In this paper, we examine affordances within the context of representing the functional behaviour of event-based systems. We provide an overview of the ontology that supports such representation. We provide an example use of the affordances that can facilitate us to describe, model, and reason about the behaviour of event-based systems in an intuitive, effective manner.
In this paper, we examine affordances within the context of representing the functional
behavior of event-based systems. The representational facility that we are going to
investigate is called the Event-Based Functional Behavior Ontology (EFBO) [
Affordances and Functionality. Originated in the field of perceptual psychology,
the term affordance refers to an action possibility formed by the relationship between
an agent and its environment [
Wpq = (Xp; Zq) and h = f (p; q); where, h 2= Xp and h 2= Zq.
In this formula above, Wpq represents a system that includes the part of its environment Xp and an agent Zq, where p and q respectively represents the properties of Xp and Zq. The potential affordance h exists between p and q through the relationship f . This formulation suggests that the affordance h exists within the system Wpq where neither the environment Xp nor the agent Zq contain the affordance h.
Based on the key notions of action, events, and change in commonsense reasoning [
Set of classes, C = fE; G; Ig, where, E is the set of Event instances, E=fe1; e2; ::; eng; G is the set of Agent instances, G=fg1; g2; ::; gng; and, I represents the set of Inter f ace instances, I=fi1; i2; ::; ing.
Set of properties, P=fR(e; g); R(e; i); R(e1; e2)g, where, R(e; g) represents the set of relations between an instance of an event, Event(e) and an instance of an agent, Agent(g), e.g., isPerformedBy; R(e; i) represents the set of relations between an Event(e) and an Interface(i), e.g.,hasInterface; and, R(e1; e2) represents the set of relations between two Event instances, e1 and e2; e.g., hasNextEvent.
The EFBO facilitates us to describe the instances of the classes in C in terms of their
associated relations in P. The key classes and the classification of the main properties at
the instances level are presented in Figure 1. The core classes of the EFBO are the Event,
Agent, and the Interface classes. Other EFBO entities are logically based on these classes.
The concept of Event within the EFBO essentially correspond to the notion of SPAN
as endorsed by the Basic Formal Ontology (BFO) [
The temporal dimension of the events within the EFBO can be reasoned through the Event-Event relations as listed in Figure 1. The concept of Agent is understood as an actor that performs some action within a system environment. An Interface is understood as a system entity that serves as an interaction point between an event and an agent. We are only concerned about the temporal aspects of the interfaces through their relations with Events. Figure 2 (left) is a depiction of the Event-Interface-Agent interaction model as supported by the EFBO.
2We adapted the notion of time from the Oxford Dictionaries.
Based on the formulation of affordances in Section 1, a common approach of
implementing affordances is by defining the relationship f in terms of the properties of the agent q
and the environment p [
A. The isPerformedBy Relation. The concept of Action within the EFBO is specified as an event that is performed by an agent. However, an action cannot be performed directly without invoking the interface of its corresponding event. We define the relational property isPerformedBy as the super property of the following chain of properties: hasInterface o interactsWith, where, the hasInterface is an eventProperty to connect an event with an interface; and,the interactsWith is an agentProperty that connects an agent with an interface. The property chain specifies that the affordance isPerformedBy could exist if an Event(e) has an Interface(i) and that interface i interacts with an Agent(g). The event e, in this case, would be inferred as an action that is performed by the agent g. The isInvokedBy relation refers to the super property of the defined relation isPerformedBy. The former is a relation between an event and agent, and the latter is between an action and an agent. The instances that hold these two properties can only be inferred and must not be asserted directly. These relations, therefore, can only be afforded by the chain of relations between the hasInterface and the interactsWith properties.
B. The isTriggeredBy Relation. The concept of Activity within the EFBO is understood to be an Action (i.e., an event performed by an agent) that is followed by a successive event. In order to activate an event, the event must be triggered by an agent through the system’s interface. We define the isTriggeredBy relation as the super property of the following chain of properties: hasPreviousEvent o isPerformedBy, where, the hasPreviousEvent is an eventProperty that connects an event with its previous event; and, isPerformedBy is a defined property that connects an action with an agent as explained before. The property chain specifies that the affordance isTriggeredby can exist if an Event(e2) has a previous Event(e1), and the event e1 is performed by an Agent(g). The event e2, in this case, would be inferred as an event that is triggered by the agent g. The instances that hold the isTriggeredBY properties can only be inferred and must not be asserted directly. In other words, the isTriggeredBy can only be afforded by the chain of relation between the hasPreviousEvent and the isPerformedBy properties.
As we can observe, the affordances within the EFBO facilitates us to have a straightforward mechanism to specify different relationships between the properties of an agent and its system environment. We can also observe that the EFBO allows nesting of affordances, as well as defining affordances in terms of the other affordances. This can be a powerful mechanism in order to have different levels of abstractions of the potential actions by an agent within the event-based systems.
Example Use of the EFBO’s Affordances. As an example, we observe the
EFBObased modelling of a typical login functionality for a smart phone application [
Based on a set of functional reasoning categories, e.g., events by triggering agents, distribution of activities by agents, events within a specific activity, and so forth, a set of instance-specific classes (the class names prefixed with underscores in Figure 4) were defined under the ‘Entity Classifier Class’ of the EFBO. For example, the class ‘ User Agent Activity’ was defined as equivalent to any Action that isPerformedBy some userAgent. Similarly, the class ‘ Event Triggered By Client Agent’ was defined as equivalent to any Event that isTriggeredBy some clientAgent. Figure 4 presents the inferred classification results after the automated reasoning on the classifier classes. It should be observed from the modelling example that the EFBO’s affordance relations allowed the reasoner to infer the set of potential actions between the system’s agents and its environment. For example, the reasoner perfectly inferred the actions that could be performed by different agents as part of their activities within the modelled system; the reasoner could also infer the events that could be triggered by each of the agents as well (see the first two columns in Figure 4). It should be noted that while we did not have the EFBO affordances directly asserted between any of the storyboard entities, the affordance relations were inferred through the automated reasoning as afforded by the relations between the properties of the events and the properties of the agents within the modelled environment. This latter strategy is critical in both observing and modelling the functional behavior of any practical application system. In the practical sense of affordances, the agents in such system must not be able to invoke the functional events of the system without the proper interactions with the system’s designated interfaces. This design choice also promotes more automation, and, ultimately, reduces the amount of human-induced errors during the modelling process. As observed in this section, the relations of affordances can be a powerful feature for the ontologies that deals with representing the functional behaviour of event-based systems.
Using the EBFO representation along with its affordances, one can effectively model the functional entities of a system in such a way so that their existence within the system can be thoroughly reasoned. It should be noted that currently, the temporal aspects of the EFBO events only supports a subset of Allen’s temporal axioms that are necessary for our functional reasoning tasks. However, the EFBO is flexible enough to incorporate the remaining axioms, if necessary in the future. Due to its affordances, the EFBO-based functional modelling process is quite intuitive and practical, which allows us to describe and reason about the behavior of a system through a series of simple statements. We have carefully engineered the EFBO ontology with a minimal set of required classes and properties in order to achieve the maximum usability of the ontology without compromising the required logical rigour to achieve an effective functional reasoning mechanism.
This work is supported by the Natural Sciences and Engineering Research Council of Canada (NSERC).