distributed Situation and Context Awareness

(1)

Hybrid approaches based on Computational Intelligence and Semantic Web for

distributed Situation and Context Awareness

P

H

D

IN

T

HEORIES

, M

ETHODOLOGIES AND

A

DVANCED

A

PPLICATIONS FOR

C

OMMUNICATION

, C

OMPUTER

S

CIENCE AND

P

HYSICS

Coordinator:

Prof. Giuseppe Persiano

Supervisor:

Prof. Vincenzo Loia

Ph.D Student Domenico Furno

XI C^YCLE – N^EW S^ERIES

A^PRIL 2013

(2)

Roadmap

Theoretical and Technological

Foundations

Proposed Approaches

and Research Objectives

Architectural Overviews Application and

Scenarios

Conclusions Introduction

 Introduction

 Some Definitions

 Research Trend

 Research Focus, Significance and Objectives

 Theoretical and Technological Foundations

 Situation/Context Awareness and Semantic Sensor Web

 From Semantic Sensor Data to Situation/Context Awareness

 Ontology-based Situation Theory

 From a Fuzzy Extension of Situation Theory to Fuzzy Situation Theory Ontology

 SPARQL

 f-SPARQL

 f-SPARQL facilities for Clustering and Classification

 Proposed Approaches & Research Objectives

CoMSA

SbESA

CAPSD

 Architectural Overviews and Application Scenarios

 Situation Awareness & Airport Security

 Situation Awareness & Smart Grids

 Context Awareness & Healthcare

 Conclusions

Some Definitions

 Situation Awareness

o is the perception of environmental elements with respect to time and/or space, the comprehension of their meaning, and the projection of their status after some variable has changed, such as time, or some other variable, such as a predetermined event.

o It is also a field of study concerning the perception of the environment critical to decision-makers in complex, dynamic areas.

 Context Awareness

o refers to the idea that computers can both sense, and react based on their environment.

• Devices may have information about the circumstances under which they are able to operate and based on rules, or an intelligent stimulus, react accordingly.

o Specifically, it is considered a process to support user applications to adapt interfaces, tailor the set of application-relevant data, increase the precision of information retrieval, discover services, make the user interaction implicit or build smart environments.

(3)

Research Trend

Foundations

Scenarios

Conclusions Introduction

Hong J.Y., Suh E. H., Kim S.J.: Context-aware systems: A literature review and classification. (2009) Expert Systems with Applications, 36 (4), pp. 8509-8522

(4)

Research Focus, Significance and Objectives

Foundations

Scenarios Introduction

 The research studies in Situation/Context Awareness have highlighted that the main issues related to these areas are:

 the need to support the acquisition and aggregation of dynamic environmental information from the field;

 the lack of formal approaches to knowledge representation and processing;

 the lack of automated and distributed systems to support the reasoning through software special purpose.

 The thesis researches new approaches for distributed Context and Situation Awareness and proposes to apply them in order to achieve some related research objectives such as:

 Knowledge Representation;

 Semantic Reasoning;

 Pattern Recognition;

 Information Retrieval.

 In particular, the proposed approaches are based on:

 semantic web technologies and languages;

 computational intelligence methodologies and techniques;

 multi-agent distributed paradigm.

(5)

Situation/Context Awareness and Semantic Sensor Web

Introduction

Scenarios

Conclusions

Foundations

 Semantic Sensor Web

 the idea is to add semantic annotations to existing standard Sensor Web languages in order to provide semantic descriptions and enhanced access to sensor data;

 this is accomplished with model-references to ontology concepts that provide more expressive concept descriptions.

I can exploit the Sensor Data Pyramid idea behind Semantic Sensor Web to support the semantic modeling of sensor data involved in Situation/Context Aware applications…

Sensor Data Pyramid

Raw Sensor (Phenomenological) Data Feature Metadata

Entity Metadata Ontology Metadata

Data Information Knowledge

(6)

Ontology-based Situation Theory

Introduction

Scenarios

Conclusions

Foundations

In ST (introduced by Barwise and Perry, 1983), information about a situation is expressed in terms of infons.

 Infons are written as

To capture the semantics of situations, ST provides a relation between situations and infons.

This relationship is called the supports relationship and relates a situation with the infons that ‘‘are made factual’’ by it.

Given an infon

σ

and a situation

s

the proposition “

s

supports

σ

” is written as

A formalization of Barwise’s situation semantics in terms of an ontology, with some parts using mathematics and rules, has been named STO (M.

Kokar, C. J. Matheus, K. Baclawski, 2009).

𝝈_𝒊 ≡≪ 𝑹, 𝒂_𝟏, … , 𝒂_𝒏, 𝝋 ≫

𝐬 ╞ 𝛔^.

(7)

From a Fuzzy Extension of Situation Theory to Fuzzy Situation Theory Ontology

Introduction

Scenarios

Conclusions

Foundations

 I propose a fuzzy extension of semantics related to Situation Theory, namely, Fuzzy Situation Theory Ontology (FSTO).

 FSTO meta-model for SA can evolve in a natural way towards the approximation and uncertainty modeling.

Thus, in my interpretation, the polarity of an infon 𝜎_𝑖 supporting a situation 𝑠_𝑗 can be one of the terms defining a linguistic variable expressing infon’s truth. For instance, let us say

(𝐼𝑛𝑓𝑜𝑛𝑇𝑟𝑢𝑡ℎ, ℑ(𝐺), 0. .1 , 𝐺, 𝑀)

 𝐺 is the grammar generating terms in ℑ(𝐺)

 𝑀 is the semantic rule which associates each linguistic value with its meaning.

The definition of the context free grammar 𝐺 involves

(𝑡𝑟𝑢𝑒, 𝑓𝑎𝑙𝑠𝑒) as primary terms, a finite number of hedges

(𝑚𝑜𝑟𝑒 𝑜𝑓 𝑙𝑒𝑠𝑠, 𝑞𝑢𝑖𝑡𝑒, 𝑟𝑒𝑎𝑙𝑙𝑦, … ) whose evaluation in 𝑀

is performed by means of concentration and dilation, the con- nectives 𝑎𝑛𝑑 and 𝑜𝑟, and the negation 𝑛𝑜𝑡.

(8)

From a Fuzzy Extension of Situation Theory to Fuzzy Situation Theory Ontology

Introduction

Scenarios Theoretical and

Technological Foundations

 I propose a fuzzy extension of semantics related to Situation Theory, namely, Fuzzy Situation Theory Ontology (FSTO).

 FSTO meta-model for SA can evolve in a natural way towards the approximation and uncertainty modeling.

As a result, an infon 𝜎_𝑖 supporting a situation 𝑠_𝑗 is written as:

𝜎_𝑖,𝑠_𝑗 ≡ ≪ 𝑅_𝑖, 𝑎₁, 𝑎₂, … , 𝑎_𝑛, 𝜏_𝜎

𝑖,𝑠𝑗 ≫ 𝑤𝑖𝑡ℎ 𝜏_𝜎

𝑖,𝑠𝑗 ∈ ℑ(𝐺) stating that 𝑅_𝑖 𝑎₁, 𝑎₂, … , 𝑎_𝑛 𝑖𝑠 𝜏_𝜎_{𝑖,𝑠𝑗} in 𝑠_𝑗.

By adopting this modeling approach, the semantic of support proposition

╞ can be stated as

𝑠_𝑗╞_𝑒𝑥𝑡 {𝜎_𝑖,𝑠_𝑗} ⟺ ∀ 𝑖 ∶ 𝑅_𝑖 𝑎₁, 𝑎₂, … , 𝑎_𝑛 𝑖𝑠 𝜏_𝜎

𝑖,𝑠𝑗

This interpretation lead us to define a modeled situation occurrence as the evaluation of a corresponding fuzzy control rule:

IF 𝑅₁ 𝑎_1,1, 𝑎_1,2, … , 𝑎_1,𝑛₁ 𝑖𝑠 𝜏_𝜎_1,𝑠𝑗 AND …

AND 𝑅_𝑖 𝑎_𝑖,1, 𝑎_𝑖,2, … , 𝑎_𝑖,𝑛_𝑖 𝑖𝑠 𝜏_𝜎

𝑖,𝑠𝑗 THEN 𝑠_𝑗 𝑖𝑠 𝑜𝑐𝑐𝑢𝑟𝑟𝑖𝑛𝑔

otherwise formalized as:

𝜇_𝑜𝑐𝑐(𝑠_𝑗) = 𝜇_𝜏

𝜎 𝑖,𝑠𝑗 𝑅_𝑖 𝑎_𝑖,1, 𝑎_𝑖,2, … , 𝑎_𝑖,𝑛_𝑖

𝑖

where ⋀ is a suitable t-norm operator and Occurrency is a fuzzy set.

(9)

SPARQL

Introduction

Scenarios

Conclusions

Foundations

 SPARQL (pronounced “sparkle”, a recursive acronym for SPARQL Protocol and RDF Query Language) is an RDF query language, that is, a query language for databases, able to retrieve and manipulate data stored in Resource Description Framework format.

 Querying RDF Graph

 Set of triples

 SPARQL sintax is similar to SQL

 SPARQL allows users to write unambiguous queries

(10)

f-SPARQL

Introduction

Scenarios Theoretical and

Technological Foundations

query SPARQL

query result query

f-SPARQL

result transformation

α-cut

(11)

CoMSA

Introduction

Foundations

Scenarios

Conclusions

Approaches Research Objectives

CoMSA Approach

Fuzzy Situation Theory Ontology

Key elements Semantic Sensor Web

Multi Agent Paradigm

Knowledge Representation

Semantic Reasoning

Pattern Recognition

Information Retrieval SbESA Approach

Fuzzy Clustering & Classification

Key elements Semantic Sensor Web Multi Agent Paradigm

CAPSD Approach

Fuzzy Matchmaking Key elements Semantic Sensor Web

Multi Agent Paradigm Sematic Web Services

(12)

CoMSA

Introduction

Foundations

Scenarios

Knowledge Base Facts Background

Knowledge

Situations Models

Situation Theory Ontology Meta-model Local

Facts

Local Facts ...

Ontology Editor

Rule Editor

1

2 3

Raw Events

Level 1: Perception 4

Level 2: Comprehension Events Management Agency

Focal Individual Infons Responsibile

5

Qualitative Infons

Level 3: Projection Situations Evaluation Agency

Focal Situation Responsibile Local Facts Local Situation Model

Inference Engine 6 7

Situations Occurrence

Decision Making

HCI Safety/Security

Operator

8

(13)

Situation Awareness and Airport Security

Introduction

Foundations

Scenarios

Conclusions

 New types of threats (terrorism, organized crime, etc.) can make risky the normal conduction of airport operations.

 So, it is necessary to achieve a common enhanced situation awareness involving:

 Relevant data sharing;

 Qualitative data sharing;

 Collaborative decisions.

 Specific Goal:

 support security operators in the detection of critical situations in the internal areas of the airport;

 Intelligent Decision Support System;

 CoMSA.

(14)

Track Data Generator

Events Generation ...

State of the Environment

Sensors Radars

Knowledge

Situations Models FSTO Meta-model

Actions Performance for Goals Satisfaction

Decision Making

Events Management

Agency

HCI

Knowledge Management Semantic

Annotation

... Situations

Evaluation Agency

Safety/Security Operator

Security

Architectural Overview

Introduction

Foundations

Scenarios

(15)

Ontological Modeling - Knowledge Base

Introduction

Foundations

Scenarios

Conclusions

Knowledge

Situations Models

Situation Theory Ontology Meta-model Ontology Editor

Rule Editor

Aircraft

Holding Point Runway

Exit Point

Clearance Point

hasClearance

isConnectedTo

Taxyway Airport

hasRunway

hasExitPoint

conductsTo hasHoldingPoint

Location Velocity

hasAttribute hasAttribute

sto:Individual

is-a

is-a is-a

Base Model Definition

 Definition and modeling of the reference

domain

 Identification of the involved static objects

 Identification of the involved dynamic objects

 Generation of Situational Models: modeling of the situations in terms of

1) Statements about relations and properties both of observed domain entities and secondary relevant situations;

2) Focal individuals for the situation.

Knowledge

Situations Models

Situation Theory Ontology Meta-model Ontology Editor

Rule Editor

AircraftCrossingOnRunway Class

subClassOf sto: Situation Class sto: supportedInfon

ObjectProperty

sto: supportedInfon ObjectProperty

domain

sto: ElementaryInfon Class

range

range range

Restriction Restriction

onProperty onProperty

isArrivedTo HoldingPoint

Infon Class

isStopBar Infon Class

allValuesFrom

sto: SIT Class

Instance of

sto: supportedInfon ObjectProperty

Restriction

allValuesFrom

isArrivedToEnd CrossingPoint

Infon Class

domain

onProperty

range

sto:focalIndividual ObjectProperty

domain

Aircraft Class

range

(16)

Simulation Results

Introduction

Foundations

Scenarios

Ground Truth

Test Case No. Sweep Radar

Duration (min.)

Max No. Traces

No.

Anomalies and Conflicts

1 356 12 21 8

2 498 17 21 5

3 604 21 16 5

4 576 20 16 6

5 583 20 18 5

6 505 17 19 8

7 645 22 17 7

Measure Definition

fairness (or precision)

𝑟𝑒𝑙𝑒𝑣𝑎𝑛𝑡 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠 ∩ 𝑟𝑒𝑡𝑟𝑖𝑒𝑣𝑒𝑑 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠 𝑟𝑒𝑡𝑟𝑖𝑒𝑣𝑒𝑑 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠

completeness (or recall)

𝑟𝑒𝑙𝑒𝑣𝑎𝑛𝑡 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠 ∩ 𝑟𝑒𝑡𝑟𝑖𝑒𝑣𝑒𝑑 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠 𝑡𝑜𝑡𝑎𝑙 𝑟𝑒𝑙𝑒𝑣𝑎𝑛𝑡 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠

timeliness 𝑀𝑒𝑑𝑖𝑢𝑚 𝑇𝑖𝑚𝑒 𝐹𝑖𝑟𝑠𝑡 𝐼𝑛𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑃𝑟𝑜𝑐𝑒𝑠𝑠 + 𝑀𝑒𝑑𝑖𝑢𝑚 𝑇𝑖𝑚𝑒(𝑆𝑒𝑐𝑜𝑛𝑑 𝐼𝑛𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑃𝑟𝑜𝑐𝑒𝑠𝑠)

 SIMULATOR: ASAS (Airport Security Agents System)

(17)

CoMSA Research Objectives

Introduction

Foundations

Scenarios

Conclusions

Approaches Research Objectives

CoMSA Approach

CAPSD Approach

(18)

SbESA

Introduction

Foundations

Scenarios

Conclusions

Approaches Research Objectives

CoMSA Approach

CAPSD Approach

Key elements Semantic Sensor Web Sematic Web Services

(19)

SbESA

Introduction

Foundations

Scenarios

Conclusions

Data Analysis

Knowledge Base Semantic Sensor Observations

Semantic Sensor Web Ontologies Obs

1

Obs N ...

Clustering

Classification Rules Automatic Design

...

Classification Rules

1 Level 1: Perception

Place Management Agency 2

3

Level 2: Comprehension Situation Classification Agency

Sensor Observation Classification

Responsibile Classifier Semantic

Obs 4

5

Level 3: Projection Observations

Classification Results

Situation Notification Agency Classification

Schema 6

7 Situations

Occurrence

Decision Making HCI

Safety/Security Operator

8

(20)

Situation Awareness and Smart Grids

Introduction

Foundations

Scenarios

Conclusions

 Smart Grids paradigm is expected to support the evolution of traditional electrical power systems.

 In particular, the constant growth of grid complexity and the need for supporting rapid decisions require paradigms:

 more scalable and flexible;

 proactive;

 self-healing.

 support the evolution of traditional electrical power systems toward web energy networks composed by distributed and cooperative energy resources;

 A situation aware distributed and cooperative monitoring system.

 SbESA.

(21)

Workflow Execution Phase Workflow Training Phase

Architectural Overview

Introduction

Foundations

Scenarios

Conclusions

Sensor

Ontology Temporal Ontology

Domain Ontology Geospatial

Ontology

Training Dataset

Semantic Data Sensing Variable N

Variable 1 Variable 2

f-SPARQL

Swarm-based Anomaly RecognitionRelevant Patterns Notification ...

sensors

(22)

Workflow Training Phase

Introduction

Foundations

Scenarios

 Building of training dataset

 Clustering of sensor observations

◦ identification of events that may occur in the system

fSPARQL

(23)

Workflow Execution Phase

Introduction

Foundations

Scenarios

Conclusions

 Swarm behavior exploiting ant foraging metaphor

 It foresees the employment of four populations of agents

Legend

Detector Agents Finder Agents

Classifier Agents

Place Agents

(24)

Workflow Execution Phase – Place Agents

Introduction

Foundations

Scenarios

 These agents provide infrastructure services in order to manage the pheromone properties (such as concentration, evaporation, and propagation) on each node of the network.

 They also implement pheromone’s logics as well as operations of data sensing.

 Moreover, they provide topological information and manage

the interactions between neighboring agents.

(25)

Workflow Execution Phase – Finder Agents

Introduction

Foundations

Scenarios

Conclusions

 These agents move in the environment in a totally random way, trying to estimate clusters density, which can be interpreted as a measure of not anomaly for clusters.

 Each Finder Agent maintains a vector with as many cells as clusters;

 Then, Finder Agents on the same cell will periodically exchange among themselves the information contained in their arrays, with the aim to refine the search.

Vector Cluster1 15 Cluster2 18

Cluster3 1

Cell Data 0.1 0.3 0.6

New Vector 15.1 18.3 1.6

Vector FA_A Cluster1 15.1 Cluster2 18.3 Cluster3 1.6

Vector FA_B 10 10 1

Vector FA_A& FA_B 12.55 14.15 1.3

f-SPARQL

(26)

Workflow Execution Phase – Detector Agents

Introduction

Foundations

Scenarios

 These agents cooperate in a stigmergic way (i.e. through the release of pheromone) and their aim is to identify local anomalies;

 Detector Agents, on their side, firstly contact a Finder Agent present in the same place in order to get the density vector.

FA Vector Cluster1 12.55 Cluster2 14.15 Cluster3 1.6

DA Density Vector 0.45

0.5 0.05

DA Anomalies Vector 0.05

0 0.45

𝑑𝑒𝑛𝑠𝑖𝑡𝑦𝑑𝑒𝑔𝑟𝑒𝑒_{𝑐𝑙𝑢𝑠𝑡𝑒 𝑟}_𝑖 = 𝑑𝑒𝑛𝑠𝑖𝑡𝑦_{𝑐𝑙𝑢𝑠𝑡𝑒𝑟 𝑖}

𝑑𝑒𝑛𝑠𝑖𝑡𝑦^𝑛_𝑗=1 _{𝑐𝑙𝑢𝑠𝑡𝑒𝑟 𝑗} 𝑎𝑛𝑜𝑚𝑎𝑙𝑦𝑑𝑒𝑔𝑟𝑒𝑒_{𝑐𝑙𝑢𝑠𝑡𝑒 𝑟}_𝑖 = 𝑚𝑎𝑥_{0≤𝑗≤𝑛} 𝑑𝑒𝑛𝑠𝑖𝑡𝑦𝑑𝑒𝑔𝑟𝑒𝑒_{𝑐𝑙𝑢𝑠𝑡𝑒 𝑟}_𝑗 − 𝑑𝑒𝑛𝑠𝑖𝑡𝑦𝑑𝑒𝑔𝑟𝑒𝑒_{𝑐𝑙𝑢𝑠𝑡𝑒 𝑟}_𝑖

 The release of pheromone by Detector Agents depends on the previously calculated anomaly vector. In particular, an Agent Detector will drop three types of pheromone:

Type Description Calculated as

ARP Detector Agents are attracted by this type of pheromone

RRP This type of pheromone is adversed by Detector Agents

 n 1

i belongingdegreecluster_i*anomalydegreecluster_i

= ARP

RRP = system_average_concentration

f-SPARQL

(27)

Workflow Execution Phase – Classifier Agents

Introduction

Foundations

Scenarios

Conclusions

 They classify anomaly sensor observations detected in the previous phase highlighting the most important ones.

 They release Classification Pheromone (CP), which spreads and evaporates slowly.

 The population of Classifier Agents moves into the environment following FP trails released by Detector Agents and according to their classification scheme, which can be local or distributed.

 The concentration of CP, released on the nodes, will depend on both the concentration of FP and the confidence that the Classifier Agents have in having found a pattern;

 confidence increases as its classification scheme proves to be exact.

(28)

Simulation Results

Introduction

Foundations

Scenarios

Simulator GUI

Sensor Network Deployment Utility

Situation Awareness Result

(29)

SbESA Research Objectives

Introduction

Foundations

Scenarios

Conclusions

Approaches Research Objectives

CoMSA Approach

CAPSD Approach

(30)

CAPSD

Introduction

Foundations

Scenarios

Conclusions

Approaches Research Objectives

CoMSA Approach

CAPSD Approach

(31)

CAPSD

Introduction

Foundations

Scenarios

Conclusions

Knowledge Base Semantic Sensor Observations

Semantic Sensor Web Ontologies Obs

1

Obs N ...

...

User Profile Ontology

User Profile

Semantic Datasets Semantic Web Services OWL-S

Repository

1

Phase 1: Semantic Web Services Definition

Domain Services Modeling

Agency

2 3

4 Phase 2: User Context Analysis and

Identification Context Sensing

Agency

5

6 Phase 3: Context Aware Brokering Context Aware

Services Providing Agency

7

Context Aware User Devices

Personalized Services 8

User Context

Context Enriched

Services

(32)

Context Awareness and Healthcare

Introduction

Foundations

Scenarios

 The rapid worldwide deployment of Internet and sensor technologies is the enabler of a new generation of healthcare applications;

 One of the main problems in this domain is the healthcare personalization;

 To achieve healthcare personalization, factors such as individual’s lifestyle, surrounding situations, device capabilities, event of happenings, etc., should be taken into account.

 support the personalized providing of healthcare services by exploiting user’s context information.

 An Enahnced Context-Aware System for the personalized providing of Healthcare Services;

 CAPSD.

(33)

Architectural Overview

Introduction

Foundations

Scenarios

Conclusions

(34)

Context Sensing Agents Workflow

Introduction

Foundations

Scenarios

Blood Pressure

Temperature

Blood Pressure

Blood Oxygen

Blood Oxygen Heartbeat

...

Blood Pressure

Context Training Data Set

Context Domain Ontologies

Blood Pressure

Heartbeat Blood Oxygen

Context Fuzzy Rule Base RULE: IF Blood Pressure is high AND …

THEN Cluster_2;

RULE: IF Blood Pressure is low AND … THEN Cluster_1;

RULE: IF … THEN

Features Selection Fuzzy Clustering Execution

Fuzzy Classifier Design

Context Training Phase

(35)

Context/Service Matchmaking Workflow

Introduction

Foundations

Scenarios

Conclusions

RULE: IF Blood Pressure is high AND Blood Sugar is high… THEN Cluster_2;

RULE: IF … THEN Body Sensor Nodes

U s e r P r o f i l e

Context Fuzzy Rule Base

OWL-S Semantic Heathcare Services

Provide First Aid

First Aid Service

Result Precondition

Blood Pressure

First Aid

...

...

Fired Context Rules ...

Semantic Service Precondition

Fuzzy Rules – Semantic Services Similarity

Context Domain Concepts Semantic Services

Concepts

Semantic Matching Wearable Devices Fuzzy Rule Engine

Execution

Hearth Attack Risk

0 130 180

1 High

Blood Pressure is High

Degree of Fulfilment

Context-Aware Service Discovery Phase

Input/Output Precondition Result

(36)

Context/Service Matchmaking

Introduction

Foundations

Scenarios

Conclusions

 The pseudo-code of matchmaking algorithm is the following:

Input:

– Fuzzy Rules (FR);

– Semantic Web Services (SWS);

Output:

– Ranked list of services;

for each rule r_i in FR

for each antecedent a_h in r_i for each sws_j in SWS

evaluate deg_MATCH(sws_f) for each sws_j in SWS

evaluate avarege degree of deg_MATCH(sws_i)

return Ranked list of services according to the average of deg_MATCH



















 

 µ ( )

) , ( max

) (

deg ¹ _i x

n B A S SWS

j n

j

f i MATCH

(37)

Simulation Results

Introduction

Foundations

Scenarios

Conclusions

Medical Care Services Data Set

No. Total Services No. Total Relevant Services

670 300

Query Context No. Relevant

Services No. Steps

1 Heart Attack 28 15

2 Fainting 37 15

3 Flu 45 15

4 Ictus 56 15

5 Tachycardia 64 15

6 Renal Insufficiency 70 15

Web Application PC Web Application Smartphone

(38)

CAPSD Research Objectives

Introduction

Foundations

Scenarios

Conclusions

Approaches Research Objectives

CoMSA Approach

CAPSD Approach

(39)

Final Remarks

Introduction

Foundations

Scenarios

Conclusions

 This research work analyzes and addresses the main issues in the fields of Situation and Context Awareness.

 As result, this work proposes to combine technologies deriving from Semantic Web and techniques of Computational Intelligence in order to overcome these challenges and meet research objectives:

 Knowledge Representation;

 Semantic Reasoning;

 The main contributions of this research work concern with:

 the exploitation of sensor ontologies to support the acquisition and aggregation of dynamic environmental information from the field (i.e. sensors, cameras, etc.);

 the definition of formal approaches to knowledge representation (i.e. situations, contexts, concepts, relations, etc.);

 the definition of formal approaches to knowledge processing (i.e. reasoning, classification, extraction, retrieval, recognition, discovery, etc.);

 the definition of multi-agents architectures capable to efficiently support the modeling and processing of a large amount of knowledge.

 Future works are going to focus on:

 Semantic Modeling of Fuzzy Control;

 Temporal Issues;

 Automatically Ontology Elicitation;

 Semantic Query Editing;

 Extension to other Application Domains;

 More Experiments.

 Pattern Recognition;

 Information Retrieval.