Modeling and Using Context

In the domain of knowledge representation as well as in Conceptual Modeling, representing part-whole relations is a long-stand-ing challenging problem. Most approaches addressing this issue rely on a set-theoretical framework, but many difficulties remain especially for disambiguating transitivity. In mathematical logic and program checking, dependent type theories have proved to be appealing but so far, they have been little applied in the formalization of knowledge. To bridge this gap, we represent part-of structures in a dependently-typed framework with the purpose of enhancing expressiveness through an explicit introduction of properties characterizing the context. We show that the dependently typed language easily captures the notion of contextualized part-of with many examples.

Referring expressions with vague and ambiguous modifiers, such as

”a quick visit”

and

”the big meeting”

, are difficult for computers to interpret because their words’ meanings are in part defined by context, which changes throughout the course of an interpretation. In this paper, we present an approach to interpreting context-dependent referring expressions that uses

dynamic binding

. During the incremental interpretation of a referring expression, a word’s meaning can be defined in part by properties from the current candidate referents—its denotation up to the previous word for the tentative interpretation.

In this paper we describe and evaluate an algorithm for generating referring expressions that uses linear regression for learning the probability of using certain properties to describe an object in a given scene. The algorithm we present is an extension of a refinement algorithm modified to take probabilities learnt from corpora into account. As a result, the algorithm is able not only to generate correct referring expressions that uniquely identify the referents but it also generates referring expressions that are considered equal or better than those generated by humans in 92% of the cases by a human judge. We classify and give examples of the referring expressions that humans prefer, and indicate the potential impact of our work for theories of the egocentric use of language.

Contextually based reasoning is an essential aspect of human cognition, permeating language, memory, and reasoning capabilities. This integral process is developed over the lifetime through experiential learning. Given the goal of artificial intelligence to mimic human intelligence, it is essential to include such contextual considerations in system design and implementation. We compare selected computational architectures and cognitive paradigms on the basis of key elements in human intelligence understanding in order to illustrate the similarities and differences between the two viewpoints and highlight the potential effectiveness of context based computing. In the literature, we discover meaningful parallels between the assessment of context in cognition and computation which have implications for both fields of study.

Using Contextual-Graphs (CxG) to represent organizational activity supports selecting key performance indicators (KPIs) that are free from causality assumptions and measuring practice-based organizational learning as distinct from organizational change. This paper presents four tools developed as part of a research program to extend CxG to

practice-based organizational learning and performance improvement. Aspectual comparison of practices represented in contextual graphs

together with a

practice-based organizational learning novelty typology

, a

practice maturity model,

and an

organizational-performance-improvement prioritization matrix

operationalize the construct of

practice maturity

that is proposed as a guide for researchers and practitioners in understanding and improving

practice-based activity management

. Results are presented for a transport planning organization that elucidate the role of context in

practice-based organizational learning and performance improvement

in the case of an activity (

light rail route selection)

that involved both political/managerial and engineering decisions.

People spend most of their time in a few significant places and often indoors in a small number of select rooms and locations. Indoor localization in terms of a user’s current place, related to a user’s daily life, routines or activities, is an important context. We implemented an automatic approach DCCLA (Density-based Clustering Combined Localization Algorithm) to automatically learn the Wi-Fi fingerprints of the significant places based on density-based clustering. In order to accommodate the influence of the signal variation, clustering procedure separately works on a list of RSSIs (Received Signal Strength Indicators) from each AP (Access Point). In this paper, the approach is experimentally investigated in a laboratory setup and a real-world scenario in an office area with adjacent rooms, which is a key challenge to distinguish for place learning and recognition approaches. From these experiments, we compare and identify the most suitable parameters for the unsupervised learning.

The experience of experts relies on the process of decision-making jointly with the progressive elaboration of a context-specific model. However, context modeling generally stays implicit because only the result of the decision-making process matters. Modeling context within a decision-making process supposes a uniform representation of knowledge, reasoning and contexts. In the Contextual-Graphs formalism, a decision-making episode is represented as a contextual graph where each path represents a practice developed by actors for making this decision in a specific working context. By incremental accumulation of the practices developed by experts, a contextual graph becomes an experience base with the decision-making process. Such an experience base may be used by humans for training future experts how to behave in the different ways to make a decision according to the variants of the working context. An intelligent assistant system exploiting such an experience base will be able to propose a more effective support to users than previous knowledge-based systems. This work is realized in the framework of a project in medicine for supporting experts in breast cancer diagnosis.

Case-based reasoning solves new problems by retrieving cases of similar previously-solved problems and adapting their solutions to fit new circumstances. The case adaptation step is often done by applying context-independent adaptation rules. A substantial body of research has studied generating these rules automatically from comparisons of prior pairs of cases. This paper presents a method for increasing the context-awareness of case adaptation using these rules, by exploiting contextual information about the prior problems from which the rules were generated to predict their applicability to the context of the new problem, in order to select the most relevant rules. The paper tests the approach for the task of case-based prediction of numerical values (case-based regression). It evaluates performance on standard machine learning data sets to assess the method’s performance benefits, and also tests it on synthetic domains to study how performance is affected by different problem space characteristics. The results show the proposed method for context-awareness brings significant gains in solution accuracy.

Complex design and engineering processes are characterized by dynamic requirements, like changing process goals or group constellations. To deal with these dynamics, a context-sensitive approach is needed to consider a changing environment and provides teams with the support they need. This paper describes research about a context-sensitive intervention approach to support collaboration in dynamic environments. Based on a review of existing approaches for context modeling in collaboration, a semantic model is presented to describe a collaboration process design as well as contextual process information. Using existing theories on collaboration performance, the paper discusses how the semantic model can be used to monitor group performance during collaboration. Thereby, a rule concept is introduced to derive interventions for dynamic collaboration processes and discusses their application to build new context-sensitive collaboration support systems.

At times, human behavior seems erratic and irrational. Therefore, when modeling human decision-making, it seems reasonable to take the remarkable abilities of humans into account with respect to rational behavior, but also their apparent deviations from the normative standards of rationality shining up in certain rationality tasks. Based on well-known challenges for human rationality, together with results from psychological studies on decision-making and from previous work in the field of computational modeling of analogy-making, I argue that the analysis and modeling of rational belief and behavior should also consider context-related cognitive mechanisms like analogy-making and coherence maximization of the background theory. Subsequently, I conceptually outline a high-level algorithmic approach for a Heuristic Driven Theory Projection-based system for simulating context-dependent human-style rational behavior. Finally, I show and elaborate on the close connections, but also on the significant differences, of this approach to notions of “ecological rationality”.

Culture is embodied in how people interact with other individuals and with their environment. It is a way of life formed under specific historical, natural and social conditions. Cross-cultural communication consists of human-to-human, human-to-machine, and human-to-environment communication in cross-cultural environments. The environment can be physical, virtual or hybrid. In our research, context is defined as a situation a user has at hand.

Cross-cultural communication environment – user – situation

is the key triplet in our context research. We introduce a context model for cross-cultural communication environments, and we give two examples of how we have applied it to the design of cross-cultural environments. Our case cultures are those of Japan and Finland.

The term ‘context,’ in software engineering, has been typically associated to mean the act of setting boundaries and setting system scope. In this paper, we revisit the concept of ‘context’ and draw ideas from other areas of engineering and the social sciences to suggest that context is a much richer concept that requires a systematic approach to model all of its relative aspects. It constitutes more complex phenomena concerning how a system interacts with its surroundings or even the world. Therefore, we outline a synthesised view of context to be used as a foundation for any approach that intends to apply ‘context’ effectively within a process or a modelling framework for software engineering. The synthesis is identified from five combined themes for any model to use context effectively.

Finding a shared understanding of context that is both theoretically coherent and operationalizable – e.g., for application in robotics, intelligent agent systems, or e-coaching products – is a significant challenge currently present in context research. This paper tries to capture the myriad of factors that together shape the multifaceted notion of context by conceptualizing the boundaries of contexts as a multitude of constraints within which actors operate. Within this ‘constraint-based approach’, context is broken down into different types, distinguishing between external and internal, as well as individual and shared contexts. In addition, it introduces vocabulary to differentiate between types of context transitions. This vocabulary is used to explain misinterpretations of context and misunderstandings between actors about the current context. Finally, the paper proposes a way of understanding context synchronization (or, context conflict resolution) between actors through context negotiation.

Hybrid tense logic is an extension of Priorean tense logic in which it is possible to refer to times using special propositional symbols called nominals. Temporal indexicals are expressions such as

now

,

yesterday

,

today

,

tomorrow

and

four days ago

that have highly context-dependent interpretations. Moreover, such indexicals give rise to a special kind of validity—

contextual validity

—that interacts with ordinary logical validity in interesting and often unexpected ways. In this paper we model these interactions by combining standard techniques from hybrid logic with insights from the work of Hans Kamp and David Kaplan. We introduce a simple proof rule, which we call the Kamp Rule, and first we show that it is all we need to take us from logical validities involving

now

to contextual validities involving

now

too. We then go on to show that this deductive bridge is strong enough to carry us to contextual validities involving

yesterday

,

today

and

tomorrow

as well.

Contextualized Knowledge Repository (CKR) is a DL-based framework for representation and reasoning with context dependent knowledge. It addresses the widely recognized need for contextualization of the Semantic Web data sources. Reasoning with CKR is possible thanks to a reduction to standard DL, and more recently a

NExpTime

tableaux algorithm was introduced for

$\mathcal{ALC}$

-based CKR. In this paper we present an

ExpTime

tableaux algorithm for

$\mathcal{ALC}$

-based CKR. The algorithm not only formally defines a tableaux decision procedure with optimal complexity, it is also presented in a form that can be effectively applied in practice employing a suitable rule application strategy together with node caching.

In a Science Museum / Center setting, a context based approach to an exhibit could provide the best results when trying to encourage young teens to pursue a career in a STEM (Science, Technology, Engineering, and Mathematics) field. An overview of our new context based exhibit that is currently being developed along with some preliminary test results indicate that the methods used in this exhibit have drawn young teens into exploring STEM careers.

Although much attention has been devoted to modeling and using context in intelligent agents, relatively little has been given to the problem for multiagent systems (MASs). Yet, just as with an individual agent, context affects how a MAS should behave. In this paper, we discuss an approach to distributed context management for multiagent systems. The approach is based on earlier work on context-mediated behavior (CMB) for single agents, which explicitly represents contexts as c-schemas that contain knowledge about how to behave in the contexts represented. We are distributing CMB for use in advanced multiagent systems. This work is just beginning, and so the paper discusses issues and potential approaches to distributing CMB.

The paper presents the contextualization of the Anatomical pathology (AP) workflow focusing on nonconformity during the reception and registration steps. Context is described by contextual elements related to heterogeneous sources such as the actor, the task, the situation and the local environment. The objective of this work is the application of contextual graph to the "workflow" of an AP exam, limited to the steps of reception and registration. The main point is the context-based representation of practices developed by actors instead of an object-centered view on the workflow to identify and ultimately avoid risky practices leading to nonconformity and prejudice. Therefore contextual graphs can be considered as a tool for monitoring the quality of work in an AP department, the actions, detection and correction of nonconformities as well as a base for intelligent software creation that links different modules in medical care.

There is a widespread health informatics vision of unlimited exchange, understanding and reuse of clinical information. However, it has also been pointed out that to understand clinical information it is to some extent necessary to know the circumstances of its production - the production-contextual clinical information.

The purpose of this study was to investigate the nature and significance of production-contextual clinical information in doctors’ everyday clinical work in order to asses whether standardization is necessary and possible. The study was performed through observation and focus group interviews at a cardiology department in a midsize Danish hospital.

It was found that production-contextual clinical information is complex, extensive, non-quantitative, and that it has an elusive structure. It is concluded that while it may be possible to standardise a limited amount of production-contextual clinical information, a general standardisation may very well be impossible.

The complexity of tasks and problems in the management of databases requires the development of tools for supporting database experts. For instance, in the database administration area, when problems occur, the database administrator (DBA) is frequently the first person blamed. Most DBAs work in a fire-fighting mode and have little opportunity to plan ahead or be proactive. They must be constantly ready to analyze and correct failures based on a large set of procedures. In addition, they are continually readjusting these procedures and developing practices to manage a multitude of specific situations that differ from the generic situation by some few contextual elements. These practices have to deal with these contextual elements in order to solve the problem at hand. This paper proposes to use “Contextual Graphs” formalism to improve existing procedures used in database administration. Up to now, this improvement is achieved by a DBA through practices that adapt procedures to the context in which tasks should be performed and the incidents appear. This work present a new version of the contextual graph platform as a basis for designing and implementing a context-based intelligent assistant system for supporting database administrators.

In this paper we analyze the relationship between context and situational awareness with the aim to get a better understanding of how context information influences situation assessment. The analysis is based on previous research on situational awareness, context and situations. We show how situation assessment could be specified more detailed with regards to sub-processes through investigating the components of a situation. Events are introduced as situational elements in themselves. The role of context in situation assessment is also analyzed and so is the information that can be treated as contextual. A case study is presented that relates the findings to the monitoring of oil well drilling. Our main contribution is an analysis of the situation assessment process and how it operates on and manipulates the components of the situation. Another contribution is the case study of how the findings apply in monitoring of the oil well drilling situation.

Cloud computing is currently receiving much attention from the industry, government, and academia. It has changed the way computation is performed and how services are delivered to customers. Most importantly, cloud services change the way software is designed, how data is handled, and how testing is performed. In cloud computing, testing is delivered as a service (TaaS). For instance, case testing (one of the most common validation approaches) could be used. However, executing test cases on a cloud system could be expensive and time consuming. Therefore, test case reduction is performed to minimize the number of test cases to be executed on the system. In this paper, we introduce a validation method called Context-Assisted Test Case Reduction (CATCR) for systems that are deployed on the cloud. In CATCR, test cases are reduced based on the context of the validation process. The results of previous test cases are used to select test cases for the next iteration. The minimized set of test cases needs to have effective coverage of the system on the cloud. To evaluate CATCR, an experimental evaluation is performed through Amazon’s Cloud and a Java validation tool. Experimental results are recorded and presented.

In the last decade, several works proposed their own list of quality of context (QoC) criteria. This article relates a comparative study of these successive propositions. The result is that no consensus has been reached about the semantic and the comprehensiveness of QoC criteria. Facing this situation, the QoCIM meta-model offers a generic, computable and expressive solution to handle and to exploit any QoC criterion within distributed context managers and context-aware applications. For validation purposes, QoCIM is successfully applied to the modelling of a set of simple and composite QoC criteria.

Nowadays, generation of user interfaces adapted to the context becomes very important in a world where technology and technical adaptations interfaces continue to increase. This paper presents an approach for the automatic and real time generation of user interface adapted to the context of use. We are particularly interested by modeling the data of context as well as the user’s task using Petri nets. The Petri nets are used for its simplicity to generate XML files, which will present the specification and generation of the human-computer interfaces adapted to the context. This approach will be illustrated with a case study that presents a diabetic patient suffering from hypoglycemia in a “smart hospital”.

In science learning, context is an important dimension of any scientific object or phenomenon, and context-dependent variations prove to be as critical for a deep understanding as are abstract concepts, laws or rules. Our hypothesis is that a context gap can be illuminating to highlight the respective general-particular aspects of an object or phenomenon. Furthermore, provoking a perturbation during the learning process to obtain the emergence of such an event could be a productive tutoring strategy. We introduce the emergence of context effects as a problem space, to be modeled in the system. We propose a model of the contextual dimension, associated with an analytical view of its modeling, based on a metaphor in physics.

One and the same inter-organizational business process - such as e-procurement - may be executed differently in different industries, geopolitical regions, etc. Thus, a standardized reference model for inter-organizational business processes must be customized to the specific business context (industry, region, etc.). In order to share, search, and (partially) re-use context specific adaptations it is essential not only to store the adaptations, but also a business context model where these adaptations are valid. In this paper we describe the Unified Context Model (UCM) introduced by the United Nations Centre for Trade Facilitation and Electronic Business (UN/CEFACT). We explain the shortcomings of the approach and show how these can be undermined by our Enhanced Unified Context Model (E-UCM). The enhanced model serves as a basis for contextualizing business documents which are exchanged between different inter-organizational business processes. Having such an approach at hand, helps prevent negative trends in today’s business, such as interoperability issues, inconsistencies and heterogeneous interpretations of the exchanged data contents.

Context is only one of several strata of meaning and we can not predict realisation at the lexical or grammatical level from context alone. Yet, there is a tendency to confuse contextual patterning with semantic patterning and allocate patterning to the contextual level that might better be dealt with on other levels. While much work has been done on theorising lexis and grammar and, more recently, on seeing these in context, much remains to be done on theorising semantics as a separate level mediating between context and lexis and grammar. This paper examines the problem of modelling behaviour and the challenge of understanding behaviour in context as well as on a semantical level. By understanding the descriptive responsibilities allocated to each level of language, we are better able to see what remains to be covered by context within a model.

According to Brézillon, context is not directly involved in solving a problem, but forced his resolution. In educational community, knowledge is transmitted from tutor to learner(s). However, two main issues are occurred. On one hand, Learner is facing resource bank and s/he has to seek the appropriate information from the relevant resource. To solve this problem, it is required to help learner by providing the context of resources in a pertinent organization. Ontologies represent the essential technology for the organization of eLearning resources. On the other hand, a massive amount of ontologies have spread among eLearning Systems. Our objective is to coordinate these ontologies in order to expand the (re)use, the search, the share of learning resources. For that aim, we propose a method which defines laws to relate automatically relevant parts of different ontologies, based on a mathematical model “Information Flow”.

The reliable, efficient and seamless exchange of business information is essential for a successful execution of the interwoven business processes. However, development of the contents for electronic data exchange is time consuming and usually can not follow the agile demands of the today’s business. If we could contextualize the pieces of the currently valid business information, we could predict its possible context specific variations. Therefore, the business contextual knowledge in which the already existing data contents are valid could be exploited to (semi-) automatically generate new, more homogeneous contents for electronic data interchange.

We aim to help the coordination of the activities of groups of users who share certain tasks. In particular, we are working towards automatically predicting the context of each user, in particular which task each user is trying to accomplish. We also intend to predict how probable it is that users will be able to successfully accomplish theirs tasks. In case a failure is likely, we help the users in negotiating task reallocation among group members. This paper presents the interaction patterns we use for information exchange among agents in order to determine the context needed to make those predictions.

The context is the set circumstances that surround an event or object as reminded by

[1]

. According to this definition, taking the context into account when analyzing an activity would require having at one’s disposal information describing the circumstances that surround each object concerned by this activity. This information is, by definition, difficult to describe precisely at the beginning of the analysis and must be progressively defined during the analysis process. We propose to gradually define the context information surrounding objects concerned by the activity,

based on the interactive learning of activity traces,

revisited by successive interpretations during the analysis. The context model of the components proposed by [1] will be considered as the basis to define the activity components, which are the objects concerned by the activity. The approach by interactive discovery of knowledge from activity traces should allow to discover the values of these components but also new knowledge specific to the observed activity. This approach allows us to

dynamically build the context

useful to the activity analysis. This methodology will be applied in the field of transportation to determine the impact of driving behaviors on fuel consumption.