-FOR-MISSION ?N1 ?N1 IS-IN [5 … 10]
Exercise Consider the following two concepts: Indicate different generalization of them.
Specialization of two concepts Definition: The concept Cs is a specialization of the concepts C1 and C2 if and only if Cs is less general than C1 and Cs is less general than C2. MILITARY-MANEUVER MILITARY-ATTACK PENETRATE-MILITARY-TASK is a specialization of MILITARY-MANEUVER and MILITARY-ATTACK PENETRATE-MILITARY-TASK Operational definition: The concept Cs is a specialization of the concepts C1 and C2 if and only if both C1 and C2 can be transformed into Cs by applying specialization rules (or Cs can be transformed in both C1 and C2 by applying generalization rules).
Other definitions The concept G is a minimally general generalization of A and B if and only if G is a generalization of A and B, and G is not more general than any other generalization of A and B. If there is only one minimally general generalization of two concepts A and B, then this generalization is called the least general generalization of A and B. The concept C is a maximally general specialization of two concepts A and B if and only if C is a specialization of A and B and no other specialization of A and B is more general than C.
Exercise Consider the following two concepts and ontology. Indicate four specializations of G1 and G2 (including two maximally general specializations).
Ontology matching Ontology matching allows one to look for instances of complex concepts in the object ontology (i.e. ask questions about the objects in the ontology). Example: Is there a deployable military unit of the red side that performs an intelligence collection military task? Yes, RED-CSOP1 is a deployable military unit of the red side that performs SCREEN1 which is an intelligence collection military task?
Ontology matching: example Is there a deployable military unit of the red side that performs an intelligence collection military task? RED--SIDE MODERN-MILITARY-UNIT--DEPLOYABLE INTELLIGENCE-COLLECTION-MILTARY-TASK IS IS SOVEREIGN-ALLEGIANCE-OF-ORG ?O4 ?O2 IS TASK ?O3 MODERN-MILITARY-UNIT--DEPLOYABLE SUBCLASS-OF MANEUVER-UNIT—MILITARY-SPECIALTY SUBCLASS-OF INFANTRY-UNIT—MILITARY-SPECIALTY SUBCLASS-OF MECHANIZED-INFANTRY-UNIT—MILITARY-SPECIALTY SUBCLASS-OF INTELLIGENCE-COLLECTION-MILTARY-TASK MECHANIZED-INFANTRY-MORTAR-UNIT—MILITARY-SPECIALTY INSTANCE-OF SUBCLASS-OF SOVEREIGN-ALLEGIANCE-OF-ORG RED-CSOP1 RED--SIDE SCREEN-MILITARY-TASK INSTANCE-OF TASK SCREEN1 Yes, RED-CSOP1 is a deployable military unit of the red side that performs SCREEN1 which is an intelligence collection military task?
Rules as ontology-based representations of PSS Condition: IF the task to accomplish is: ASSESS-SECURITY-WRT-COUNTERING-ENEMY-RECONNAISSANCE FOR-COA ?O1 ?O1 is a COA and ?O2 is a deployable military unit of ?O4 (which is the red side) and performs the task ?O3 (which is an intelligence collection military task) Condition: ?O1 IS COA-SPECIFICATION-MICROTHEORY ?O2 IS MODERN-MILITARY-UNIT--DEPLOYABLE SOVEREIGN-ALLEGIANCE-OF-ORG ?O4 TASK ?O3 ?O3 IS INTELLIGENCE-COLLECTION--MILITARY-TASK ?O4 IS RED--SIDE Then accomplish the task: ASSESS-SECURITY-WHEN-ENEMY-RECON-IS-PRESENT FOR-COA ?O1 FOR-UNIT ?O2 FOR-RECON-ACTION ?O3 This rule will be applicable only if the current ontology contains an instance of the complex concept (?O1 ?O2 ?O3 ?O4) represented by the condition. A rule is an ontology-based representation of a problem solving step.
Ontology maintenance Maintaining the consistency of the object ontology is a complex knowledge engineering activity because the object and feature definitions interact in complex ways. Example: Deleting an object concept requires the updating of all the knowledge base elements that refer to it (e.g. the rules that contain it in their conditions; the features that contain it in their ranges or domains; the concepts that inherit its features).
Potential consequence of editing operations: Illustration domain domain f A f A C can no longer have the feature f because it is no longer in the domain of f A A B B f f C C 7 7 Initial State Modified State