A Semantic-based Approach for Facilitating Arbovirus Data Usage

Aparecida Santiago

, André Alencar

, Amanda Souza

, Erika Araruna

, Isabel Fernandes

and Damires Souza

Academic Unit of Informatics, Federal Institute of Paraiba, João Pessoa, Brazil

Medical Science Center, Federal University of Paraíba, João Pessoa, Brazil

damires@ifpb.edu.br

Keywords: Semantics, RDF Data, Data Usage, Metadata Reuse, Arbovirus Data.

Abstract: Today’s continuous growth for healthcare information entails an increasing need for using large amounts of

data. Particularly, the incidence of arboviruses has been on the rise in some countries, what causes specific

needs for studies and definitions of public strategies. In this light, providing a computational platform for

usage and reuse on arboviruses related data may help matters. The idea is that different applications and

users can make use of that data in diverse ways. In this work, we propose a semantic-based approach for

facilitating use and reuse of arboviruses related data. We present the definitions underlying our approach,

examples illustrating how it works, and some promising results we have obtained.

1 INTRODUCTION

Today’s continuous growth for healthcare

information entails an increasing need for using

large amounts of data, which may come from

different data sources. Particularly, the data scenario

associated with vector-borne diseases is currently

causing specific needs for studies and definitions of

public strategies in some countries.

Vectors are living organisms that can transmit

infectious diseases between humans or from animals

to humans. Mosquitoes are examples of vectors. The

Aedes aegypti mosquito is a vector, which transmits

four different diseases commonly called as

arboviruses (Fletcher, 2017), as follows: Dengue

fever, Yellow fever, Zika virus and Chikungunya

fever. Given the spread of these diseases in some

countries, it is necessary to improve control

strategies. Providing data sharing and usage, by

means of a computational platform, may enhance the

development of applications and data analytics in

such a way that healthcare managers and doctors

may take important decisions.

A large amount of information on arboviruses is

available on the Web in sources such as sites. Data

are also found in specific databases, many of them

local to some hospitals. With this diversity of data

sources, with their own terminological definitions, it

is hard for a computational application to solve the

conflicts arising from the existing heterogeneities

and achieve a common understanding of the data

(Bansal and Kagemann, 2015).

To help matters, it is necessary to collect and

integrate existing data on arboviruses and share them

in a way that makes them feasible for easier usage.

In order to make the computational effort smaller in

the development of a solution, some principles and

technologies derived from the Semantic Web (Heath

and Bizer, 2011) can be employed. At first, the data

should be described semantically, i.e., according to a

common understanding, what facilitates their

processing and reuse (Bansal and Kagemann, 2015).

To this end, it is necessary to choose and employ a

domain vocabulary in order to provide semantic

reference to the data. These data are usually

converted to the RDF data model in order to enable

the semantic description (Lóscio et al., 2017).

With this scenario in mind, we define two main

research problems that have guided our work, as

follows: (i) How to provide a standard vocabulary

on arboviruses so that researchers, doctors,

healthcare managers as well as software agents can

use it as a reference for data conversion and

sharing? And (ii) Given arbovirus related data,

semantically described in RDF, can they be used in

order to facilitate the development of useful

applications on diseases control?

Santiago, A., Alencar, A., Souza, A., Araruna, E., Fernandes, I. and Souza, D.

A Semantic-based Approach for Facilitating Arbovirus Data Usage.

DOI: 10.5220/0006799706710678

In Proceedings of the 20th International Conference on Enterprise Information Systems (ICEIS 2018), pages 671-678

ISBN: 978-989-758-298-1

671

In this work, we present a semantic-based approach

that aims to facilitate usage and reuse of arboviruses

related data and metadata. Semantic technologies are

employed for modelling relevant information by

means of an ontology, which implements the domain

vocabulary. The approach includes a tool, which is

able to convert CSV data into RDF. In order to

verify the usefulness of the converted data, a web

application, which provides arbovirus information

visualization, has also been developed and

evaluated. In addition, some experiments have been

accomplished.

Our contributions are summarized as follows: (i)

we introduce the ARBO ontology; (ii) we propose a

semantic-based approach to convert arbovirus data

into RDF ones; (iii) we present an application, which

provides useful information based on the produced

RDF data; and (iv) we describe accomplished

evaluations w.r.t. the proposed approach.

The remainder of this paper is organized as

follows: Section 2 introduces some background

concepts, a motivating scenario and related work;

Section 3 presents the proposed approach; Section 4

shows some obtained results and describes the

accomplished evaluations; Finally, Section 5 draws

our conclusions and points out some future work.

2 CONCEPTS, SCENARIO AND

RELATED WORK

In this section, we provide some concepts and

recommended practices for sharing data on the Web.

We also provide a motivating scenario and discuss

some related works.

2.1 Data on the Web

The Web has evolved into an interactive information

network, allowing users and applications to share

data on a massive scale. To help matters, the

Semantic Web and the Linked Data principles define

a set of practices for publishing structured data on

the Web aiming to provide an interoperable Web of

Data (Heath and Bizer, 2011). These principles are

based on technologies such as HTTP, URI and the

RDF data model. By using the RDF model, data or

resources are published on the Web in the form of

triples (composed by a subject, a predicate and an

object). Each resource is identified by means of an

URI. In order to achieve this, it is necessary to

convert data, which are originally in other format

(e.g., CSV), to RDF data.

In order to make data available and feasible for

reuse, another semantic web principle is to organize

data in such a way that they can be interpreted and

used meaningfully without human intervention

(Bansal and Kagemann, 2015). This is achieved by

adding data about data, i.e., by adding metadata to

refer semantically the data.

To clarify matters, the World Wide Web

Consortium (W3C) defines some best practices to

facilitate sharing data on the Web (Lóscio et al.,

2017). These best practices cover diverse aspects

related to data publishing and consumption, like data

formats, data access, data identification and

metadata provisioning. One of the recommendations

regards the use of open domain vocabularies in order

to semantically refer the data, when data are

converted to RDF. To this end, it is essential to take

into account the knowledge domain (e.g., “Health”,

“Music”) in which the data exist and choose the

appropriate domain vocabularies. Vocabularies are

usually developed as ontologies, which represent a

formal, explicit specification of a conceptualization

(Gruber, 2009). An ontology provides definitions of

terms in a given data domain as well as the

relationships that link these terms to each other.

Other W3C recommendation regards facilitating

data consumption. In this sense, it is important to

make data available through APIs (Application

Programming Interfaces), developed for such

purpose, especially if data are large, frequently

updated, or highly complex.

2.2 Motivating Scenario

Collecting and integrating data on diseases, such as

arboviruses, become relevant to some specific

applications, particularly in times of their high

incidence in some countries. We have observed the

need of data analytics on these diseases not only by

healthcare agency managers but also by healthcare

professionals. They have to plan and study

preventive measures in order to fight diseases

occurrences and consequences.

Some data on arboviruses are already published

on the Web as open data. Nevertheless, in some

governmental states as ours, there are no open data

portals with such data. In this work, we have

obtained data directly from the state healthcare

agency. As an illustration, excerpts from the

obtained data are depicted in Figure 1.

Lines in Figure 1 represent patients and

occurrences of disease notification (dengue or

chikungunya). For each patient, symptoms (most

columns) are set according to medical anamnesis.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

672

Existing symptoms are included as “1” value; on the

other hand, if a given symptom is not present in

patient complains, it is defined as “2” value. To

facilitate understanding, we present the english

meaning of the symptom terms present in the data

sets (properly ordered), as follows: fever (febre),

myalgia (mialgia), headache (dor de cabeça),

exanthema (exantema), vomit (vômito), arthritis

(artrite), arthralgia (artralgia), petechiae (petequia),

leukopenia (leucopenia) and tie proof (prova do

laço).

Figure 1: Excerpts from Real Arbovirus Data.

In order to have an integrated view of the data from

the data sources, it is necessary to deal with phases

such as data Extraction, Transformation and Load

(ETL) (Bansal and Kagemann, 2015). Each phase

has specific technical issues to be addressed. To

facilitate this process, identifying the relevant data to

extract, creating feature extractors and converters,

and building a domain vocabulary to align the data

are usual steps to be done.

We use the presented data scenario for

motivating this work and also for demonstrating how

the proposed approach works in a real-world data

environment. Nevertheless, the proposed approach

may be instantiated in any arbovirus data scenario.

2.3 Related Work

Literature about disease ontologies is not new and

some works already provided useful artifacts. In this

section, we briefly resume some relevant work in

this data domain. We also discuss works regarding

data conversion, semantic platforms for health and

use of data on applications and analytics.

Some studies have been carried out on the

creation of health ontologies, such as the IDODEN

(ontology for Dengue) and the IDO (Infectious

Disease Ontology) regarding viruses in general

(Bioportal, 2017). In addition, other ontologies

related to this knowledge domain are IDOMAL, an

ontology on malaria information, which extends

IDO, and the MEDDRA, MESH and SNMI, which

are ontologies that provide a terminology for

cataloging medical information (Bioportal, 2017).

The ontologies most related to our work are the

IDO, IDODEN and DOID. IDO is a consortium of

infectious disease ontologies, among which are

currently being developed ontologies for Dengue

and Malaria. IDODEN is an ontology for Dengue,

addressing clinical aspects, which extends the IDO

ontology. The DOID ontology is defined by

following a scope of human diseases in general.

With respect to general computational

approaches, Dragoni et al. (2017) developed an

architecture for supporting the monitoring of people

and for persuading them to follow healthy lifestyles.

To this end, they used semantic technologies for

modeling relevant information and for fostering

reasoning activities. Chun and MacKellar (2012)

developed a system, which integrates information

from some sites such as PatientsLikeMe (Patients,

2018) and PubMed (Pubmed, 2018). It can be used

to annotate a variety of text based blogs.

Regarding tools developed to provide data

conversion to RDF, the LinkMapia application

(Sacenti and Fileto, 2014) is an example. It converts

geographic data into linked data. It also filters data

to align them with existing collections of linked

data. Regarding applications which consume data on

diseases, two examples are provided. Kaieski

developed the Vis-Health application, an open

source system in which records from public health

are used to provide some analyses (Kaieski, 2014).

Varela (2016) implemented an application for

tracking and presenting data on arboviruses, with the

purpose of informing, by using maps and charts the

list of hospitals that received infected patients.

These last two applications do not deal with RDF

data, differently from ours. Also they produce

distinct kinds of data analyses. Our approach uses a

specific ontology to assist real data to be converted

to integrated RDF data as part of the transformation

step of an ETL process. Based on the produced RDF

data, an information visualization application has

been developed as a means to validate data

consuptiom and usage.

3 PROPOSED APPROACH

Vocabularies provide the semantic glue enabling

data to become more meaningful data. With the

emergence of open vocabulary repositories, many

vocabularies are being published and similar ones

are being grouped together usually on the web.

Examples of such repositories are the Linked Open

A Semantic-based Approach for Facilitating Arbovirus Data Usage

673

Vocabulary (LOV, 2017) and, more specifically, the

Bioportal, which is related with the health

knowledge domain (Bioportal, 2017).

As a result, finding a suitable vocabulary for

publishing a specific dataset in RDF has become

easier, although it is usually necessary to select one

with a wide consensus in the community. However,

in case of unavailability of a suitable one, or when it

does not cover completely a given set of knowledge

domain terms, it is necessary to build a new one and

reuse terms which already have been defined.

Although there are some specific vocabularies

regarding arboviruses such as dengue, to the best of

our knowledge, we could not find specific ones

related to the recent arboviruses, i.e., to chikungunya

fever and zika virus. As a result, we have worked on

an ontology, which covers the domain of arboviruses

in terms of general kinds of diseases, symptoms,

signals, exams, severity of the diseases and other

additional information. All developed ontology

terms were suggested by medical specialists. In the

following, we present the ARBO ontology. Then we

describe how the ontology is used along with the

data conversion and publication process.

3.1 The ARBO Ontology

Based on some methodologies of ontology

engineering (Sure et al., 2009), we have instantiated

an iterative and incremental process to develop the

ARBO ontology. The ontology building process

includes the following steps:

I. Determination of the knowledge domain

and its scope: in this work, the knowledge domain

refers to the viruses group named as arboviruses.

II. Enumeration and definition of important

terms w.r.t. concepts and properties of the domain at

hand (conceptual domain model).

III. Survey of existing and relevant

vocabularies for allowing the reuse of some terms.

IV. Definition of classes, hierarchies and

properties. A mapping between the candidate terms

and the terms identified in the domain conceptual

model was performed.

V. Validation of the ontology terms by

domain experts. In our case, healthcare doctors and

researchers have provided such validation.

The ARBO ontology makes reuse of terms from

existing ontologies such as IDOMAL, IDODEN and

DOID instead of creating duplicates of terms. Thus,

it ensures interoperability with already existing

infectious disease ontologies. Table 1 presents the

list of reused ontologies and the number of terms

reused from each one. Additionally, it shows the

number of specific domain terms (155), which have

been originally created in the ARBO ontology. By

adding all the terms reused and created, the ARBO

ontology is now composed by 218 terms, of which

63 are reused and 155 were newly created for this

ontology. The main concepts of the ARBO ontology

are depicted in Figure 2, according to the ontograf

notation (Ontograf, 2017).

Table 1: Vocabularies and the Number of Reused or

Created Terms.

Ontology

Number

of Terms

DOID 15

IDODEN 8

IDOMAL 3

SYMP 13

MESH 5

MEDDRA 17

SNMI 1

SNOMEDCT 1

ARBO 155

The ARBO vocabulary comprises some primary

concepts such as Disease, Disease by Infectious

Agent, Viral Infectious Disease, Arboviruses, Pre-

existing diseases, Patient, Symptom, Signal,

Chikungunya, Zika, Dengue and Yellow Fever.

Properties and relationships are defined by means of

data and object properties, respectively.

We have identified these terms with the

assistance of two medical experts in virus caused

diseases. In accordance with their guidance, we have

also differentiated symptoms from signs or signals.

The signs and symptoms described in this work

correspond to the observations from the doctor and

the complaints presented from the patient in the

medical appointment, respectively. The signs refer

to more objective and direct data, which can be

noticed by the doctor, nurses and relatives, along the

physical exam or at home. The symptoms, in

contrast, are subjective and only can be described by

the patient, as the characteristics or manifestations of

the disease in his/her body.

At the ARBO documentation (ARBO, 2017), we

depict terms that have been defined not only to

symptoms but also to signs. Both are very important

since they provide means to doctors to understand

patient complains. The ARBO ontology will be

published in the main open vocabulary repositories

such as LOV and Bioportal. Descriptive metadata

have been included in the ARBO ontology in order

to provide information such as the ontology creators,

publisher, version number, and date of publication.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

674

Figure 2. Main Concepts of the ARBO Ontology.

3.2 The Data Publication Process

The main idea underlying our approach is to bring

the knowledge domain semantics into the data ETL

process aiming to facilitate data publication and

consumption. The activity of converting different

data sources on arboviruses produces an integrated

view of the data defined in terms of a given domain

vocabulary. In this work, we use the ARBO

ontology to provide that means.

The proposed semantic data publication process

consists of the three ETL major phases along with

the data publication and consumption phases, as

depicted in Figure 3. The use of semantic

technologies is introduced in the Transformation

phase as a means to enhance data conversion. Phases

are discussed in the following.

Figure 3: The Data Publication and Consumption Process.

Data Extraction

In the data extraction step, instance data along with

their properties (metadata) are extracted from

existing datasets. In this work, CSV datasets are

considered. Data cleaning tasks are applied in order

to prepare the data. Metadata are identified

according to the names of the properties (columns),

which compose the CSV file. A developed extractor

provides the selection of the properties and data to

be converted.

Data Transformation

In this step, the extracted metadata along with their

corresponding data are converted to RDF triples. At

first, the matching of extracted CSV metadata

against the ARBO vocabulary terms is done. To

assist this process, a user known as Domain Expert

(DE) is needed. The DE has an understanding of the

content to be converted and the knowledge domain

underlying the data. The DE assists the matching

activity by pointing out the correspondences

between the extracted metadata and the ontology

properties. The output of the matching process is

called an alignment. It contains a set of equivalence

correspondences indicating which properties

correspond to each other. Examples of these

correspondences are shown in Figure 4. This

alignment is saved and used later. Then, based on

the identified correspondences, for each property

(e.g., symptoms, signs) and, for each row (e.g.,

patients), RDF triples are generated.

Figure 4. Examples of correspondences between original

data properties and the ARBO terms.

Data Load and Publication

The generated RDF dataset is persisted in an RDF

store and made available on the web as linked data.

This means that it is available for querying via a

SPARQL endpoint.

Data Consumption and Visualization

The RDF dataset on arboviruses has a SPARQL

endpoint that allows its consumption. Developers

ID_MUNICIP  arbo:hasCity

DT_NOTIFIC arbo:dataOfNotification

A Semantic-based Approach for Facilitating Arbovirus Data Usage

675

have programmatic access to the data on arboviruses

for use in their own applications. As an initial

example of a consumption application, the arboviz

application (described in Section 4.1), has been

developed.

4 RESULTS AND EVALUATION

In this section, we describe some implementation

and evaluation results.

4.1 Developed Tools

We have developed the data conversion process

within a tool implemented in PHP. In this version, it

is able to convert CSV files to the RDF model, using

information from the knowledge domain of the data.

Although in this work we have used the ARBO

ontology and datasets provided by the state agency,

the tool is able to receive as input any CSV file

along with the domain ontology to be considered

(any) in order to provide the data conversion. Thus,

it may be used in any data domain.

Regarding the scenario illustrated in Figure 1

(part of the provided datasets), the tool is able to

generate the RDF dataset for each one of the CSV

files. The datasets refer to the years 2015, 2016 and

2017 and to the disease notifications for Dengue and

Chikungunya fever.

As an illustration, we provide an excerpt from an

RDF dataset with respect to the occurrences of

Dengue in 2015 (Figure 5). The dataset is serialized

in RDF/Turtle. In this example, there is one patient

with related data. The patient refers to the resource

idoden:IDOMAL_0000603#0 (subject), which has

two predicates and their respective objects:

arbo:has_city

http://siderg.com.br/arbo/ID_MN_RESI/LASTRO,

and doid:has_symptom

http://siderg.com.br/arbo/CEFALEIA.

We have implemented a data consumption

application named as arboviz (arboviz, 2017). It was

developed in PHP, and it consumes data from the

RDF dataset with information about arboviruses

occurrences in the state of Paraíba, Brazil. It uses a

SPARQL endpoint to this end.

The main goal of arboviz is to provide easy and

accessible information visualization, with

explanatory texts and analytics generated from the

data present in the produced RDF dataset. One of the

produced views is depicted in Figure 6. It depicts the

main symptoms, regarding dengue and chikungunya

Figure 5. Excerpt from a generated RDF dataset.

diseases. To this end, it uses a word cloud, which is

built by considering the most cited symptoms in the

RDF dataset. Each symptom is printed in a given

font and scaled by a factor roughly proportional to

its number of occurrences in the underlying dataset.

Regarding dengue disease, the most referred

symptoms are (properly ordered) the following:

fever, headache, myalgia, arthralgia, arthritis,

nausea, back pain and vomit. Regarding

chikungunya, the most cited symptoms are (properly

ordered) the following: fever, myalgia, headache,

arthralgia, back pain, nausea, arthritis, and vomit.

Figure 6. Example of a Data View in the Arboviz App.

4.2 Evaluation

We have conducted two main types of evaluation to

verify the effectiveness of the proposed approach

and applications. The former regards evaluating if

the ARBO ontology indeed makes difference when

converting the data. The latter is concerned with the

usefulness of the arboviz application and,

consequently, the produced RDF data.

To verify the former claim, two experiments

were done. Firstly, we performed a comparison

among the domain terms which have been used to

compose the ARBO ontology (Figure 7), shown in

Table 1. In addition to the number of terms depicted

in Table 1, we have also considered the number of

common terms between the IDO and the ARBO

ontology, what results in 03 terms. We observe that

some domain terms are rather important to the

composition of the final one (e.g.,

doid:hasSymptom), although most terms (70%) had

to be indeed created in the ARBO ontology, in

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

676

accordance with recommendations provided by

healthcare experts.

Figure 7. Number of Used Terms w.r.t the ARBO

Ontology Composition.

We have also conducted an experiment to verify the

degree of recall regarding the data conversion when

considering some ontologies versus the ARBO one.

In this particular evaluation, we used four CSV files

regarding arboviruses notifications. The first one is

identified as dengue2015 and contains 30.359 rows.

The second one is identified as dengue2016 and

contains 45.114 rows. The third dataset is named as

chikungunya2016 and is composed by 14.026 rows.

The last dataset is called chikungunya2017 and holds

884 rows. As domain vocabularies to be verified, we

have used the IDODEN, IDO and ARBO.

We consider recall as the ratio of correctly found

resources (true positives) over the total number of

expected resources (true positives and true

negatives) (Rijsbergen, 1979). To achieve the

expected number of resources, we have produced

gold standards regarding the RDF data generation

for each evaluated vocabulary versus dataset to be

converted. These gold standards have been manually

produced by participants of our research group. The

recall formula is presented in the following.

Where

#CorrectResources is the number of correct returned

resources (URIs);

#ExpectedResources is the total number of all possible

resources (URIs) that could be generated.

A summary of the results regarding the recall

measure along with the usage of IDODEN, IDO and

ARBO vocabulary for the four datasets is shown in

Figure 8. We are able to observe that the usage of a

suitable domain vocabulary makes all the difference.

In this work, we have defined a specific vocabulary

by making reuse of recommended terms when

possible. New terms which belong to health data

sources have been defined in the ARBO Vocabulary,

based on expert advice. As a result, it has covered

almost 100% of the required data.

Figure 8. Recall w.r.t. the choice of a domain vocabulary

on arboviruses.

In addition, we have invited some users to evaluate

the arboviz application w.r.t. its usefulness. The user

group was composed by a total of 24 persons by

means of general users (56%), healthcare managers

(22%) and healthcare professionals (22%). To

perform the evaluation, they used the arboviz

application and then they filled out a questionnaire.

They filled out a questionnaire stating their

opinions on the interface usability, the provided

information views and also on the ontology terms (in

case of healthcare professionals). They were also

asked to provide comments pointing out what they

most liked or not, along with suggestions. Among

the presented questions, four main ones are depicted

in Figure 9 with their respective answers.

Figure 9. Evaluation of the arboviz application w.r.t. its

usefulness.

In terms of easiness of use, information views and

content presentation, the users provided a great

impression. Particularly in terms of whether there is

useful information, only two persons considered the

application not providing that feature. In case of

healthcare professionals, they were also asked if the

ARBO vocabulary covers the major terms in the

area. All of them confirmed that the ontology is

complete w.r.t. the arboviruses domain area.

0,820

0,840

0,860

0,880

0,900

0,920

0,940

0,960

0,980

1,000

1,020

1234

Recallw.r.t.thechoiceofadomainvocabulary

RecallIDODEN

RecallIDO

RecallARBO

A Semantic-based Approach for Facilitating Arbovirus Data Usage

677

Regarding what the users most liked, they said:” the

application is simple and easy of finding

information; the symptoms chart, which shows the

symptoms associated with the body parts, is really

interesting, the presented content is useful and

integrates information from different sources in

unified views”. The biggest complaints were

regarding the “lack of responsiveness of the

application interface”.

5 CONCLUSIONS AND

FURTHER WORK

The publication and consumption of data on

arboviruses is indeed an important issue. Based on a

built domain ontology, we have developed an ETL

data process, which provides publication and

consumption of arbovirus related data and,

consequently, facilitates their usage. Data published

in RDF have potential to be used in many ways and

to facilitate the creation of data-driven applications

such as the described arboviz application.

Accomplished experiments show that our

approach is promising. By using the proposed

domain ontology, it is able to produce a recall

measure regarding data conversion of almost 100%

w.r.t. the source data. In addition, an evaluation

carried out with real users showed that the arboviz

application is useful since it provides kinds of

information views that unify data by presenting them

in different perspectives.

For future work, we intend to convert more real

data on arboviruses and deal with other data

analytics on them. In addition, the ARBO ontology

will be exposed as a web service.

REFERENCES

ARBO Documentation, 2017. Available at

http://siderg.com.br/arboviz/ontology/views.php. Last

access on December, 2017.

Arboviz, 2017. Available at http://siderg.com.br/

arboviz/index.php. Last access on December, 2017.

Bansal, S., Kagemann, S., 2015. Semantic Extract-

Transform-Load framework for Big Data Integration.

In Computer 48(3):42-50.

Bioportal, 2017. Available at https://bioportal.

bioontology.org/. Last access on December, 2017.

Chun, S., MacKellar, B., 2012. Social Health Data

Integration using Semantic Web. In Proceedings of the

27th Annual ACM Symposium on Applied Computing.

SAC.

Dragoni, M., Bailoni, T., Eccher, C.,Guerini, M.,

Maimone, R., 2017. A Semantic-enabled Platform For

Supporting Healthy Lifestyle. In Proceedings of the

ACM Symposium on Applied Computing. SAC.

Fletcher, J., 2017. Arboviruses: Types, symptoms, and

transmission. Available at

https://www.medicalnewstoday.com/articles/318645.p

hp. Last access on December, 2017.

Gruber, T., 2009. Ontology. Encyclopedia of Database

Systems, pp 1963-1965.

Heath, T., Bizer, C., 2011. Linked Data: Evolving the Web

into a Global Data Space, Morgan & Claypool

Publishers. 1

edition.

Kaieski, N., 2014. Vis-Saúde - Uma metodologia para

visualização e análise de dados de saúde pública.

Master Thesis, UNISINOS.

Lóscio, B., Burle, C., Calegari, N., 2017. W3C Data on

the web best practices. Available at:

https://www.w3.org/TR/dwbp/. Last access on

December, 2017.

LOV, 2017. Linked Open Vocabularies. Available at:

https://lov.okfn.org/dataset/lov/. Last access on

December, 2017.

Ontograf, 2018. The Ontograf plugin. Available at

https://protegewiki.stanford.edu/wiki/OntoGraf. Last

access on January, 2018.

Patients, 2018. Available at https://patientslikeme.com/.

Last access on January, 2018.

Pubmed, 2018. Available at

https://www.ncbi.nlm.nih.gov/pubmed/. Last access

on January, 2018.

Rijsbergen, C., J., 1979. Information Retrieval, MA:

Butterworths. Stoneham, 2

edition.

Sacenti, J., Fileto, R., 2014. LinkMapia: Uma Abordagem

para Converter Dados Geográficos Livremente

Anotados em Dados Ligados. In: X Escola Regional

de Banco de Dados, 2014.

Sure, York., Staab, Steffen., Studer, Rudi., 2009. Ontology

Engineering Methodology. In Staab, Steffen & Studer,

Rudi (eds.) Handbook on Ontologies. Springer-Verlag,

Heidelberg, 2

edition. ISBN 978-3-540-70999-2.

Varela,V., 2016. Rastreamento endêmico da dengue, Zika

e chikungunya via Android e sistema de informação

geográfica (SIG). Undergraduate Work. UNB.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

678