An Ontological View
In [Fig. 1] we show an ontological view of our collaborative situation-aware scheme. Here, base concepts are enclosed in gray oval shapes and are connected by properties, represented with directed edges in the figure. The core properties are User moves into the Environment and User is in a Situation. As these properties cannot be directly sensed (i.e., instantiated) by the system, they are shown with a dotted edge, as abstract properties. Indeed, the overall system is aimed at indirectly discovering them, by observing the collective behavior of the users starting from data provided by personal devices.
Let us consider a set of U Users. We assume that each user owns a Device able to provide current Location and Time (e.g., a smart phone equipped with clock and GPS reader), as well as a number of Services. A User Agent (UA) recommends a subset of such Services observing the current user Situation inferred by a Situation Agent (SA). The UA is based on an ontology which allows connecting situations to tasks, and then to specific services.
We will not further discuss how the UA works since the main aim of this thesis is to show how a situation can be recognized by using an emergent paradigm rather than discussing an overall service recommender.
The information that allows to identify the occurrence of a specific situation may be indirect (such as implicit inputs or context descriptions) and, in general, uncertain and imprecise. As a consequence, the situation occurrence is determined with a certainty degree. This inferential process is based on fuzzy logic, and is shown in italic bold style in figure. Due to the intrinsic vagueness of fuzzy inference, it is possible to recognize more than a current Situation with a related degree of certainty. As a set of specific situations we considered four key typical situations related to social events (collaboration in the following) namely, pre-collaboration (PreC), on-going collaboration (OngC), post-collaboration (PstC), and collaboration pause (PauC). In the figure, edges with white arrow head, and white oval shapes represent the classical inheritance and a specialized concept, respectively.
The SA infers situations by observing Events. The situation inference process exploits fuzzy if-then rules fined on the certainty degrees with which the events are detected. There are two types of events: Grouping and Disjoining events. The grouping event arises when two or more users tend to be close to each other. The disjoining event occurs when a user separates from a formed group of users. The detection of the two events performed by, respectively, two types of Event Agents (EA): the Grouping Agent (GA) and the Disjoining Agent (DA). A unique GA is associated with the group, while a DA is connected to each user belonging to e group. The EAs observe the intensities of the Marks and exploit these intensities for associating a certainty degree to the event by a fuzzy granulation process. Marks are located in the Environment, and are produced by Marking Agents (MAs). Each user is associated with an MA which observes Time and Location her/his Device in order to produce marks. Whenever the MA leaves its mark in the environment, it receives from the Environment the information whether the mark is going to be superimposed on marks left other MAs. If this occurs, the MA generates one Grouping Agent (GA) and one Disjoining Agent (DA). Possibly, GAs corresponding to the same group of users are fused. In the next section, we will discuss in detail the marking process, the fuzzy granulation process and the situation inference process, which are the core processes of, respectively, the MAs, the EAs and the SAs.