Authors: Xiao Hang Wang, Da Qing Zhang, Tao Gu, Hung Keng Pung Institute for Infocom Research, Singapore Some slides adopted from earlier presentation of Sangkeun Lee; IDS Lab Akmal Khan Multimedia and Mobile Communication Lab, SNU, Korea

Agenda Background Introduction CONON(Context Ontology) Ontology Reasoning & User Defined Reasoning Experiments & Comparison Conclusions & Questions

Pervasive computing

Pervasive computing…also known as:

What is Context? Context is any information that can be used to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and application themselves.

Context Types Location – position, orientation, velocity, trajectory, etc. Identity – preference, profile, social relationship, biometrics, etc. Time – Sequence of events, duration, etc. Activity – walking, sleeping, sitting, etc. Task – meeting, reading, working, etc. Environment – temperature, humidity, brightness, loudness, etc. Computing resources – device, appliances, etc. Emotion

What is Context-Aware? A system is context-aware if it uses context to provide relevant information and/or services to the user, where relevancy depends on the users’ task. Goal To acquire and utilize information about the context of a device to provide services that are appropriate to the particular people, place, time, events, etc. EX) A cell phone will always vibrate and never beep in a concert, if the system knows the location of the cell phone (i.e., user) and/or the concert schedule.

Acquiring Context Explicitly By requiring the user to specify e.g. Current Location Implicitly By monitoring user and computer-based activity e.g. monitoring of user interaction to turn of a device after a period of inactivity e.g. monitoring of battery power for adaptation of power- intensive applications Acquisition of context Smart environments Embed sensors in ultra-mobile devices

Context Models

Context Model Context model is one of the infrastructure and is essential for efficient manipulation of context information. Context model includes various categories and high complexity–Due to comprehensibility and manageability Enables adaptation of complex architecture It is important not only to define the range of context correctly–To understand the characteristics of each context model

Context Representation Key-Value models: use a set of attributes and their associated values. Markup models: structure context into a hierarchy using tags. Graphical models: express relationships between context entities. Object-Oriented models: structure context into object classes and their implicit relationships. Logic models: express context in terms of facts and rules. Ontologies: combination of logic models and O-O models they structure context into object classes and their explicit relationships.

Agenda Background Introduction CONON(Context Ontology) Ontology Reasoning & User Defined Reasoning Experiments & Comparison Conclusions & Questions

Introduction Context-awareness an important step in pervasive computing Increasing need for developing formal context model to facilitate Context Representation Context Sharing Interoperability of heterogeneous systems

Introduction: Previous Works Various context data models Context Toolkit: Attribute-value Tuples CoolTown: Web based data model each object has a corresponding Web description Karen et al: ER and UML Gaia: First-order pridicates written in DAML+OIL However, None of them has addressed Formal knowledge sharing Quantitative evaluation for the feasibility of context reasoning in pervasive computing environments

Introduction: What’s in this paper? In this paper, the authors present An ontology-based formal context model to address critical issues Formal context representation Knowledge sharing Logic based context reasoning Detailed design of their context model and logic based reasoning scheme Quantitative evaluation for context reasoning in pervasive computing

Why Ontology Model? Ontology The shared understanding of some domains Often conceived as a set of entities, relations, functions, axioms and instances Reasons for developing context models based on ontology Knowledge sharing The use of context ontology enables computational entities to have a common set of concepts about context Logic Inference Context aware computing can exploit various existing logic reasoning mechanisms Knowledge reuse We can compose large-scale context ontology without starting from scratch

Agenda Background Introduction CONON(Context Ontology) Ontology Reasoning & User Defined Reasoning Experiments & Comparison Discussion & Conclusions

CONON: The Context Ontology Fundamental: Location, User, Activity, Computational Entity Skeleton of context Act as indices into associated information Upper Ontology Context in each domain shares common concepts Encourages the reuse of general concepts Provides flexible interface for defining application- specific knowledge

CONON Upper Ontology

Specific Ontology for Home Domain

Agenda Background Introduction CONON(Context Ontology) Ontology Reasoning & User Defined Reasoning Experiments & Comparison Discussion & Conclusions

Context Reasoning The authors present a smart phone scenario E.g. when the user is sleeping in the bedroom or taking a shower in the bathroom, incoming calls are forwarded to voice mail box The use of context reasoning has two folds Checking the consistency of context Deducing high-level implicit context from low-level explicit context Two categories of context reasoning Ontology reasoning User-defined reasoning

Ontology Reasoning

Example: Ontology reasoning

User-defined Context Reasoning

Agenda Background Introduction CONON(Context Ontology) Ontology Reasoning & User Defined Reasoning Experiments & Comparison Discussion & Conclusions

Experiment  The prototype context reasoners are built using Jena2

Summary of Context Models Adopted from : A Survey of Context Modeling By: Seungseok-Kang ;IDS Lab

Comparison

Ontologies Evaluation Based on usage of ontology languages Using ontology design principles

Ontologies Evaluation With respect to pervasive computing themes Themes reference by reference ontology based models

Agenda Background Introduction CONON(Context Ontology) Ontology Reasoning & User Defined Reasoning Experiments & Comparison Discussion & Conclusions

Discussion Three major factors Size of context information Complexity of reasoning rules CPU speed The authors insist that it is feasible for non-time-critical applications For time-critical applications such as security and navigating systems We need to control the scale of context dataset and the complexity of rule set Off-line manner static complex reasoning tasks De-coupling context processing and context usage is needed in order to achieve satisfactory performance The design of context model should take account of scalability issue

Questions The major factors Size of context information Enhanced CoCA: heuristics (loading only relevant context data) Complexity of reasoning rules CPU speed: Not our concern  How can we control the complexity of reasoning rules? We need to define the minimal set of rule language Expressively powerful enough to be used in actual context-aware system Guarantees acceptable performance Is there a way of applying only relevant reasoning rules?  What happen if the user-defined rule becomes no longer satisfied? Presented system doesn’t consider

Technology Roadmap

Conclusions OWL encoded context Ontology (CONON) Modeling context in pervasive computing environment Logic based context reasoning Upper Ontology + Domain-specific Ontology Prototype implementation and Experiment Feasible for non-time-critical applications Discussion: what we need to care for time-critical applications Center for E-Business Technology IDS Lab. Seminar - 37

Thank you