PSI Upper-Level ontology
The Role of the PSI Upper-Level Ontology
PSI Upper-Level ontology  is the upper level theory for the set of the Core ontologies of the PSI Suite of Ontologies . Its main function is putting the components of the Suite in line with the commonly accepted metaphysical and cognitive framework of the common sense represented by several reference ontologies like Suggested Upper Merged Ontology (SUMO), Descriptive Ontology for Linguistic and Cognitive Engineering (DOLCE) and highly reputable linguistic resources like WordNet Linguistic Ontology (WordNet). One more objective of introducing the upper-level of the Suite is providing semantic bridges to mainstream enterprise, business, and process modeling frameworks like the Enterprise Ontology (EO), Toronto Virtual Enterprise (TOVE) Ontology, Process Specification Language (PSL).
In difference to the mentioned enterprise, business, and process modeling frameworks, which are to a certain extent Domain independent (TOVE, PSL) or model manufacturing Domain (EO), PSI Upper-Level ontology defines an upper-level theory for the Domain of Engineering Design Processes and Environments in Microelectronics and Integrated Circuits.
As many foundational ontologies the PSI Upper-Level ontology has a clear cognitive orientation in the sense that it does not pretend being strictly and rigorously referential to the theories describing nature. Instead, it captures ontological categories and contexts based on human common sense reflecting socially dominant views on the Domain – characteristic at least to engineering design professionals. As such, the categories introduced in the PSI Upper-Level ontology are not related to the intrinsic nature of the world but are rather thought of as “cognitive artifacts ultimately depending on human perception, cultural imprints and social conventions” (c.f. ). Therefore, these categories assist in making already formed conceptualizations of the PSI Suite of Ontologies explicit and referenced by the common sense. PSI Upper-Level ontology also plays an integration and harmonization role of a foundational ontology because it represents a rather domain-independent descriptive theory based on formal principles for harmonizing and integrating the underlying domain dependent modules with other relevant ontologies.
In contrast to foundational ontologies the PSI Upper-Level ontology is not foundational in the sense that it is not a profound and a complete theory in philosophical or, more precisely, cognitivistic sense. For example the PSI Upper-Level ontology does not deal with many problems that are topical for foundational theories like:
- Differences between abstract and concrete oblects
- Particulars and universals
- Spatio-temporal co-localization of things
- Mereological axiomatization, etc
It also does not provide rich axiomatic sets for rigorously describing the semantics of the contained entities. Instead, other highly reputable foundational ontologies are used as reference sources for defining PSI Upper-Level ontology components. The mappings of these components to those reference sources are explicitly specified.
PSI Upper-Level ontology v.2.3 commits to the ontological choices of DOLCE and therefore is descriptive, possibilistic, multiplicative, and perduranistic. As far as the upper level of DOLCE taxonomy is reused, the PSI Upper-Level ontology is, as DOLCE, the ontology of particulars.
1. Ermolayev, V., Jentzsch, E., Keberle, N., and Sohnius, R.: Performance Simulation Initiative. Upper-Level Ontology v.2.3. Reference Specification. Technical Report PSI-ONTO-TR-2009-2, 06.10.2009, VCAD EMEA Cadence Design Systems, GmbH, 75 p.
2. Ermolayev, V., Jentzsch, E., Keberle, N., and Sohnius, R.: Performance Simulation Initiative. The Suite of Ontologies v.2.3. Reference Specification. Technical Report PSI-ONTO-TR-2009-1, 23.09.2009, VCAD EMEA Cadence Design Systems, GmbH, 173 p.
OWL Code of the PSI Upper-Level Ontology v.2.3
Could be acquired from Media:PSI-ULO-v.2.3.pdf
The Taxonomy of the PSI Upper-Level Ontology
The Overview of the PSI Upper-Level Ontology
The PSI Upper-Level Ontology is the upper-level formal representation of the model of stateful creative dynamic processes, pro-active agents, and objects situated in nested dynamic environments based on the formal representation of time, events, and happenings. Apart of the PSI project it may be used as an upper-level theory for developing, aligning, harmonizing ontologies in different application domains that have common features. The PSI Upper Level Ontology is designed as a semantic bridge facilitating to mapping the PSI Core ontologies to abstract ontological foundations and common sense. It is also used as semantic “glue” for bridging the PSI ontologies with other theories, widely accepted in the domains where processes, states, and participating objects are the major entities. These mappings and semantic bridges are supposed to ease further commitment of potential users to our Suite.
The overview of the ontology is given by the presentation of the semantic contexts of its key concepts: a Process, an AtomicAction, an Environment, a State, an Object, an Agent, a Rule, a Characteristic, a Value, and a Measure.
PSI approach to domain modeling could be qualified as “environmentalistic”. Indeed, it is considered in PSI that:
- Design systems – the aggregations of the objects of various kinds like designers, tools, resources, design methodologies – belong to their environments, have their environments, and actually form the environments for design processes
- Design processes – the aggregations of lower-level process chunks and atomic actions wrapped by process chunks – have their environments and therefore flow through design systems
Another important aspect of the PSI model is that an environment is changed in design processes. These changes are brought into effect by the agents that execute atomic actions applied to different constituents of the environment. Atomic chunks of change are therefore actualized as results of atomic actions and in those actions that bring the affected environment to a different state. Hense, a design process is considered a transformation path connecting several states of the environment of this design process. It is specified in the PSI modeling framework that states could be requirement- and quality-sensitive. Such a sensitivity is the mechanism for checking if the requirements (a quality requirement for instance) to the transformed objects in the environment are met. The requirements are formulated with respect to the characteristics of the objects in question.
Yet another important domain feature in PSI is that not all possible paths in the state space of an environment are allowed or could be considered correct. Correct ones are suggested by the design technology that is used in the house and are represented as process patterns in the descriptive model. These process patterns are built using the patterns of the states to be connected, the patterns of the atomic actions that could be executed after a particular state is reached. These actions could be executed by the agents that possess specific capabilities – i.e. are capable of following particular behavior patterns.
A Process, an AtomicAction, and an Environment
A Process is a specialization of an Event that is stateful and possesses pro-active character. A Process has its Environment – the part of the world which may influence the course of the Process or may be changed in the course of the Process. A Process is pro-actively directed by the Agent who manages it. Pro-activeness of the Agent is understood in the sense that the Agent pursues a particular Goal in the managed Process. This Goal is the State of the Environment, which the Agent desires to reach. It should also be mentioned that the change in the Environment is not produced by the Process, but by the entities who act in this process – those Agents who execute AtomicActions wrapped by the Process. In general, it is considered that changes may only be applied by Agents through execution of AtomicActions. For example, it is wrong to say that a multimedia controller layout has been designed by the process of logical design. In fact the appearance of the layout for the multimedia controller in a certain state of the Environment has been achieved by the team of Agents who executed a particular sequence of AtomicActions. By that the Agents applied the sequence of particular changes to the Environment and guided the environment through the sequence of States towards the Goal. Processes in an engineering Environment can not connect any arbitrary State to any other arbitrary State because it is senseless with respect to the technology or the methodology. Some sequences of States may therefore be withdrawn from the engineering design routine while some other sequences may be suggested or prescribed by an industrial standard or a company policy. These prescriptions are ProcessPatterns. From the point of view of an Agent a Process could be relevant to a particular working Context as well as it may have its Context.
The model of an Environment is further refined in the PSI Core Environment, Event, and Happening ontology (E2H). ProcessPatterns are further elaborated as GenericTasks, BehaviorPatterns – as Roles, an ActionPattern – as a GenericActivity in the PSI Core Process Pattern ontology. The concepts of a Process and an AtomicAction are further refined in the PSI Core Process ontology.
The State of an Environment
Any Process as a pro-active stateful manifestation of a change in the Environment is managed by an Agent for reaching the State of affairs in which the constituents of the Environment possess the properties partially or fully matching the Goal of that Agent. It is considered that a Process has reached its target State if such a state of affairs is reached. Otherwise the Process fails to reach its target state. A Goal, if complex, can be decomposed to simpler partial Goals as often done in problem solving. Such partial Goals are in fact the states of affairs that should be reached before the overall compound Goal can be attacked. States are the configurations of the constituents of an Environment. It is considered that a State is reached when the Characteristics of the constituents of the Environment have Values in the ranges satisfactory matching the corresponding Goal or partial Goal of an Agent. In engineering design the mentioned goals are technologically controlled. For example, a technology of digital front-end design in microelectronics and integrated circuits prescribes that an overall goal of a digital back-end design is the development of a design artefact in the GDSII layout representation. At the same time the technology suggests that the netlist, floorplan, placement and routing representations should be developed before the overall goal can be reached. In these settings the States can be seen as technological milestones on the path through the problem solution space leading to the overall goal. The requirements to the ranges of the property values of the constituents of the environment are denoted by StatePatterns. StatePatterns are controlled by the Policy of a company that are normally be based on the standards of the particular industrial sector. Goals and corresponding partial goals may be pursued by taking different alternative paths going through different States. If a problem solution space is represented as a directed graph, a State may have several alternative outgoing edges. These edges correspond to different admissible AtomicActions applying different changes to the Environment. A Decision on the choice of an admissible AtomicAction should be taken for choosing the continuation of the path at any State. In particular, a Decision in the target state chooses among the alternative to terminate the process in success and the alternative to refine the values of the properties of the constituents of the environment heading to the same target state. Hence, a Decision is a specific AtomicAction which applies changes to the Environment indirectly – by choosing the alternative on the solution path. A Decision is also the mechanism for altering the course of the Process when the Goal or the sub-Goals are dynamically changed. In difference to an Environment, which is a DOLCE: Perdurant, a State is a DOLCE: Endurant because all its parts should be present at any instant of time of the presence of a State.
An Object, an Environment, and the Change
An Object is a Holon that may be changed by an Agent. An Object has its Environment and belongs to an Environment as a part – is situated in the Environment by other words. An Object may be changed in the course of an AtomicAction executed by an Agent. An Object may have Characteristics. The relationship between an Object and a Characteristic is further refined in the PSI Core ontologies by the relationships between the subclasses of and Object from one side and the subclasses of a Characteristic from the other side. An Object could be either material or immaterial. MaterialObjects are those Objects which are physically or legally substantial in the sense that they possess tangible physical non-temporal properties like mass, color, shape, size, speed, usage right and can not be copied or duplicated without borrowing a definite amount of physical or legal substance for it. The law of conservation of matter is applicable to material objects. MaterialObject subclasses are an Agent, a MaterialArtifact, and a ConsumableResource. In difference to a ConsumableResource that is always material, a NonConsumableResource subsumes directly to an Object because it could be either material or immaterial. ImmaterialObjects in contrast to material ones are not substantial in a physical or legal sense. Hence, they can be copied or duplicated without consuming physical or legal substance. ImmaterialObject subclasses are an ImmaterialArtifact, a Rule, a Fact, a Belief. From the point of view of an Agent an Object could be relevant to a particular working Context as well as it may have its Context.
The model of an Object is further refined for its subclasses in several PSI Core ontologies. For instance the model of an Agent for individuals is refined in the PSI Core Actor ontology and for teams – in the PSI Core Organization ontology. The representations of different sorts of resources and Tools are further elaborated in the PSI Extension ontologies. The model of a Context is further elaborated in the PSI Core Environment, Event, and Happening ontology.
An Agent is a MaterialObject that possesses pro-activity, is able to execute AtomicActions and to manage Processes. The pro-activity of an Agent is revealed in pursuing the Goals of changing an Environment to a desired State. An Agent is the only entity which can change an Environment by executing AtomicActions applied to the Objects in the Environment. An Agent has Beliefs about the Environment(s) that are the hypotheses believed to be true. These Beliefs may further become Facts if confirmed by the happenings perceived by the observers. Beliefs together with desires and intentions are important basic elements forming the behaviour of an Agent. This behaviour is regulated by BehaviorPatterns specified as Rules. An Agent is an abstract entity which is a generic model for an individual person (a manager, a designer), a group of persons or artificial agents acting on behalf of physical persons (a team or an organizational unit), or an external pro-active entity influencing the Environment of an observed Process in a definite way.
A Rule is an ImmaterialObject which is a principle, a condition, a procedure, a generic pattern, or a norm shaping out a possible Process, AtomicAction, behaviour, or State. A Rule may be an atomic proposition or a more complex composition of other Rules due to the inherited structural parthood relationship to self. As far as a Rule is a DOLCE: Endurant no temporal parthood relationships are allowed for its proper parts – the composition of a rule can not be changed in time. A Rule itself still has a temporal property of validity – it is valid within a particular TimeInterval or several TimeIntervals. In PSI a Pattern, a Policy, and a Metric subsume to a Rule.
Pattern models are further elaborated in the PSI Core Process Pattern ontology. The refined model of a Policy is the part of the PSI Core Organization ontology. The model of time is further elaborated in the PSI Core Time ontology.
A Characteristic, a Value, and a Measure
A Characteristic is a distinguishing property that has Values and may be checked by one or more Requirements. A Characteristic is not an attribute of a concept because the latter inhers to a particular concept, but the former may have different semantic scents for different concepts. For example, a Dependency (a subclass of a Characteristic) may mean a particular form of co-execution for AtomicActions, a causal relationship for Events, or a commitment for Agents. Associations of a Characteristic with the other concepts of the upper-level model are not defined due to this very reason. These relationships may have different semantics for different counterparts or even in different ontological contexts. Therefore, these associations are defined lower on in the knowledge hierarchy of the PSI Suite of Ontologies – either in the Core or in the Extension ontologies where more appropriate. The following specializations of a Characteristic are defined at the upper-level because of their generic character and multiple semantic scents.
A Dependency is a Characteristic that defines the qualities of a relationship among different entities of the same type. For example, we shall say that: one Event depends on the other Event if the latter causes the occurrence of the former; one AtomicAction depends on the other AtomicAction if the former can not be executed without the latter; one Agent depends on the other Agent if the former committed to execute an AtomicAction in the process managed by the latter and so on.
A Location is a Characteristic that defines spatial or affiliation properties of an Object or a Process. A Process may be located within one organization, but also may involve several organizational entities. A Policy may have a local sphere of power within a team, but also may have a more widespread character – a National regulatory policy is obligatory for all organizations in a country, etc.
A Value is the mapping of an individual Characteristic to the corresponding “quality space” pointing to the position of this individual Characteristic in this space. A Value is semantically equivalent to DOLCE’s concept of a Quale. For example, the Value of an Agent’s Characteristic of utility may be the quantity of the Units of Welfare collected by the individual Agent in his life time. As a Value maps a particular Characteristic to a specific value space, we are interested in those Characteristics that may have Values. For a Characteristic “having a Value” may mean that:
- This value is measured
- or This value is assigned
- or This value is computed (e.g. inferred)
The models of a Characteristic and its upper-level subclasses are further elaborated in the PSI Core Design Artifact Complexity and Quality ontology (DASpecific), PSI Core Design Artifact ontology, PSI Core Process Pattern ontology, PSI Core Actor ontology, and in the PSI Extension Ability ontology.