A Formal Metamodel for Context Design: from Identification to Modeling and Analysis to Integration in Intelligent Systems
Context variability influences the requirements to be satisfied in a system, their alternatives and the quality of each alternative. This drives a variability in system behaviors to keep fulfilling the requirements and accommodating the varying context. The integration of context variability into the system will allow for the creation of intelligent applications. However, an approach that formally associates the software variability with contextual information is still missing. In this paper, we propose a unified approach that facilitates a formal treatment and integration of context variability in the requirements and operational levels. [the advantage of our approach is to be demonstrated with its applicability & usage in model-driven engineering as a promising software development paradigm]
Context influences the requirements to be satisfied in a system, their alternatives and the quality of each alternative. This drives a variability in system behaviors to keep fulfilling the requirements and accommodating the varying context. Hence, identifying and representing context information will allow for the creation of intelligent applications. However, an approach that formally conceptualizes contextual variability for supporting the system adaptability is still missing. In this paper, we propose an approach that facilitates a formal treatment of context and relationshipswith system goals and operations. We apply this approach on a smart conference scenario.
A formal metamodel for context modeling and analysis in intelligent systems
In order to operate in complex environments with a highly varying context, the software systems demand smart engineering that would allow for the provision of intelligent interactions with users and of more appropriate services and information. Context is a key element to the engineering of intelligent systems. Hence, several approaches consider context in the system being developed from different perspectives. However, a methodology that formalizes and integrates the main notions and concepts related to, and influenced by context in a unified framework is still missing. In this paper, we propose a formal metamodel for context modeling and analysis. The architecture of our modeling framework is based on four levels of abstractions, and aligned with the OMG Meta-Object Facility (MOF) specification. A formal method is used to define the abstract syntax and the static semantics of the metamodel.
Advanced Information and Computing Technologies (ICT) have led to emerging systems and took the computing paradigms out of the standalone and desktop. Context-aware systems, Ambient Intelligence (AmI), context-sensitive systems, adaptive systems, ubiquitous systems, cyber-physical systems and Internet of Things (IoT) bring miniaturized technologies closer to the user’s daily activities [ref#]. From a broad perspective, the intelligence is a common ground between such systems. An Intelligent system is capable of (i) knowing current user’s needs (e.g., a goal, a task being performed), (ii) knowing contextual changes (e.g., locations, environmental conditions) and (iii) responding appropriately at the information and the service levels, and in real-time (e.g., by executing an adaptive action or notifying the user). To do this, the system should observe and reason about the context that is pertinent to the user’s current interests. Here, context is any information that characterizes the situation of an entity that is relevant to the interaction between a user and an application (e.g., place, object, person, device, user and application themselves) [Dey]. To this end, context is a key element should be considered when developing intelligent applications.
At runtime, a varying context is main factor contributing to the variability of the system [1], to unpredictability, uncertainty, and weak controllability, thus causing a failure in meeting users’ needs and software requirements [ref#]. To deal with these challenges, several researchers have made contributions in the endeavor of supporting the design and engineering of intelligent systems and applications considering context at design time and runtime, and across the different phases of the software engineering process, namely requirements engineering, design, implementation, testing, deployment, maintenance and feedback (for a survey see [2,3,4]). Other researchers developed techniques for modelling and reasoning about context information (for a survey see [5]). Also, several approaches address the dynamicity and context particularity when developing the modern generation of software systems (for a taxonomy see [6]).
Requirements Engineering (RE), as a preliminary phase in software projects, concentrates on investigating and precisely specifying the problem world that a machine-based solution is intended to improve [Requirements Engineering: a roadmap, Bashar Nuseibeh, Steve Easterbrook] [7]. The scope of investigation includes problems, opportunities and domain knowledge in the system-as-is (i.e. the system as it exists before developing the machine), and objectives, assumptions, services, constraints, and responsibility assignments in the system-to-be (i.e. the software to be developed and its environment such as people, physical devices, and pre-existing software) [7]. In Goal-Oriented Requirements Engineering (GORE), unlike the traditional RE approaches, the concept of goal plays a central role in the elicitation, modeling, analysis, evaluation and evolution of system and software requirements, where a goal is an objective that the system should achieve through the cooperation of system agents [8]. An agent is an active component playing a specific role in goal satisfaction. The agents may restrict their behavior to ensure such a role by monitoring and controlling system items [8].
KAOS, “Keep All Objectives Satisfied” or “Knowledge Acquisition in autOmated Specification”, was firstly proposed by Dardenne and Van Lamsweerde as a GORE language associated with a set of formal analysis techniques [9,10]. Alternatively, in his PhD thesis Eric Yu proposed the i* language [11] for modeling the social aspects of requirements. Tropos [12] project was proposed by Mylopolos. NFR [13] was firstly proposed by Lurance chang as qualitative analysis framework for modeling and analysis of non-functional requirements.
In intelligent environments, system requirements overlap with context information which is drawn from domain knowledge. The demarcation of the boundary between the system and context in a particular domain is crucial for determining what to be considered as context. To do this, we have to define (1) the relation between domain knowledge and the structural part of context (( i.e., extending domain engineering/analysis with a context theory [])), and (2) the mutual influence relation between the dynamic context and system aspects including user’s intentions and software functionalities (( i.e., exploring together the problem variability (by the users and the environment of the software) and the variability of software (i.e., functional and quality requirements, and operations [])), [14] , [Modeling domain variability in requirements engineering with contexts, ER, 2009], [Towards a computer-aided problem-oriented variability requirements engineering method, CAiSE 2014 workshops], [Exploring the dimensions of variability: a requirements engineering perspective, Sotirios Liaskos// OR// On Goal-based variability Acquisition and analysis, Sotirios Liaskos], [A framework for combining problem frames and goal models to support context analysis during requirements engineering, CD-ARES 2013], [Situ: a situation-theoretic approach to context-aware service evolution], [Analyzing monitoring and switching problems for adaptive systems, Mohamad Salifu, the journal of systems and software, 2012]
Several researchers consider context for engineering computer systems from different perspectives. Problem Frame [PFx] (PF) specifies variations of a class of software development problems in terms of biddable domains and lexical domains (i.e., the world/W), a requirement R, a machine domain (i.e., the specification/S) and interfaces/connections between the domains. To keep the requirement satisfied, each variant represents an alternative specification which corresponds to a particular context [PFx]. For context modeling and analysis. CODA [15] is context-oriented domain analysis method for capturing contextual information, variation points and context-dependent adaptations. In his PhD thesis [16], Ali proposed an extension to Tropos models by weaving context variability with variation points on the goal level.
In [Context-driven requirements analysis, Jongmyung Choi, 2007] the author proposed context-aware artifacts such as use case and context-switch diagrams. Elicitation techniques for identifying contextual requirement at design time [Eliciting contextual requirements at design time: a case study, Alessia Knauss, Daniela Damian, Kurt Schneider, EmpiRE, 2014]. [Situation-oriented requirements elicitation, Nimanthi L. Atukorala, 2016]
To the best of our knowledge, an approach that formally conceptualizes the dynamic context and its associated concepts centered on user’s interests, and systematically facilitates a formal treatment of context variability in the intentional and functional levels of a system is still missing.
Our research aims at proposing such an approach with the motivation of building a unified framework for identifying and representing contextual information in order to ensure formal and precise specifications that pave the way for smart engineering of software for intelligent systems, and the guidance of five main research questions: (1) How can we identify relevant context in both the user’s world and the software-to-be? (2) How can we unify and formalize the identification process? (3) How can we formally contextualize a system functionality? (4) How can we systematically formalize the dynamic context and its variability including diverse dimensions, sources, and parameters like quality, validity, certainty, volatility, and access types? (5) How can we systematically facilitate the adopting of context-enabled software engineering? the formalized concepts in context modeling and analysis for?
This paper aims to unify the formalization of context and requirements modeling and analysis languages featuring context, entity, goal, operation and agent as main concepts.
Decide what RE language is used, what context approaches are used
RQ1 = identify? relevant context /// Goal-driven and Entity-centered Identification + relevance + relevance level
Goal-driven, entity-centered context identification method: Context is any information that is (1) relevant either to a specific human goal (resp. a system/software requirement) that designates the current interest from the perspective of the user (resp. the system), or to a specific environmental operation (resp. a software operation) that designates how to reach and achieve the current interest in the user’s world (resp. the software system) and (2) characterized and formalizable with respect to an entity of interest.
RQ2 = formalize? and unify? identification///
Introduce the concept of focal element.
Definition. 1. (Focal Element). An interest, capability, constraint or service is a focal element FC to an agent Ag iff Ag directs the most attention on achieving, undertaking or executing FC.
Definition. 1. (Focal Element). A goal or operation is a focal element FC to an agent Ag iff Ag directs the most attention on achieving or executing FC.
Context is any information that is relevant to a focal element on which the most user’s attention is that is a specific goal or operation which encapsulate the current most.
Definition. 2. (Context). Any information that is relevant to a focal element FC and Formalizable in terms of any of interest.
both the user’s world with respect to his interests, and the software-to-be with respect to execution of software services.
= model and represent? identified context /// contextualization
=
Introducing the focal element:
A user’s interest might be in a non-communicative form like a cognitive goal. In the light of the perceived context, the user (re)acts in order to realize his interest which might be evolved to a task undertaken within the environment and/or an interaction with the system. Indeed, the interest takes different forms that are influenced by diverse dimensions of context ranging from environmental conditions, preferences, available technologies, temporal events and so forth. On the other hand, the system dynamically adapts its behavior regarding both the user’s current interest and the current state of context. This requires the acquisition and augmentation of contextual information.
RQ3– In other words, how can we formalize the contextualization of a system at the goal and operation levels? //LF challenge
A context-sensitive system encompasses two types of variability:
A system that is developed to operate in intelligent environments, in contrast to traditional ones, centers its functionality on users’ needs in order to provide smart services, interactions and information taking into account its varying context. Researchers are working to deal with complexities caused by contextual variability (e.g., locations, profiles, conditions, technologies etc.). Several approaches have been proposed to associate the system functionality with context. This section introduces the concept of context and presents different approaches for context modeling and analysis.
There is a lack of consensus about the definition of context. one single, unified and multi-disciplinary definition for context. Context is a multi-defined concept in many disciplines. In computer science, many researches introduced different perspectives on understanding and using context. Dey defined context as any information that characterize the situation of an entity (e.g., a person, a place or an object) []. One of the wide accepted definition is
Regarding intelligent systems, context-awareness.
| Works | Definitions and Perspectives |
| Brezillon and Pomerol, 1999 | Context is relevant to the focus that is a step in solving problems, making decisions or performing tasks []. |
| Dey, 2000 | Context is any information that can be used to characterize the situation of an entity. |
| Dourish, 2004 | Representational & interactional views [] |
| Bradley, 2005 | A multidisciplinary model of context for representing contextual dimensions of both the user’s world (i.e., human actions and goals) and the application’s world (i.e., software services) []. |
| Zimmermann et al., 2007 | An operation definition of context |
| Vieira et al., 2007 | Extending the notion of focus to include a task and an actor together with a role []. |
| Ali et al, 2011 | Context is a partial state of the world that is relevant to an actor ’goals [14]. |
Tab. 1. Definition and perspectives on Context
We discuss in this section how researches (RE approaches, design approaches, analysis approaches) take context (internal system states, external environment and users states) in their methods for engineering software systems.
We propose the following criteria for discussing and comparing the approaches:
| Works | Considerations |
| Jackson, 2000 | Problem frames characterize variations of a class of software development problems in terms of contexts W (i.e., biddable domains and lexical domains), a requirement R, the specification S (i.e., machine domains) and interfaces (i.e., connections between domains) []. Each variant represents an alternative specification which corresponds to a particular context []. |
| Ali et al, 2011 | Contextual requirements weave context variability with variation points at the goal level [x]. |
| REUBI | [A requirements engineering method for ubiquitous systems] |
| [ Using Activity theory to identify relevant context parameters] | |
| [ Managing Dynamic context to optimize smart interaction and services |
Tab. 1. The relationship between the system and context
Contextual goal model
Problem frames
REUBI
RELAX
{System and Context are in a persistent state of mutual influence}
There is a persistent state of mutual influence between system aspects and the context.
Give an informal introduction to the notion of tracing requirements volatility to changes on contexts elements, and classifying the rationale behind the evolution/volatility of requirements.
(statement #) We state our findings on this section.
Context intervenes with and propagates its influence to all system items in their traceability link that originates form human goals to system requirements to software services.
Give an introduction to the notion of tracing context changes to adoptions of specific systems alternatives and classify the rationale behind the variability of context elements.
(statement #) We state our findings on this section. CVSt
Context is a partial state of the world that is relevant to an actor ’goals.
To the best of our knowledge, a formal and unified approach that integrates the concept of context and its related concepts into the intentional, requirements and functional views of a system is still missing. Note: It would be better if we put this statements in the introduction
The proposed metamodel is based on the contextual classifications and dimensions of domain knowledge in context-sensitive, context-aware and ubiquitous systems engineering [], and uses the dimensional information for modeling and analysis in order to (1) identify contexts relevant to a certain goal or a required functionality, and (2) evaluate the types and levels of impact from contextual factors on functional requirements and software operations in a system. We use the term “Context Design (CD)” to refer to the activities for identifying, analyzing, representing, reasoning about and embodying (i.e., integrating) context information in a system at both the design time and runtime. A “framed context” refers to a formal specification of contextual information as a mean to assist the engineering of a system.
Based on concepts and notions introduced in research endeavors for context framing, we have developed a formal metamodel for engineering context in intelligent systems. The main analysis and design aims of such metamodel are to support and provide: (1) Un explicit and formal basis for understanding the concept of context, (2) A unified framework in which conceptual abstractions related to context are defined and interrelated, (3) Context enrichment modeling (i.e., artifacts furnished with context), (4) A ground for integrating context-enabled requirements analysis into other phases across the software development lifecycle, (5) A basis for toot support, and (6) A logical schema of the specifications database in which specifications (e.g., semi-formal and formal structures) can be stored and queried for validation and analysis (e.g., structural consistency and competence checks). and analyzing the syntax of the context models and the corresponding context-enriched artifacts in order to check their correctness with respect to inter-view consistency rules. (e.g., constraints between the intentional view and the contextual view).
We propose a metamodel-based framework (MICoIS) for Modeling and Integrating Context within Intelligent Systems.
A meta-model is a model that rises the abstraction level for system engineers including analysts, designers and developers. A metamodel defines a set of high-level conceptual abstractions in terms of which other specifications are constructed. MDA OMG 2003 the meta-modeling techniques (i.e., the abstraction, classification, generalization, specialization, etc.) allow to build, reuse and transform source models into different target implementations. In other words, to design one platform-independent model, and build it on diverse platforms (e.g., Java/EJB, C#/.NET and XML/SOAP). The Meta Object Facility (MOF) standard [x], recommended by the Object Management Group (OMG), provides a metamodeling architecture based on four layers; (i) Meta-metamodel (M3) defines a language for specifying metamodels, (ii) Metamodel (M2) define a more elaborated language for specifying models, (iii) Model level (M1) defines a language for describing information domains and systems, and (iv) Instance level (M0) defines a specific information domain. The architecture of our modeling framework is based on four levels of abstractions and aligned with MOF as follows (see figure 1).
Fig. 1. TheMOF-aligned modeling architecture: the meta-meta, meta, domain and instance levels
This level represents the meta-language within which the highest-level abstractions (i.e., language constructs) are defined for structuring other language metamodels (e.g., a domain-specific language (DSL) metamodel, UML metamodel). MOF meta-meta-model is composed of object-oriented constructs; MetaClass, MetaAttribute, Meta-Reference, Inheritance, and Composition. In our modeling architecture, M3 is aligned with MOF, and is made of the Meta-Concept, Meta-Relationship, Meta-Attribute and Meta-Constraint.
This level represents the language within which domain-independent abstractions are defined for structuring formal specifications of the multiple views and aspects of a system. contextual, structural, intentional, functional and behavioral views of a system. It is composed of meta-concepts(e.g., Context, Entity, Goal, Operation, Agent), meta-relationships linking meta-concepts (e.g., Characterization, Contextualization, Responsibility), meta-attributes of meta-concepts or meta-relationships (e.g., Definition of Context, Type of Contextualization), and meta-constraints on meta-concepts and meta-relationships (e.g., ‘A Context Contextualizing a Goal must has a Entity)
This level represents concepts specific to the modeled domain or system, for example, projector, light, meeting room in a system for smart meeting rooms. It is composed of concepts that are instances of meta-level abstractions. For example, the HostMeeting concept in Figure 1 is an instance of the Context meta-concept; the MeetingRoom concept is an instance of the Entity meta-concept; the concept Achieve [LightsSwitchedOn] is an instance of the Goal meta-concept.
Domain-specific concepts are linked through instances of the meta-relationships linking the corresponding meta-concepts of which they are an instance, for example HostMeeting Contextualizes Achieve [LightsSwitchedOn]; HostMeeting Characterizes MeetingRoom. Domain-specific concepts must satisfy instantiations of the meta-constraints on the corresponding meta-concepts of which they are an instance, for example, …. A model of a system view is structured from domain-specific concepts according to instances of the corresponding meta-relationships inherited from the meta level.
This level represents specific instances of domain-specific concepts in the running application (see figure 1). This level is composed of sensed data and acquired information as well as concrete instances of observed entities, places, devices, people, software and hardware components.
As we have seen before, M3 is made of meta-meta-objects to construct the meta level. Thus, a formal method that support both the object-orientation and the reuse of constructs is appropriate to formalize the context and the associated concepts. We use a UML class diagram [x], and the formal expression language OCL [x] to define our formal metamodel for context modeling (see sections 3.4), and specify its well-formedness rules (see sections 3.5), respectively. We a use natural language (NL) to define semantics (i.e., the meanings) of the formalized constructs (see section 3.6).
The abstract syntax is presented in a UML class diagram illustrating the meta-concepts, meta-attributes and meta-relationships. The diagram also presents some of the meta-constraints and well-formedness rules, particularly the multiplicity constraints of the meta-relationships. We use a natural language to define the semantics of (1) the meta-concepts and (2) the meta-relationships as well as their meta-attributes as follows:
This association should be changed into the opposite direction
Fig. 1. The metamodel for context modeling and analysis
| Goal |
You have to be 100% sure of the quality of your product to give a money-back guarantee. This describes us perfectly. Make sure that this guarantee is totally transparent.
Read moreEach paper is composed from scratch, according to your instructions. It is then checked by our plagiarism-detection software. There is no gap where plagiarism could squeeze in.
Read moreThanks to our free revisions, there is no way for you to be unsatisfied. We will work on your paper until you are completely happy with the result.
Read moreYour email is safe, as we store it according to international data protection rules. Your bank details are secure, as we use only reliable payment systems.
Read moreBy sending us your money, you buy the service we provide. Check out our terms and conditions if you prefer business talks to be laid out in official language.
Read more