In this paper we refine an organizational model to describe the execution of business activities. The model offers a dynamic, actor--centered, context--based and business process oriented perspective of the organization that explicitly addresses information, cooperation and collaboration needs derived from human multi--tasking capabilities. This modeling approach aims to facilitate the design of collaboration services capable of providing a more appropriate support for groups of interacting individuals. We illustrate our model with an example from an on--going case study, which was elaborated applying speech act theory to case study observations. Professional, service and administrative areas of the organization involve the execution of semi-structured and ad--hoc activities. Technological support for cooperation and collaboration in these business activities is provided by a variety of groupware applications (commonly called collaboration services) such as e--mail, chat, discussion boards and file sharing systems. All these systems offer a partial support to human business actors. Designing an appropriate technological support for groups of interacting people is not trivial (Slagter 2004). First, the required task--technology fit (Zigurs and Buckland 1998) means to achieve a complicated and delicate balance between providing a sufficiently rich support and avoiding information overload. Workers' information needs depend on action contexts defined by personal, role, task and time factors (Zacarias et al. 2005). Current work dynamics forces human business actors to frequently “switch” among several action contexts. Human multi--tasking capabilities and limitations are studied by Experimental Psychology (Rubinstein, Meyer and Evans 2001). Several cognitive theories claim the existence of mental executive processes that supervise and manage the execution of tasks. An analogy of executive processes with multi--tasking operating systems is established in (Kieras et al. 2000). Improving the task-technology fit entails devising collaboration services that provide an integral, personalized and timely information support to human business actors. This kind of support requires (1) determining action context specific needs and (2) a better understanding of action context--changing patterns. Second, the social technical--gap, i.e. the gap between social requirements and what technology is able to support, is considered one of the fundamentals challenge of computer supported cooperative work research (Ackerman 2000). Interactions among human actors involved in the execution of business
activities create social contexts reflecting the knowledge shared by these interacting actors. These business contexts include knowledge about current “state of affairs” i.e., current state of activity, participating actors and objects. Business context behavior is determined by shared assumptions, expectations, habits, protocols and rules, which support and regulate interactions among actors. Social contexts are dynamic constructs with different behavior among instances of organizational settings, activities and participants (Dourish 2004). Groupware social--technical gap can be reduced with the provision of context models enabling “context--informed” collaboration services that act according to the specific knowledge and behavior of each business context.
In this paper, we refine an organizational model introduced in (Zacarias et al. 2005) to describe the execution of business activities. This model, which offers a dynamic, actor--centered, context--based and business process oriented perspective of the organization, explicitly addresses information, cooperation and collaboration needs derived from human multi--tasking capabilities. The model seeks to facilitate: (1) improvement of task--technology fit, (2) reduction of social--technical gap and (3) balance between individual and group needs, when designing collaboration services. Moreover, since groupware often fails for lack of an explicit task control (Houben 1998), explicit business context models provide a means for supporting the execution of ad--hoc activities. The remaining of this paper is structured as follows: section 2 reviews related work on context and speech--act theory. Section 3 refines our model. Section 4 illustrates our model with an example from an on--going case study. The example was elaborated applying speech act theory to case study observations. Section 5 offers our conclusions and future directions.
This section presents work supporting our concepts and modeling approach. Section 2.1 contrasts the engineering and sociological approaches to context and briefly discusses two context--based organizational models. Section 2.2 summarizes Speech Act Theory concepts.
Although the notion of context plays an important role in multiple disciplines, there is no standard concept or theory. This notion varies according its area of application. However, there is consensus around one single idea: its relational nature i.e., context is always related to something else.
Classical Engineering Approach: In engineering, context has been approached as a stable and discrete notion. Context is viewed a collection of things (sentences, propositions, assumptions, properties, procedures, rules, facts, concepts, constraints, sentences, etc) associated to situations (environment, domain, task, agents, interactions, conversations, etc). This notion is reflected by the “box metaphor” (Benerecetti, Bouquet and Ghidini 2001). The intuition is that context is a “box” where its content 2 depends on a set of parameters. The specific parameter set varies according to the area of application. In Artificial Intelligence, parameters (called dimensions) such as time, location, culture, topic, granularity and modality have been proposed as defining elements of context (Lenat). Context--aware applications identify localization, user identity, activity and time as context parameters (Dey 1999). A proposal for a workflow context in Maus (2001) includes the following parameters: function, behavior, causality, organization, information, operation and history.
Sociological Approaches: Seeking to improve system sensitivity to specific settings, research in context-aware computing is focusing on a view of context inspired by sociological investigations of real--world practice (Dourish 2004). This work contrasts the objective account of engineering and the subjective account of phenomenology and discusses the implications of approaching context from the latter perspective. The phenomenological perspective argues that (1) context is a relational property among objects, (2) the scope of contextual features are defined dynamically, (2) context is relevant to particular settings, instances of action and participants and (4) context and activity are not separable i.e. context is embedded in activity and arises from it. Under this perspective, the focus moves from context representation to context support.
The apparent contradiction between the objective and subjective positions is denied by Structuration Theory (Giddens 1984), which seeks a balance between both positions. According to this theory, interactions among subjects both produce and reproduce social practices. On one side, social practices are produced from interactions among subjects. On the other side, from these interactions emerges an objective structure (interaction context) which provides rules and resources that simultaneously support the reproduction of social practices and constrain subject interactions.
Context Use in Organizational Models: Distributed knowledge management projects (Bonifacio and Bouquet 2002) recognize the subjective, social, and nature of knowledge, and approach organizations as a network of contexts. These projects are developing theoretical frameworks and technical solutions under this organizational perspective. A context--based approach of organizations is also proposed in (Gachet and Brézillon 2005). This work makes a case for the need of making explicit several kinds of organizational contexts. Although context--based, these approaches are not business process--oriented and do not address human context changing dynamics.
Developed by Austin (1975) and extended by Searle (1968), speech act theory focuses on the use of language to modify states of affairs, rather than to describe states of affairs. Austin claimed the existence of sentence that were in themselves actions and called them performatives. Performatives express a speaker’s intention with his hearer. Performatives are not true or false, they succeed or fail. Performatives succeed if they satisfy a set of social conventions called by Austin as felicity conditions. The action involved in a performative is a speech act. Thus, performatives are speech acts subject to felicity conditions. According to Austin a speech act is composed of three kinds of acts: the locutionary act is the act of uttering the words, the illocutionary act reflects the speech act intended effects and the perlocutionary act reflects the speech act actual effects. Perlocutionary acts are less frequently tied to verbal expressions Searle does not distinguish among the locutionary and the illocutionary acts. Rather, he describes speech acts in terms of a propositional content and its illocutionary force, which expresses the speaker’s intention. The propositional content is a dimension of the illocutionary act. Since its origins, speech act theory has been applied in several areas such as discourse analysis, message structuring and categorization, agent communication language design and organizational modeling .
We start this section defining business contexts in terms of three overlapped, but still different, concepts: activity, action and interaction contexts. Later we use these concepts in refining our model.
Contexts are used in several areas. The context notion of engineering has been mainly used for automation purposes, i.e. enhancing automated actors or reasoning mechanisms. Nevertheless, when aiming at supporting human actors, social sciences offer a more appropriate approach to context. The main components of action contexts are illustrated in table 1. An action context reflects the actual behavior and information needs of a human actor when he performs a specific task under a given role. Since these needs change in time, an action context defines an actor’s state during some time interval. Such state represents a subset of actor’s properties and their values and it is determined by his on--going actions and interactions. Although we here use the “box metaphor” of the engineering approach, this is not a discrete notion since we will consider several relevance degrees.
Human actors participating in a business activity act and interact from specific action contexts. These interactions are partially determined by the activity context i.e., pre--defined activity behavior and activity state. At execution time, successive interactions among action contexts create a collective object that we define as the interaction context. This definition is mostly influenced by sociological approaches to context. The interaction context reflects actual activity--related behavior and state of affairs. The shared part of interacting action contexts and the activity context belong to the interaction context. Thus, these concepts are somewhat overlapped. Interaction context behavior is determined by a number of elements such as interaction rules and protocols. Interaction protocols describe valid sets of action and interaction sequences among action contexts that meet specific purposes related to tasks or activities. Interaction rules define the set of conditions required for successful interactions among action contexts. Both cooperation and collaboration in business activities are supported and regulated by the specific rules and protocols of the interaction context where they take place. Figure 1 illustrates the interaction context main components.
SShhaarreedd AAcctitoionn CCoonntteexxtt
InIntteerraacctitoionn
PPrroottooccoolsls&&RRuuleless Interaction Context
SShhaarreedd AAccttioionn
CCoonntteexxtt
Fig 1. Interaction Components main components In summary, business contexts are approached as dynamic and composite concepts formed by the interplay of several action contexts mediated by an activity and interaction context. Business context particular properties and behavior will depend on the action and interaction history among participating actors. Different human actors performing identical tasks and roles behave differently. Consequently, business contexts of identical activities and role sets but with different participants, will exhibit different properties and behavior.
Due to their multi--tasking nature, human business actors are capable of handling several action contexts and participating in several interaction contexts .We propose a model to that explicitly addresses human context identification and management capabilities. Figure 2 illustrates our model. The organization is viewed as a network of individual and collective multi--tasking actors. The model explodes the actor notion and approaches it as a network of contexts managed by actor’s own “operating system”. Three kinds of actors are defined: human, business process actor and organization actor. Human actors are modeled as a network of action contexts. Human “operating system” determines individual habits and action context changing patterns. The business process actor is a collective actor modeled as a network of activity contexts. Its “operating system” acts as an “engine” managing the execution of pre--defined flows of work. The organization actor is another collective actor which represents sets of organized groups of interacting individuals with specific purposes such as teams or departments. It is modeled as a network of interaction contexts. The organization “operating system”: (1) provides a communication environment based on shared expectations, rules and protocols which support and regulate activity--related actions and interactions and (2) manages ad--hoc behavior and exceptions not handled by business processes. The inclusion of two kinds of collective actors explicitly acknowledges the organization’s capability to act flexibly by combining pre--defined and structured behavior with ad--hoc behavior. It also acknowledges that operational procedures handled by business processes, are ultimately modified by individual features and social practices of the human actors involved in its execution.
Human “Operating System”
Interaction Context Action Action Context Interactions Context
Human “Operating
System”
Organization “Operating
System” Interaction Context
Management
Activity
Context Activity Context Management Business Process “Operating System”
The “Business Context”
Identification
Process
Fig 2. The proposed model based on the Operating Systems Metaphor We use the operating systems metaphor for several reasons. First, it provides a conceptual framework with a clear cut separation between actual execution and its management. Second, it allows a dynamic view of actors. Third, it offers a straightforward abstraction of human multi--tasking capabilities. In order to act or interact, human actors must first identify and activate the corresponding action context. This metaphor allows an explicit representation of context identification and management operations. From our point of view, this organizational perspective facilitates a more personalized and timely support to human business actors. Moreover, an overall view of individual and collective actors facilitates the design of services seeking a balance between individual and group needs. The model assumes a flexible business process management. However, we do not focus on business process issues. Rather, we address model context--related concepts from a business process perspective. In our work, we aim at building dynamic business context models from activity--related action and interaction history. Specifically, we seek to capture particular elements of: (1) interaction contexts, i.e. shared action contexts (2) human “operating systems”, i.e. individual habits and action context--changing patterns, and (3) organization “operating system” i.e., supporting and regulating elements of action context interactions, such as shared interaction rules and protocols.
In order to illustrate our model, we present an example from observations of an on--going case study, which involves a software development team of 4 programmers and their project manager. The team develops web applications for a commercial bank. Due to space limitations, we do not describe the collecting procedure in detail. Computer and non--computer based actions and interactions of the team were registered in a chronological order. This was manually done by team members and complemented by the project manager. Each action or interaction was described with separated sentence. Three weeks of actions and interactions were registered, encompassing 534 sentences. among illocutionary and perlocutionary acts. The latter were determined establishing cause--effect relationships among related actions and interactions. We identified illocutionary force (in uppercase) and propositional content of illocutionary acts. We used the terms sender--receiver, rather than speaker--hearer since several interactions were computer--mediated. By appropriately grouping actor actions and interactions, we may discover actual roles, tasks and activities. This grouping is illustrated in table 3 for actor Ann. We inferred the manage project role from the following actions and interactions: (1) request project control meeting, (2) do prepare project report status and (3) answer application related information. Also, two Ann’s tasks were identified: (1) control team software projects and (2) provide application--related information. The identification of tasks and roles required analyzing illocutionary force and propositional content of illocutionary acts. action contexts and their evolution in time. Table 4 shows the evolution of two action contexts: (1) Ann executing task prepare control meeting under the role of manage projects between 9 and 14:00 and (2) Ann executing task provide application--related information (under the same role) between 11 and 14:00. " " " " 1 1 ' ( "
! To appropriately identify human and organization “operating system” elements, more complex analysis techniques are required. In table 5 we show some basic elements of a human (5a) and business context (5b) operating systems identified in table 2 and complemented by Ann. In this example we are not showing any interaction rule. Whilst modeling action contexts (table 4) enables to support actors’ specific execution needs, capturing human “operating systems” allows helping actors to manage his several tasks (table 5a). Collaboration services based on the organization’s particular interaction rules and protocols (organization operating system ---- table 5b) would provide a more appropriate support to group interactions.
In this paper we refine a dynamic, actor----centered, context----based and business process oriented modeling approach to describe the execution of business activities introduced in (Zacarias et al. 2005). Business contexts are defined as the interplay of three kinds of context: activity, action and interaction contexts. The model uses an operating systems metaphor to address information, cooperation and collaboration needs taking into account human multi----tasking capabilities. We illustrate our model with an example elaborated using a speech act----based analysis of action and interaction logs of an----going case study. This method offers several benefits. It eases log analysis as it makes explicit shared information and collaboration needs of human actors. It also provides a common format and it will enable to decompose interactions into machine----understandable elements.
The overall goals of our research are: (1) provide an organizational view to facilitate the design of “context informed” services that improve task----technology fit, reduce socio----technical gap and balance individual and group information and collaboration needs, (2) provide groupware with means for supporting the execution of ad----hoc activities and (3) provide a bottom up approach for the specification of tasks and roles.
Our work is in a preliminary stage. At this present moment we are focusing on defining ways to describe and represent business contexts. Manual analyses pose several limitations. Automated support for log storage and analysis must be developed. Analyzing actions and interactions with speech act theory enables the retrieval of contextual knowledge, but it does not convey valuable tacit knowledge. Observation and collection procedures must be enhanced with techniques to elicit this knowledge. Specifically, ethnographic methods will be explored. We aim at creating dynamic business context models from action and interaction logs. This will require further structuring the logs and researching on appropriate analysis and modeling techniques. Model validation methods must also be devised. Additional case studies are required to further test, adjust and refine our model as well as our observation and collection procedures.
Zigurs I, Buckland (1998), B. K., “A Theory of Task/Technology Fit and Group Support Systems Effectiveness”, MIS Quarterly, 22, 313----334, Ackerman M. (2000), “The Intellectual Challenge of CSCW: The Gap Between Social Requirements and Technical Feasibility”, Human----Computer Interaction, 15:181----205 Houben G., “A New Modeling Approach for Group Communication”, wwwis.win.tue.nl/~houben/respub/etce98.ps Zacarias M., Caetano A., Pinto H.S., Tribolet J.(2005), “Modeling Contexts from Business Process Oriented Knowledge Support”, Proceedings of 2nd Workshop on Knowledge Management for Distributed Agile Process, Benerecetti M., Bouquet P., Ghidini C. (2001), “On the dimensions of context dependence: partiality, approximation, and perspective”, Third International and Interdisciplinary Conference on Modelling and Using Context, Springer----Verlag, LNAI 2116: 59----72, Maus H. (2001), “Workflow Context as a Means for Intelligent Information Support”, Third International and Interdisciplinary Conference on Modeling and Using Context, Springer----Verlag LNAI 2116: 261– 274,.
Gachet A., Brézillon P. (2005), “A context----based Representation of Organizational Structures”, Proceedings of the 18th International Flairs Conference, Invited Special Track “AI for Social Networks, Social Networks in AI”, Dey A., Abowd G. (1999), “Towards a Better Understanding of Context and Context----Awareness”, GVU Technical Report, ftp://ftp.cc.gatech.edu/pu b/gvu/tr/1999/99----22.pdf Lenat D., The Dimensions of Context----Space, CycCorp, http://casbah.org/resources/cyccontextspace.shtml Dourish P. (2004), What we talk about when we talk about context, Pers Ubiquit Comput, 8:19----30 Giddens A. (1984), The Constitution of Society, University of California Press Austin J. L. (1975), How To Do Things With Words, 2nd ed. Harvard, University Press Searle, John R (1968) ‘Austin on Locutionary and Illocutionary Acts’. The Philosophical Review 77: 405---424 Rubinstein J., Meyer D., Evans J. (2001), “Executive Control of Cognitive Processes in Task Switching”, Journal of Experimental Psychology: Human Perception and Performance 27(4): 763--797 Bonifacio M., Bouquet P.(2002), “Distributed Knowledge Management: a Systemic Approach, Emergence in Complex, Cognitive, Social and Biological Systems”, Kluwer Academic Publishers Kieras, D., Meyer, D., Ballas, J., Lauber, E. (2000), “Modern computational perspectives on executive mental processes and cognitive control: Where to from here?”, In Control of Cognitive Processes: Attention and Performance XVIII, 681--712., MIT Press