Proactive Domain Data Querying based on Context Information in

Ambient Assisted Living Environments

Vinícius Maran

1,4

, Alencar Machado

1,2

, Iara Augustin

Leandro Krug Wives

and José Palazzo M. de Oliveira

Institute of Informatics, Federal University of Rio Grande do Sul, Porto Alegre, RS, Brazil

Polytechnic School, Federal University of Santa Maria, Santa Maria, RS, Brazil

Technology Center, Federal University of Santa Maria, Santa Maria, RS, Brazil

Academic Coordination Office, Federal University of Santa Maria, Cachoeira do Sul, RS, Brazil

Keywords: Situation, Prediction, Context-awareness, Data Query, Ambient Assisted Living, Ubiquitous Computing.

Abstract: Ubiquitous computing defines a set of technologies to make computing omnipresent in real life environ-

ments. In the area of ambient assisted living, ubiquitous technologies have been used to improve the life

quality and expectancy for elderly people. Recently, researches have shown that the use of context-

awareness combined with proactive actions can cause systems to act more appropriately in assistance to the

user. In this paper, we present a new persistent and proactive data retrieval model for ambient assisted living

systems. This model provides an architecture that is able to integrate information that is gathered from the

user environment and considers the current user context to act in a proactive manner. The model was im-

plemented on a service integrated in a Situation as a Service middleware and it was applied in a case study

for evaluation and validation.

1 INTRODUCTION

Presently, we perceive an increased demand of tech-

nology for domestic environments, either as the

world population is aging (WorldMeters, 2013), and

by the fact that companies often prefer to keep em-

ployees at home working remotely. This context

emphasizes the possibility of expanding Smart

Homes and Ambient Assisted Living’s (AAL) appli-

cation domain. The vision presented in this article is

focused on Smart Homes. The home infrastructure

and AAL layer provides services dedicated to home

care for people who need help in daily activities.

People and devices interact to perform specific

activities, which are oriented to develop daily activi-

ties in these environments. In this sense, different

situations arise and are often related to the user pro-

file and also related with the activity being executed.

For example, people with impaired cognitive abili-

ties due to aging may unintentionally place them-

selves in dangerous or uncomfortable situations in

simple activities such as cooking, or leave a room

without checking if important actions such as turn-

ing off the shower were taken. Software and hard-

ware solutions related to ubiquitous computing are

being proposed to support this demand. For instance,

in a previous paper the Situation as a Service Mid-

dleware (SIaaS) was presented (Machado et al.,

2014). In the current paper, a solution to store in-

formation and make contextualized selection of

information using SIaaS middleware is presented.

This solution is supported by a set of require-

ments established in other works (Maran et al.,

2014) (Makris et al., 2013). Current solutions do not

clearly address how an information query can be

constructed with contextual information in high

level to generate the most appropriate decision based

in the acquired data to manipulate current or future

situations. The definition of a model to integrate

ontologies and JavaScript Object Notation for

Linked Data (JSON-LD) language (Lanthaler and

Gutl, 2012) is presented in this work. Thus, a signif-

icantly decrease in the overhead information pre-

sented by other tools is expected. Furthermore, we

also developed a model to integrate ontological

representation of context information, serialized in

JSON-LD, and domain information, which are often

used in ubiquitous systems to allow querying this

information based on context. A model to query

domain information based on context information

610

Maran V., Machado A., Augustin I., Krug Wives L. and Palazzo M. de Oliveira J..

Proactive Domain Data Querying based on Context Information in Ambient Assisted Living Environments.

DOI: 10.5220/0005365006100617

In Proceedings of the 17th International Conference on Enterprise Information Systems (ICEIS-2015), pages 610-617

ISBN: 978-989-758-097-0

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

was also defined. With these models, a service to

persist and retrieve data related to reactive and pro-

active situations in the SIaaS was created.

This paper is organized as follows: Section 2 dis-

cusses the main concepts related to this work; Sec-

tion 3 presents the models and service to query do-

main data based on context information; Section 4

presents a case study and discusses the application

of the models on it; Section 5 presents the conclu-

sions and future works in this area.

2 BACKGROUND

In the future, Ambient Intelligence (AmI) will ena-

ble environments to support people to inhabiting

them, being sensitive to their needs and capable of

anticipating behaviors (Sadri, 2011). Ambient As-

sisted Living (AAL) and Smart Homes are emerging

as AmI focused on specific characteristics of users.

For instance, existing researches present conceptual

models to transform homes into AAL environments,

modeling their context and services (Klein et al.,

2007). AAL characterize a domestic and automated

environment as one in which different users interact

personally and with physical objects.

These physical objects can be managed by spe-

cific systems developed for AAL. As the user behav-

ior change over the time, situations that involve

users also change over time. In our case, AAL’s

situations are related to health and, for this reason

experts’ knowledge for detection and handling these

situations is necessary. Besides, AAL systems need

knowledge about the world around users they moni-

tor and, in order to perform actions, they need to

interact with users through interaction devices (Au-

gusto et al., 2009). Moreover, services have to cope

with medical guidelines in a context-aware way in

order to provide users with instructions that are

appropriate to the patient’s situation. Systems for

AAL can choose most appropriate actions when they

are supported by mechanisms to query historic con-

text, current and futures situations involving the

user. With this information, the system may act in

advance, which characterizes a proactive system.

The implementation of context awareness in compu-

ting systems is a key requirement for it to occur.

2.1 Context-awareness

According to Dey et al. (2001) context may be de-

fined as "any information that might be used to

characterize the situation of entities (person, place

or object) that are considered relevant to the inter-

action between a user and an application". Recent

works proposed variations on definition of context

(Makris et al., 2013), where context is defined as the

flow of information, measured and inferred about

the general state of the related entities.

Consequently, a system considered context-

sensitive should be able to deal with various levels

of abstraction involved in sensitivity to context (Ma-

ran et al., 2014). It should start in the lowest level of

abstraction, collecting and aggregating information

from sensors, passing through intermediate levels,

making inferences about data and information gen-

erated from sensors with semantic annotations that

define contexts themselves, until the highest level of

abstraction, where situations define semantic rela-

tionships between contexts of interest. In ubiquitous

architectures, sensitivity to context directly inter-

feres in many operations.

Integration between ontologies and databases has

been presented in several researches, and some of

them are described in next section. Due to the large

amount of data collected from the environment and

the necessity of subsequent queries to these data, it

is necessary to integrate ontologies with databases.

2.2 Related Work

Comparative analyses between tools to integrate

ontologies and databases were presented in recent

works (Klein et al, 2007). Integration tools can be

classified into: (i) integration with relational data-

bases, (ii) integration with NoSQL databases, and

(iii) integration with distributed file systems.

Regarding integration with purely relational da-

tabases, it was observed a significant increase in

amount of information created for maintaining set-

tings of converted OWL-DL files to relational sche-

ma (Batzios and Mitkas, 2009). Furthermore, there

has been a decrease in performance of queries and

operations in database, primarily due to necessity for

constant conversion of query languages. More recent

approaches have worded with the integration of

ontologies with distributed file systems, primarily to

allow better distribution of information. These ap-

proaches contribute to the scalability of systems, but

suffer since they need external agents to work with

the original file system to perform queries and infer-

ences (Neumann et al., 2010).

NoSQL databases have been used in various ap-

plications, including integration with ontologies.

Recent works (Neumann et al., 2010) describe the

integration of XML databases with ontologies serial-

ized in XML or RDF files. The main problem with

this approach is that XML databases often need too

ProactiveDomainDataQueryingbasedonContextInformationinAmbientAssistedLivingEnvironments

611

much memory. This is a big problem when systems

use large ontologies. In DB4OWL, Batzios and

Mitkas (2009) present an object-oriented representa-

tion of OWL-DL ontologies, integrating ontologies

with object-oriented databases.

There are a number of important requirements

that must be considered in the management of in-

formation in AAL environments that are not met by

the studied solutions: (a) Context information is

often represented in ontologies, and domain-specific

information is often represented in relational data-

bases. Therefore, it is necessary to provide ways to

integrate this information and to query them in an

integrated manner (Bolchini et al., 2013); (b) Usage

of proactive methodologies together with reactive

action to manipulate situations of interest demon-

strated good results to execute services (Klein et al.,

2007). However, the focus of these solutions is the

implementation and execution of services, not data

retrieval about the domain and context information

itself. As result of this analysis, we developed a

proactive data retrieval model for AAL.

3 PROACTIVE APPROACH FOR

DATA QUERYING BASED ON

CONTEXT INFORMATION

The proactive approach here presented for data que-

rying based on contextual information is applied in

the SIaaS Middleware showed in (Machado et al.,

2014). In a home care system for smart homes, the

main issue is to identify the essential characteristics

that an AAL system needs to manipulate situations

of interest involving users in their living environ-

ment. Thereafter, the system must: identify different

situations along time in an extensible manner, reac-

tively manipulate the current situation and proactive-

ly manipulate future situations.

In this sense, SIaaS (Figure 1) is responsible for

the management of environmental resources and for

the detection of current and future situations of in-

terest for pervasive applications. In this context,

pervasive applications are software developed by

specialized companies in specific fields, such as

health, surveillance and energy, and they run in the

SIaaS. In fact, they are deployed in the SIaaS, and

these applications contain knowledge regarding

decision-making processes necessary to manipulate

situations of interest. In fact, applications can state

the situations they are interested in and the system

will activate these applications when one situation is

detected. The middleware architecture has three

levels. The lowest level comprises the physical envi-

ronment, where sensors and appliances are located.

The intermediate level is the SIaaS itself, a system

that manages the environment and provides situa-

tions for pervasive applications (the highest level).

As depicted in Figure 1, the SIaaS is composed

of four modules. Besides, it uses a conceptual model

(inside the application manager), which is based on

Ontology Web Language (OWL). The Applications

Manager module allows the installation (i.e., de-

ployment) of a pervasive application and obtains its

context and situations of interest. After one applica-

tion is deployed, this module informs these situa-

tions of interest to the Prediction and Inference

Manager module and notifies the application’s con-

text of interest to the Context Manager module.

Figure 1: SIaaS Middleware.

The Context Manager has tree subsystems: the Con-

text Persistence and Querying Management Service,

the Monitor, which manages sensor produced raw

data, and the Complex Event Processing system.

The Context Persistence and Querying Manage-

ment Service is a new component in the SIaaS archi-

tecture, and generates event instances according to

what is described in the application´s context of

interest. Thus, each time an event occurs, it is sent to

the Complex Event Processing, which uses a pattern

presented in (Machado et al., 2014) to describe

events, and processes event flows to determine if an

event is of interest for an application already de-

ployed in the SIaaS. If a pattern is detected, it noti-

fies the Prediction and Inference Manager module.

The Prediction and Inference Manager is responsi-

ble for performing inference and predictions, and

has two subsystems. The Inference subsystem con-

tains rules to detect the current situation, and the

Prediction tries to determine the probability of a

situation to happen in the future. If a situation hap-

pens now or is predicted to happen in the future, its

ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems

612

corresponding application is started. In this moment,

the pervasive application chooses the most relevant

actions to manipulate the situation detected and

requests the Action Manager component to execute

these actions.

The Action Manager component then selects the

services that result in the most appropriated func-

tionalities to execute the actions specified by the

application. This selection process takes into account

the user context, comprising user disabilities, for

instance. In this paper, we will focus on the Context

Manager module, more specifically in the Context

Persistence and Querying Management Service. We

will present a solution for data querying based on

contextual information to generate high-level infor-

mation that results in situations of interest.

3.1 Context Persistence and Querying

Management Service

To allow persistence and retrieval of context and

domain information, a set of services that can be

accessed through an API provided by the SIaaS

middleware was defined. Contextual information is

considered dynamic, inferred on raw data from read-

ings taken on environment (Makris et al., 2013).

Domain information is usually more static compared

contextual information and represents information

that describes specific domain of application. This

type of information is usually represented and stored

in relational databases (Klein et al., 2007).

The Context Persistence and Querying Manage-

ment Service consists of four functionalities, which

allows context information to be stored and retrieved

according to the occurrence of specific events that

determine situations of interest. The subsystem con-

sists on the following components: (i) Database

with native JSON support: a database that supports

JSON serialization to store ontologies implemented

in OWL-DL that describes contextual information

was used. In addition, this database must support the

relational model, which allows domain information

to be stored in the same database; (ii) Management

Service: a service that performs the management of

databases used by the Context Persistence and Que-

rying Management Service. It implements routines

that manage the communication between persistence

of information and OWL-API (Horridge and Bech-

hofer, 2011), and Pellet (Parsia and Sirin, 2004); (iii)

Query Processor: Implements the control of que-

ries. This module receives requests made by the

SIaaS and translates the queries to a form compati-

ble with the database used by the subsystem. Thus, it

is possible to perform the integration between con-

textual and domain data, a problem researched by

other works (Bolchini et al., 2013); (iv) Communi-

cation Service: Provides a communication interface

to external modules, and is able to perform query

operations like insert, update and delete data in the

service.

3.2 Context Modeling and

Serializations of Ontologies

For context modeling, we have used the model pre-

viously presented by Machado et al. (2014) and

Silva et al. (2014) (see Figure 2). The conceptual

model begins with Entity. Entities are concrete or

abstract concepts used to reason about a domain of

interest, for instance, person, space, time, and sen-

sor. So, the environment is represented by a set of

entities and their semantic relations, which charac-

terize the context of the environment. Then, seman-

tic relations are a very important concept, and are

represented by triples in the form

<Es, p, Eo>. In

this triple, Es represents the subject of the relation,

and Eo represents the object of the relation. Subjects

and objects are linked by a contextual predicate.

The contextual predicate was first described by Ye

et al. (2012), and it links two contextual entities

through a relation. Using this kind of semantic rela-

tion, pervasive applications can define contexts of

interest.

Thus, pervasive applications can use this kind of

statement to define a set of semantic relations of

interest. Therefore, the context of interest is a subset

of instances with their corresponding semantic rela-

tions. When these relations are evaluated, it is possi-

ble to determine which actions the applications will

use. Figure 2 shows the core of the model, including

domotic, user and proactive domains. Initially, when

users are involved with his daily activities, they

perform (human) actions. These actions result in

external events collected by the AAL system.

Events start and finish the current situation and

influence predictive situations that involve the user

at the current time. The events that influence a pre-

dictive situation (future situation) that will involve

the user are always processed by the Pervasive Ap-

plication. In this contextual model, Semantic Web

Rule Language (SWRL) rules can be used to specify

a current situation to be detected by the system. In

addition, probabilistic values can be modeled using

PR-OWL in order to determine if a situation is hap-

pening now (isSituationOf) or will happen in the

future (willBeSituationOf). Using information about

the current or future situations, the system can select

Automated Actions to manipulate situations.

ProactiveDomainDataQueryingbasedonContextInformationinAmbientAssistedLivingEnvironments

613

Figure 2: Ambient Assisted Living Ontology Network.

For instance, if an interaction with a Person is neces-

sary to manipulate a situation, the system may

choose an Automated Action of the type Regarding

a Person to be executed by a Functionality provided

by the Device of type InteractionWithPerson. This

Functionality needs to be sensitive to Disability of

Person. The Automated Action produces an Internal

Event and using it, the system can detect if the cur-

rent or future situation change or will change in

relation to a Person. In AAL, information related to

the domain refers to resources, people and services,

and may be used by systems (Sadri et al., 2011).

We modeled personal information about patients

according to a list of information considered confi-

dential for patient according to CDT (2014). Figure

3(a) shows an ER diagram of data structure of per-

sonal information about patients in AAL. The dia-

gram (presented in Figure 3(a)) includes information

related Patients and their medical information, rep-

resented by PhoneNumber, MedicalRecord and

Email. In addition, people can use devices that are

registered in the Device table and can use health

services of external companies, defined in

HealthService table. Thus, implemented service can

provide confidential health information only if the

health service is considered safe, or it can filter this

information according to the service used by the

patient. To integrate context with domain-specific

data, we defined an alignment between context in-

formation ontology and a specific domain ontology,

which describes structure and individuals; this

alignment is persisted in the relational database. To

represent the domain structure, we exported the

schema of relational database for a simple ontology,

where Device, Patient and HealthService tables

were defined as classes in ontology (Figure 3(b)).

Furthermore, normal and reverse relationships were

modeled between classes. To align the ontologies

that describe the application domain (a) and the

context definitions (b), a third ontology was imple-

mented (c).

It imports other ontologies and implements

sameAs relationships between context.Device and

domain.Device, and between context.Person and

domain.Patient in addition to the relation between

domain.HealthService and context.InteractionWith-

Organization. Thus, we make connection between

definitions of context and domain information, both

used in data querying. Figure 4 shows the different

levels of abstraction used in SIaaS. The Situation

level presents a situation of interest for pervasive

applications. Situations are represented on a time

interval (T1 and T2) where a situation is valid, and

start and end Events are described using SWRL.

The definition of contexts, and rules that describe

the starting and ending events of situations, as well

as the definition of temporal questions regarding

situations are modeled in a context Ontology.

A form of serialization of OWL-DL ontologies

was implemented using JSON-LD. Based on this

modeling, we extended the model to a scheme de-

scribed in JSON-LD, where each axiom defined in

OWL-DL was converted to JSON-LD in a segment-

ed manner. This file segmentation allows the data-

base to be easily replicated thus increasing scalabil-

ity. In addition, segmentation of axioms allows them

to be shared through REST requests. To perform the

persistence of instances and ontologies, we defined a

set of database schemas, one for domain data persis-

tence (shown in Figure 3(c)), and another for storage

of ontologies that represent context, domain infor-

mation and the alignment between them (Figure

3(b)). Each table in this schema represents a set of

individuals in ontologies separated by their structur-

al type (according to the OWL-DL definitions). All

tables have an attribute used as a unique identifier

(_id), and content (_content), which represent the

definition of axiom in JSON.

ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems

614

)

 

Figure 3: (a) Relational diagram with information about patients and health plans in AAL Systems; (b) Alignment between

context and domain data; (c) Database schema used in ontology serialization.

Figure 4: Abstraction levels of SIaaS.

To manage the model, a set of algorithms to deal

with database schema integration were implemented.

The schema was implemented with PostgreSQL,

which supports the relational model and deals with

JSON. We also implemented an integration with

OWL-API and Pellet. Thus, both axioms of ontology

as information relating to application domain could

be persisted and retrieved in the same database.

3.3 Management of Persistence and

Retrieval of Information

To allow the recovery of information in the service

that implements the model developed in this paper,

we consider the following situations: (i) The Moni-

tor module performs constant monitoring of infor-

mation from sensors. This information should be

grouped according to a set of rules and stored as

inferred context for analysis and subsequent use by

other services and applications. (ii) Based on an

analysis made by Monitor service, it informs Com-

plex Event Processing certain context information,

in conjunction with domain information, which

needs to be used by the Prediction and Inference

Manager to detect situations of interest for the per-

vasive applications deployed on the middleware.

The models presented earlier, of context, domain

and alignment, are implemented in OWL-DL, and

later converted to JSON-LD in the Management

Service of Context Persistence and Query. After,

these settings are inserted into the database. Thus,

ontological representations, implemented in OWL-

DL, of domain and context are persisted and inte-

grated into the OWL API by loading the same in-

formation for inferences in memory.

To insert new context information, the Monitor

performs insertions using the Communication Ser-

vice. When a new context is entered, the service

stores and updates the model in memory, because

this new context can directly affect inferences. To

retrieve domain information based on contextual

information, the management service queries the

database and relates the definitions present in the

ontology with the definitions used in the database.

To validate the model we considered a case study

based on an AAL environment. This case study is

detailed in the next section.

4 CASE STUDY

A case study that represents a set of conditions in an

environment was defined for evaluation purposes.

The case study is based on the union of two applica-

tion domains: Intelligent Environments for health

care (Ambient Assisted Living - AAL), and ubiqui-

tous hospital systems (e-HealthCare Systems).

Let’s consider John, an elderly patient about 70

years old that has a history of cardiovascular disease.

Let’s imagine he is at home, more specifically in his

living room (his current localization). In addition,

his house has a set of sensors, which collect data and

transmit them to the SIaaS middleware, which con-

stantly monitors the environment. John also has a

ProactiveDomainDataQueryingbasedonContextInformationinAmbientAssistedLivingEnvironments

615

caregiver, i.e., a person to help him if something

wrong happens. Suddenly, the sensors start produc-

ing a stream of data and the middleware interpret it

as someone falling and infer that John has just felt.

Immediately, the middleware reacts by notifying

John’s caregiver through his smartphone. In parallel,

as John is in his living room, and his TV (device) is

turned on and is capable of providing alerts, an alert

is sent to it. Since John manages to get up by himself

and the middleware recognizes he is ok, people

registered as next to him are informed.

As John has a history of cardiovascular disease

(e.g., this information is in his profile), the SIaaS,

through an analysis of his past situations, infers a

recurrent pattern, i.e., John frequently falls. In this

behavior pattern, John needs hospitalization. Thus, it

is necessary to inform to a hospital service previous-

ly hired by him to deal with this situation. At the

time a call to the health service is made, the middle-

ware sends a semantic description of his recent his-

tory of activities and of the contexts of interest

(heartbeat and temperature) involved in recent simi-

lar situations. This semantic description is shared

with the system of the hospital, which appends this

information to the hospital record of John (along

with other sensitive medical information). Infor-

mation actions and contexts, as well as a summary

of medical records are transmitted by some ubiqui-

tous system of the hospital to an ambulance, which

performs a call to John. When an ambulance arrives

at John’s residence, the SIaaS confirms the arrival of

the ambulance to the pervasive application and to

the hospital’s system.

4.1 Reactivity

Let’s say that John hired a monitoring service that

works with the SIaaS middleware. In addition, he

acquired a pervasive application that deals with

Falling and notRaised situations. All models are

loaded in SIaaS and inferences can be performed on

these representations because they are integrated to

OWL-API and to Pellet by the service described in

this work. The situation “John’s Fall” is defined by

a starting event, at time T1, and a final event at T2.

Table 1(a) presents this situation and its associated

SWRL rules defining the starting and finishing

events. As John takes a long time to rise, notRaised

situation is inferred by the system. Table 1 (b) pre-

sents rules that define the events of that situation.

These rules are persisted in the service and loaded to

reasoner, so it is possible to infer when events occur

in the environment. When John falls, the service

triggers the corresponding event, converting models

to JSON, and inserting it in the database. When a

situation is inserted in the database, another trigger

sends a message to module CEP to warn the corre-

sponding pervasive application that a situation of

interest has just happened.

Table 1: (a) Initial and final events of the falling situation;

(b) Initial and final events of the notRaised situation.

Starting

Event (a)

Patient(?p) ∧ SensorFalling(?s)∧

hasDevice(?p, ?s) ∧ SensorFallingOn

(true)∧ TimeMinutesSensorFallingOn

(?t) ∧ LivingRoom(?l) ∧ hasLocation

(?p, ?l) ∧ swrlb:greaterThan(t,10) →

isSituationOf (?p, Fall_i)

Finishing

Event (a)

Patient(?p)

∧ SensorFalling(?s)∧

hasDevice(?p, ?s)

∧

SensorFallingOn(false) → isSituation-

Of(?p, Fall_f)

Initial

Event (b)

Patient(?p) ∧ isSituationOf(?p,

Fall_i)∧TimeMinutesSensorFallingOn

(?t) ∧ (not) isSituationOf(?p, Fall_f)

∧ swrlb:greaterThan(t,20) → isSitua-

tionOf (?p, notRaised_i)

Final

Event (b)

(?p, notRaised_i) → isSituationOf (?p,

notRaised_f)

4.2 Proactivity

To query proactively, we used a proactive method-

ology (Machado et al., 2014). There, rules are com-

bined with Bayesian networks to determine the

probability that undesired situations may occur.

Constantly, the module consults the database to

verify if, according to current context, situations of

interest occurred previously. This is done with a

query about situations occurring in similar contexts

in a given time, using concept of time windows

(Machado et al., 2014). In this concept, situations of

interest have a limited time frame and occur in a

limited amount of time before or after the current

situation. The code snippet below shows a query of

situations arising in contexts similar to the current

one. In this case, time entities are taken into consid-

eration, with an interval of 2 hours from the current

time.

SELECT count(*) as situationOccourrenceNum, ontol-

ogies."olindividual"._id,ontologies."olindividual"

._content as SemanticDescription FROM do-

main."Patient", ontologies."olindividual" WHERE

CAST(domain."Patient"._id as text) like ontolo-

gies."olindividual"._content->> 'isSituationOf'

AND ontologies."olindividual" ._content ->> 'has-

StartTime' BETWEEN '10:00' AND '12:00' GROUP BY

ontologies."olindividual"._id

The query result is informed by the CEP service

with a number of occurrences of unwanted situations

that occurred in the given time window and a seman-

tic description (presented below) of the information

regarding the patient if they need to be sent to the

hospital system.

ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems

616

{ "@context": { "#situation1":{"@id":

"https://www.inf.university.br/~main/ontologies/aa

l#situation1.jsonld","@type":"@id"},"aal":

"https://www.inf.university.br/~main/ontologies/aa

l.jsonld",

"rdfs": "http: //www.w3.org/2000/01/rdf-schema#",

"owl":"http: //www.w3.org/2002/07/owl#", "#Pa-

tient":"https://www.inf.university.br/~main/ontolo

gies/aal#Situation.jsonld"},

"@type": "owl:Individual","rdfs:label":[

{ "@value": "situation 1", "@language": "en"},

{ "@value": "situacao 1","@language":"pt-br"}],

"owl:individualOf":[ {"owl:Class":"#Situation"}],

"isSituationOf": "1", "hasStartTime": "10:30",

"hasEndTime": "10:45" }

CEP, in turn, analyzes information sent by the

service, and infers that there are positive probabili-

ties of the unwanted situation CardiacAttack to

happen. Thus, CEP informs the application the in-

formation about situations and medical patient in-

formation, which should be sent to the hospital sys-

tem, with a request for an ambulance.

5 CONCLUSIONS

The combination of actions taken reactively and

proactively increases the efficiency of control sys-

tems in assisted living environments. SIaaS middle-

ware was implemented and offered a model for

making reactive and proactive decisions. In this

paper, we extended the SIaaS middleware adding

new models to perform the integration of contextual

definitions stored in ontologies and domain-specific

information stored in relational databases; and re-

trieves this information in an integrated and proac-

tive manner. These contributions were implemented

in SIaaS middleware in form of a new Context Per-

sistence and Querying Management Service, which

allows SIaaS to query about domain and context

data proactively.

As future work, we intend to expand the integra-

tion between tools of inference of ontologies and the

persistence solution allowing the developers to use

the approach transparently, using the API as ontolo-

gy management interface and to compare this solu-

tion against solutions based on triple stores that are

often used for the persistence of ontologies.

REFERENCES

Augusto, J. C., Nakashima, H., Aghajan, H. 2009. “Ambi-

ent Intelligence and Smart Environments: a State of

the Art”. In: Handbook of ambient intelligence and

smart environments. p. 3–31. Springer New York.

Batzios, A., Mitkas, P. A. 2009. "db4OWL: An Alterna-

tive Approach to Organizing and Storing Semantic

Data." In: Internet Computing, IEEE. v.13, n.6, p.48-

55.

Bolchini, C., Quintarelli, E., Tanca, L. 2013.“CARVE:

Context-aware Automatic View Definition over Rela-

tional Databases”.In:Information Systems,v.38,i.1.

Center for Democracy & Technology Website. Available

at: https://www.cdt.org/files/healthprivacy/20090625

_deidentify.pdf.

Dey A. K., Abowd G. D., Salber D. 2001. “A Conceptual

Framework and a Toolkit for Supporting the Rapid

Prototyping of Context-aware Applications” In: Hum.-

Comput. Interact. n. 16, p. 97–166.

Horridge, M., Bechhofer, S. 2011. "The OWL API: A Java

API for OWL Ontologies." In: Semantic Web. p. 11-

21.

Klein, M., Schmidt, A., Lauer. R. 2007. "Ontology-

centered Design of an Ambient Middleware for As-

sisted Living: The Case of Soprano.". In: 30th Annual

German Conference on Artificial Intelligence.

Lanthaler, M., Gutl, C. 2012. "On Using JSON-LD to

Create Evolvable RESTful Services" In: Proceedings

of the Third International Workshop on RESTful De-

sign. p. 25-32.

Machado. A , Lichtnow, D., Pernas, A. M., Wives, L. K.,

Palazzo Moreira de Oliveira, J. 2014. “A Reactive and

Proactive Approach for Ambient Intelligence”. In:

16th International Conference on Enterprise Infor-

mation Systems, Lisbon. v. 2. p. 501-512.

Makris, P., Member, S., Skoutas, D. N. 2013. "A Survey

on Context-Aware Mobile and Wireless Networking:

On Networking and Computing Environments' Inte-

gration." In: Communications Surveys & Tutorials,

IEEE. v. 15.1. p. 362-386.

Maran, V., de Oliveira, J., Augustin, I. 2014. "Are The

Integrations Between Ontologies and Databases Really

Opening the Closed World in Ubiquitous Compu-

ting?" In: International Conference on Software Engi-

neering & Knowledge Engineering. Vancouver, CA.

Neumann, C. P., Fischer, T., Lenz, R. 2010. “OXDBS –

Extension of a Native XML Database System with

Validation by Consistency Checking of OWL-DL On-

tologies” In: Proceedings of the 14th International

Database Engineering & Applications Symposium.

Parsia, B., Sirin, E. 2004. "Pellet: An OWL DL Reasoner."

In: Third International Semantic Web Conference-

Poster. no. 18. Pp. 51-53.

Sadri, F. 2011. “Ambient Intelligence: A survey”. In:

ACM Computer. v. 43, n. 4, p. 36-70.

Silva Kambara. J, Machado Medeiros, G, Thom, L. H.,

Krug Wives, L. 2014. “Business Process Modeling

and Instantiation in Home Care Environments” In:

16th International Conference on Enterprise Infor-

mation Systems, Lisbon. v. 2. p. 513-525.

Ye, J., Dobson, S., McKeever, S. 2012.“Situation Identifi-

cation Techniques in Pervasive Computing: A Re-

view”.In: Pervasive and mobile computing,v.8,n1.

WorldMeters. Current World Population. Website. Avai-

lable: http://www.worldometers.info/world-

population/

ProactiveDomainDataQueryingbasedonContextInformationinAmbientAssistedLivingEnvironments

617