The Ontology Model and Reasoner to Build an Autonomic System for U-Health Smart Home

Transcription

1 The Ontology Model and Reasoner to Build an Autonomic System for U-Health Smart Home Jin Kim DPNM Lab. Department of Computer and Communications Engineering POSTECH, Korea December 21, 2009 DPNM, POSTECH Master Thesis Defense 1/24

2 Agenda I. Contents and Motivation II. Objective III. State of Arts IV. Proposed Solution V. Implementation and Tests VI. Conclusion & Perspectives DPNM, POSTECH Master Thesis Defense 2/24

3 Contents and Motivation Aging society and Health cost problem Reduce the healthcare costs of elderly Abuse and neglect of senior citizens Bring healthcare to remote locations Among elderly, falls(60% at home) is leading cause of injury and hospital admission U-health Smart Home is one of the possible solutions Why is it complex? Understand the health support for elderly at home Understand the behavior of elderly at home Heterogeneity of technologies at home Medical symptoms are complex => Requires the convergence of technologies and knowledge DPNM, POSTECH Master Thesis Defense 3/24

4 Objective General question: How web semantic technologies can help to build a U-Health smart Home? Question 1: Is it possible to create model to capture all the details of the smart home and the elderly situation (including health) at home? Question 2: Is it possible to build some kind of autonomic decision making system that can use this semantic model to manage any elderly situation at home without any human intervention? DPNM, POSTECH Master Thesis Defense 4/24

5 State of Arts(1/3) Smart Home Projects Project Organization Year Target Characteristic Center for Future Health Rochester Univ ~ Medical care Proactive health system that supports the personal health care needs of all Aware Home Research Initiative Georgia Institute of Technology 2000 ~ User`s convenience Aimed at addressing the fundamental technical, design, and social challenges for people in a home setting MavHome project Texas Univ ~ User`s convenience Sense and predict the occupants mobility habits and their use of electrical appliances Medical automation Research Center Virginia Univ ~ Medical care Molecular automation, technologies to promote dignified aging, medical efficiency technology, etc. Smart home program Duke Univ ~ User`s convenience Student actually reside in home CodeBlue project Harvard Univ ~ Medical care infra Develop the wireless infrastructure for emergency medical care CASAS Washington State Univ ~ User`s convenience Intelligent home environment, multi-disciplinary research project IRIS Institute for Rebabilitation Republic of Slovenia 2007 ~ Medical care Persons with disabilities and elderly persons Nokia Smart Home Nokia Research Center 2007 ~ User`s convenience Building an open ecosystem for digital home control House_n Massachusetts Institute of Technology - User`s convenience Focus on how the design of the home and its related technologies, products, and services should evolve to better meet the opportunities and challenges of the future DPNM, POSTECH Master Thesis Defense 5/24

6 State of Arts(2/3) Autonomic Computing What is Autonomic Computing? Initiated by IBM in 2001 To develop computer systems capable of self-management To overcome the rapidly growing complexity of IT management Define the MAPE architecture MAPE: Monitor, Analyze, Plan, Execute Introduce the concept of Knowledge Base (KB) Database that contains all the knowledge necessary to the autonomic computing to achieve self-management Based on a semantic model (*) Other approaches are model less: AI approaches (Bio-inspired, etc.) DPNM, POSTECH Master Thesis Defense 6/24

7 State of Arts(3/3) Semantic Web Technologies What is Semantic Model? It is an abstract view that captures all the concepts and relationships within a domain What is Ontology? It is a formal representation of the semantic model Ontology is used to formally define domains What is Description Logic (DL)? Description logics are a family of knowledge representation language which can be used to represent the concept definitions of an application domain in a structured and formally well-understood way What is OWL? It is an Ontology language for defining and instantiating Web Ontology Allows to define concepts, relations, and facts (based on DL) Standard from W3C What is Reasoner? The Reasoner is a piece of software to infer logical consequences from a set of asserted facts or axioms It is interdisciplinary, covering research in, e.g., philosophy, logic, AI, statistics, cognitive science, law, psychology, mathematics, and the sciences DPNM, POSTECH Master Thesis Defense 7/24

8 Synthesis Smart Home Semantic Web Technologies Autonomic System DPNM, POSTECH Master Thesis Defense 8/24

9 Proposed Architecture for Smart Home (1/3) What are the requirements? Health care part Hospital / doctor Specialized organization Remote diagnosis Autonomic computing part Information filtering / aggregation Situation / context modeling Intelligence reasoning Decision making Home networking part Information gathering Service discovery Appliance discovery Smart Home Sensing part Nano / Bio sensor Environment sensor Actuator Home control unit Home automation DPNM, POSTECH Master Thesis Defense 9/24

10 Proposed Architecture for Smart Home (2/3) Home Sensor Network Autonomic System Sensor Monitor Sensor Sensor Aggregation node Analyze Plan Execute Knowledge Medical Sensor Network Internet Home Gateway Hospital ( Control System ) Appliances DPNM, POSTECH Master Thesis Defense 10/24

11 Proposed Architecture for Smart Home (3/3) Detailed Architecture for the Autonomic System KB Reasoning Interacting Health-care Provider Distributed Framework Monitor Analyze Plan Execute Autonomic System Wired and Wireless Network in the Smart Home Physical Environment DPNM, POSTECH Master Thesis Defense 11/24

12 Specification of an Ontology Model for U-Health Smart Home: General Approach 1. CIM Classes & Relations Selection We inspired from the DMTF Common Information Model CIM is an open standard that defines how managed elements in an IT environment are represented as a common set of objects and relationships between them Advantage: CIM is a Open and Free standard model in IT It is well documented The model is rich ( a lot of classes, relations ) Limitation: Oriented IT and not U-Health Smart Home No definition of context or situation Many concepts required in the U-Health Smart Home do not exist CIM is defined in UML but not in a semantic language Understand the semantic of the classes and their relations and select those CIM classes and relations which are useful to build our model DPNM, POSTECH Master Thesis Defense 12/24

13 Specification of an Ontology Model for U-Health Smart Home: General Approach 2. Select a Smart Home Specification Language : OWL (Web Ontology Language): OWL-Lite, OWL-DL, OWL-Full 3. Convert CIM classes/relations into OWL ontology model 4. Enhance the resulting model with new concepts and relations to be able to perform autonomic decisions in the U-Health Smart Home The result is called the SHOM (Smart Home Ontology Model) DPNM, POSTECH Master Thesis Defense 13/24

14 CIM Classes and Relations Selection CIM Schema CIM Schema = Core model + Common models The core model captures notions that are applicable to all areas of management and represents a starting point for determining how to extend the common schema The common models are information models that capture notions that are common to particular management areas, but independent of any particular technology or implementation Which classes are useful for the project? CIM_Core of Core Model CIM_Device of Common Model CIM_System of Common Model CIM_User of Common Model CIM_Service of Common Model Other not used models (may be useful in the future) CIM_Application, CIM_Database, CIM_Event, CIM_Interop, CIM_Policy, CIM_Network DPNM, POSTECH Master Thesis Defense 14/24

15 Converting CIM to OWL Converting the main CIM concepts into OWL CIM Class Generalization Property REF Property Min Max Required <owl:class> <rdfs:subclassof> <owl:datatypeproperty> <owl:objectproperty> <owl:mincardinality> <owl:maxcardinality> OWL <owl:mincardinality rdf:datatype= &xsd;int >1 </owl:mincardinality> Add semantic information to the relationships between concepts Functional Inverse Transitive Symmetric DPNM, POSTECH Master Thesis Defense 15/24

16 Enhancing the Resulting Model CIM classes are not sufficient to model the U-Health Smart Home world Added several concepts and relation between concepts TBox (TerminologicalBox) such as Patient (Monitored and Not Monitored) Sensors (Bio, Environment) Situation (Activity, etc.) Smart Home, Room, etc. Used several ABox (AssertionBox) to create relations between individuals and concepts SHOM: Proposed Smart Home Ontology Model DPNM, POSTECH Master Thesis Defense 16/24

17 SHOM: Smart Home Ontology Model DPNM, POSTECH Master Thesis Defense 17/24

18 Reasoning on the SHOM What is reasoning? Reasoning is the cognitive process of looking for reasons, beliefs, conclusions, and actions Reasoning is a kind of test for context recognition and right decision making How do we reason on an ontology? To define reasoning rules in the ontology To use a reasoner that is able to reason on the ontology (applying inner logic rules + the defined rules) DPNM, POSTECH Master Thesis Defense 18/24

19 Reasoning on the SHOM How to define rules? SWRL (Semantic Web Rule Language) is a proposal for a Semantic Web ruleslanguage, combining sublanguages of the OWL Web Ontology Language (OWL DL and Lite) with those of the Rule Markup Language SWRL uses logic operators to define rules AND, OR, NOT, {,,,, } SWRL rules are stored in the ontology as concepts DPNM, POSTECH Master Thesis Defense 19/24

20 Reasoning on the SHOM Example of rule to identify elderly situation and take an autonomic decision: If the elderly is in his bed and the TV is off and the Light is Off, then the elderly is sleeping If the elderly is sleeping, then set the temperature in the room to 27 degrees SWRL Formal Specification: isinroom(?x, bedroom) hasappliancestate(tv_bedroom, false) hasappliancestate(light_bedroom, false) hasuseractivitystate(?x, SleepingActivityState) DPNM, POSTECH Master Thesis Defense 20/24

21 Implementation and Tests Development environments Protégé and 4.0 JAVA SE Runtime environment Eclipse SDK Pellet API & OWL API Defined classes in the SHOM Whole classes (103) Defined TBox and Concepts Properties Object property (26) + Data property (37) DPNM, POSTECH Master Thesis Defense 21/24

22 Conclusion(1/2) Question 1: Is it possible to create model to capture all the details of the smart home and the elderly situation and health? We have proposed an ontology model (SHOM) to capture the concepts of the U-Health Smart Home. We have used successfully the ontology concepts to capture the semantic of the relations between several entities in the home Question 2: Is it possible to build some kind of autonomic decision making system that could take care of the elderly at home without any external help for many situations We have used SHOM to define semantic rules to capture the situation of the elderly at home. And based on this situation we have shown how we can enforce some autonomic decisions DPNM, POSTECH Master Thesis Defense 22/24

23 Conclusion(2/2) Contribution Initial specifications of a semantic model for U-Health Smart Home and initial autonomic decision-making Limitation The model and autonomic decision are only a first initiative that should be developed extensively taking into account real situations Future work Enhance the model and reasoning part Interface the autonomic system with the other components of the autonomic system Perform some real tests in the POSTECH Smart Home to understand elderly behavior at home DPNM, POSTECH Master Thesis Defense 23/24

24 Thank you Q&A DPNM, POSTECH Master Thesis Defense 24/24

25 Appendix DPNM, POSTECH Master Thesis Defense 25/24

26 Reasoner Comparison OWL-DL entailment Supported expressivity for reasoning Reasoning algorithm Consistency checking Bossam Pellet KAON2 RacerPro Jena FaCT++ yes yes yes Yes SROIQ(D) SHIQ(D) SHIQ(D-) SROIQ(D) Rule-based Tableau Resolution & Datalog Tableau Rule-based Tableau yes yes Incomplet for OWL DL DIG support no yes yes yes yes Yes Yes Rule support SWRL & own rule format SWRL SWRL SWRL-Not fully support & own rule format Own rule format No Licencing Free/closedsource Free/opensource & nonfree/clo sed-source Nonfree/clo sed-source Free/closedsource Free/opensource Free/opensource DPNM, POSTECH Master Thesis Defense 26/24

27 CIM MOF File Style MOF File Contents (ex: PhysicalElement ) // Copyright (c) 2005 DMTF. All rights reserved. // <change cr="sysdevcr " type ="change">update of // descriptions based on Tech Edit review.</ // <change cr="archcr " type="add">add UmlPackagePath // qualifier values to CIM Schema.</change> // ================================================================== // CIM_PhysicalElement // ================================================================== [Abstract, Version ( "2.10.0" ), UMLPackagePath ( "CIM::Core::Physical" ), Description ( "Subclasses of CIM_PhysicalElement define any component of a " "System that has a distinct physical identity. Instances of " "this class can be defined as an object that can be seen or " "touched. All Processes, Files, and LogicalDevices are " "considered not to be Physical Elements. For example, it is not " "possible to touch the functionality of a \'modem.\' You can " "touch only the card or package that implements the modem. The " "same card could also implement a LAN adapter. PhysicalElements " "are tangible ManagedSystemElements that have a physical " "manifestation of some sort. \n".. class CIM_PhysicalElement : CIM_ManagedSystemElement { [Key, Description ( "An arbitrary string that uniquely identifies the " "Physical Element and serves as the key of the Element. " "The Tag property can contain information such as asset " "tag or serial number data. The key for PhysicalElement MaxLen ( 256 )] string Tag;. DPNM, POSTECH Master Thesis Defense 27/24

28 CIM UML example(1/2) DPNM, POSTECH Master Thesis Defense 28/24

29 CIM UML example(2/2) SettingDefineState - Initial value setting ElementSettingData - current value setting SettingsDefineCapabilities - limitation of value DPNM, POSTECH Master Thesis Defense 29/24

30 Whole Classes Taxonomy DPNM, POSTECH Master Thesis Defense 30/24

31 Whole Classes with Object Properties DPNM, POSTECH Master Thesis Defense 31/24

32 ManagedElement Class DPNM, POSTECH Master Thesis Defense 32/24

33 PhysicalElement Class DPNM, POSTECH Master Thesis Defense 33/24

34 LogicalDevice and State Class DPNM, POSTECH Master Thesis Defense 34/24

35 UserEntity and UserActivityState Class DPNM, POSTECH Master Thesis Defense 35/24

36 ManagedSystemElement and Organizaiont Class DPNM, POSTECH Master Thesis Defense 36/24

37 Snapshot Using Protégé Tool Protégé 4 For OWL 2 Protégé For SWRL DPNM, POSTECH Master Thesis Defense 37/24

38 Autonomic System by IBM Four common managed resource arrangements Autonomic computing reference architecture DPNM, POSTECH Master Thesis Defense 38/24