A Novel Query Language for Data Extraction from Social Networks

Francesco Buccafurri

, Gianluca Lax

, Lorenzo Musarella

and Roberto Nardone

DIIES, University Mediterranea of Reggio Calabria, Reggio Calabria, Italy

Keywords:

Multiple Social Networks, Web Analysis, Query Language, Data Extraction.

Abstract:

Online Social Networks (OSNs) represent an important source of information since they manage a huge

amount of data that can be used in many different contexts. Moreover, many people create and manage

more than one social proﬁle in the different available OSNs. The combination and the extraction of the set

of data from contained in OSNs can produce a huge amount of additional information regarding both a single

person and the overall society. Consequently, the data extraction from multiple social networks is a topic of

growing interest. There are many techniques and technologies for data extraction from a single OSN, but there

is a lack of simple query languages which can be used by programmers to retrieve data, correlate resources

and integrate results from multiple OSNs. This work describes a novel query language for data extraction

from multiple OSNs and the related supporting tool to edit and validate queries. With respect to existing lan-

guages, the designed language is general enough to include the variety of resources managed by the different

OSNs. Moreover, thanks to the support of the editing environment, the language syntax can be customised by

programmers to express searching criteria that are speciﬁc for a social network.

1 INTRODUCTION

Over the past decade, Online Social Networks (OSNs)

have become widely used by a large set of users in ev-

eryday life. Most people create and manage proﬁle in

one or more OSNs like Facebook, Twitter, Linkedin,

Instagram. Many people spend a lot of time creat-

ing many contents on every-day life, latest news, pol-

itics, economics, sport, and publish many informa-

tion (posts, photos, videos, etc.) about their passions,

travels, friends. As a consequence, OSNs represent

a source of a huge amount of data, offering a great

variety of information spanning from personal infor-

mation to users interests. For this reason, OSNs are

recognised as an important phenomenon from a so-

cial and economic point of view, and, thus, it is mat-

ter of study for the design and the development of

innovative and modern web applications. In many

cases, both personal information and social interac-

tions coming from social network proﬁles can be part

of innovative software solutions. Among these, social

Web applications are the most signiﬁcant example in

which both peoples’ identities and contents they pro-

https://orcid.org/0000-0003-0448-8464

https://orcid.org/0000-0002-5226-0870

https://orcid.org/0000-0002-9616-7256

https://orcid.org/0000-0003-4938-9216

duced are involved in the business process and data

are mostly owned by users, strongly interlinked and

inherently polymorphic (Bell, 2009).

To allow the data extraction, some of the social

network providers offer different APIs and frame-

work that reﬂect their internal metadata. In addition,

some other tools and well-known techniques (e.g.,

web scraping) have been proposed in literature with

the aim of extracting data from OSNs. Hence, when

a developer needs to interact with different OSNs,

she/he needs to manage different techinques and lan-

guages. In fact, there is a lack of standard languages

and protocols to query both heterogeneous OSNs and

multiple OSNs with the same language. In this di-

rection, this paper introduces a novel query language

for the extraction of data from multiple OSNs. The

proposed language is, according to Martin Glinz’s in-

sightful remarks (Glinz, 2002), “as simple as possible

and as rich as needed”, in the sense that it provides

simple keywords and powerful mechanisms to allow

writing queries. Moreover, this language offers the

possibility to be customized by its users in order to

add keywords that better reﬂect the metadata used by

the single OSN.

The paper is organized as follows. Section 2 dis-

cusses the related work in the ﬁeld of data extraction

from OSNs. Section 3 describes a data model that is

Buccafurri, F., Lax, G., Musarella, L. and Nardone, R.

A Novel Query Language for Data Extraction from Social Networks.

DOI: 10.5220/0008362503650371

In Proceedings of the 15th International Conference on Web Information Systems and Technologies (WEBIST 2019), pages 365-371

ISBN: 978-989-758-386-5

365

necessary for the formal representation of the multiple

social network scenario as a direct graph. Section 4

describes the structure of the query language and its

concrete realization in Xtext, also reporting the for-

mal graph. Section 5 reports some usage scenarios in

order to clarify the ﬁnal structure of the queries and

to highlight the potentialities of the developed envi-

ronment. Finally, our conclusions and possible future

work are summarized in Section 6.

2 RELATED WORK

As stated previously, some OSNs offer some APIs

as a part of their business (Lomborg and Bechmann,

2014). In this way they allow third-party companies

to use these APIs to develop complex applications.

To the other hand, some works as (Marres and Wel-

tevrede, 2013; Buccafurri et al., 2015) analyzed and

exploited web scraping for automated capture of data

from OSNs. Web scraping (also known as web har-

vesting) is a well-known technique to extract online

data. This technique is supported by different tools

such as Selenium, cURL, Firebug, Node js (Rotruck

et al., 1973).

As a consequence, the data extraction and the con-

sequent realization of innovative web applications on

the top of social networks is technically feasible. Nev-

ertheless, its complexity is due to different and vari-

able organization of data from the OSNs and the con-

sequent variety of APIs and continuous updates they

offer to programmers and/or the different organization

of web pages. Indeed, despite the conceptual uni-

formity among the structures, mechanisms, features

of existing social-networks, each platform adopts in

practice its own terms, resources, actions. This is a

strong handicap for the design and implementation

of applications enabling internet-working functions

among multiple social networks, and, then, for the

achievement of the above goal. As a matter of fact,

a few exist in terms of models and languages to sup-

port the interaction with multiple social networks.

Up to now a lot of effort has been spent in the

deﬁnition of standards and models for the data ex-

change on web. For example, the Resource Descrip-

tion Framework RDF(rdf, ) is a W3C standard model

for data exchange on web, in particular proposed for

Semantic Web. Based on RDF, many researchers pro-

posed different ontologies, because it guarantees in-

teroperability. Indeed, the authors of (Choi et al.,

2009) proposed SPIDER, a query processing system

for RDF data which supports the SPARQL query lan-

guage (spa, ; Prud et al., 2006). In the literature, many

works use RDF and/or SPARQL approaches for the

manipulation and extraction of general data from web

sources ((Silva et al., 2017; Diamantini et al., 2018;

Thushar and Thilagam, 2008; Barbieri et al., 2010) to

cite a few).

Some of the social network providers offer

query languages and/or extraction techniques as well.

For example, Facebook offers a simple query lan-

guage (fac, 2008) in order to process mainly your

own and your friends’ relational data. This language

does not give any support for complex data extraction.

From 2016, Facebook offers also a speciﬁcation and

a reference implementation of a framework, named

GraphQL, which introduces a new type of web-based

data access interface (Buna, 2016). This framework is

seen as an alternative to the standard REST-based in-

terface (Richardson et al., 2013). The query language

at the basis of GraphQL queries uses the JavaScript

Object Notation (JSON) (Crockford, 2006) and asks

its users to deﬁne a schema in JSON. In particular,

this schema deﬁnes the types of objects and the ﬁelds

that the query has to populate. Consequently, this ap-

proach is very powerful but it is limited to complete

knowledge of the possible types and it is also not ex-

tensible to multiple-social network scenario. The au-

thors of (San Martın et al., 2011) proposed SNQL, a

data model and a query language for Social Network

by representing data as graph database. Although

the idea and the work are very interesting, these lan-

guages cannot be easily extendend to multiple social

network scenarios given the different organization of

the data.

In the context of multiple social networks, it is

more appropriate to analyse the graph query lan-

guages. As an example, Neo4J (Miller, 2013) is

one of the most complete query languages for Graph

Databases, where data can be structured as edges and

vertices. GOQL (Sheng et al., 1999) is a different

graph query language and it is based on an object-

oriented data model. GOQL uses the traditional se-

lect... from... where... statement for querying and

offers also temporal operators such as next, until and

connected. The graph query languages can be bet-

ter adapted to manage complex data structures, where

nodes have properties and are organized by relation-

ships, which can also have properties. However, these

approaches are not tailored for extracting data from

multiple social networks and they are not able to man-

age the different organization of data from the avail-

able sources, which represents, as said before, one of

the the main issues in multiple social network.

With respect to the considered works, our ap-

proach is different because it aims at developing a

complete framework for the creation, veriﬁcation and

execution of queries that are able to extract data from

WEBIST 2019 - 15th International Conference on Web Information Systems and Technologies

366

multiple OSNs. This framework has to offer the fol-

lowing functionalities:

1. Enable users to deﬁne customized keywords

and the mapping towards the speciﬁc metadata

adopted by the OSNs;

2. Support to users during the query editing and val-

idate the results;

3. Enable users to execute queries by chosing among

the usage of APIs, of the web scraping techniques

or of mixed approaches.

With respect to existing languages, the designed

language aims at being general enough to include the

variety of resources managed by the different social

networks. Moreover, it allows users to extend its syn-

tax, adding keywords to the language that map spe-

ciﬁc metadata offered by a single social network. The

design of this language started from the formaliza-

tion of a conceptual data model for the generaliza-

tion of the structure of commonly known social net-

works. Then we deﬁned the syntax and we realized

the corresponding grammar ﬁle into the Xtext frame-

work (Eysholdt and Behrens, 2010).

3 THE BASIC DATA MODEL

Each Online Social Network has been designed for a

particular (main) purpose: for example, Facebook for

keeping in touch with friends, Twitter for microblog-

ging, LinkedIn is business-oriented, Instagram for

sharing pictures. However, all social networks have

some common properties on which they are built:

some examples are (1) user’s proﬁle, which stores

personal data of the user, (2) the concept of rela-

tionship, which is necessary to build the connections

among users, (3) the presence of shared resources,

which can be short texts, Web links, images, videos,

and so on, and (4) actions that can be done from

users, such as appreciating, commenting, sharing of

resources. It is worth noting that social networks usu-

ally implement in a different way (or use a different

name for) similar properties: for example, the friend-

ship is symmetric in Facebook, whereas it is asym-

metric in Twitter, it is built by the concept of cir-

cle in Google+, it is called connection in LinkedIn

whereas in Twitter the terms followers and follow-

ing are adopted. Again, appreciation is implemented

by like in Facebook, endorsement in LinkedIn, +1 in

Google+ (recently closed).

From these examples, it is clear that, in spite of the

intrinsic similarity of social networks, each one de-

ﬁnes its own terminology, thus making social network

depending on the reference to an entity when we want

to write a high-level language for querying social net-

work data: for example, to refer to the contacts of a

social network user, we should ask for connections

if the user is on LinkedIn and followers/following if

the user is on Twitter. Consequently, the need to have

an ontology describing and integrating the character-

istics of social networks arises.

In this section, we solve this problem by present-

ing the ontology used to model social network con-

cepts at an abstract level. The data model presented

in this section is general enough and it is independent

of the speciﬁc social network.

Among the several approaches proposed in the lit-

erature to represent social networks, we used an on-

tology based on the model presented in (Buccafurri

et al., 2016), which has been designed with the goal

of integrating information coming from different so-

cial networks.

Social network data are modeled by a direct graph

G = hN, Ai, in which the set of nodes is N = P∪R ∪B,

with P ∩R∩B = {}, and the set of arcs is A = F ∪ M ∪

Pu ∪ S ∪ T ∪ Re ∪ L ∪ Co with F ∩ M ∩ Pu ∩ S ∩ T ∩

Re∩L∩Co = {}. Now, let’s deﬁne the sets introduced

above.

Each element of the set P represents

the proﬁle of a user and consists in the tu-

ple hurl, socialNetwork, screen-name,

[personalInformation], [picture]i, where

url is the Web address that identiﬁes and localizes

the proﬁle, socialNetwork is the commercial name

of the Online Social Network which the proﬁle

belongs to (the same value is shared by proﬁles in

the same social network), screen-name is the name

chosen by the user who registered the proﬁle to ap-

pear in the home-page of the proﬁle or when posting

a resource, and, ﬁnally, personalInformation and

picture are the information and the personal image

which the user inserted as related to the proﬁle.

The set R models resources of the Web or created

by users. A resource is represented by a tuple hurl,

type, [description], [date]i, where url is the

Web address to access the resource, type indi-

cates the type of the resource content, and ﬁnally,

description and date represent the string inserted

by who published the resource and the publishing

date, respectively.

The set B models bundles, a set of resources han-

dled simultaneously by a user, and represented by a

tuple huri, [description], [date]i, where uri

is the identiﬁer of the bundle, description is the

string chosen by the user to be shown with those re-

sources and, ﬁnally, date represents the publishing

date.

The follow arcs set F ⊆ A = {p

, p

| p

, p

∈ P}

A Novel Query Language for Data Extraction from Social Networks

367

models the fact that in the (source) proﬁle p

, it has

been declared a certain type of relationship towards

the (target) proﬁle p

The me arcs set M ⊆ A = {p

, p

| p

, p

∈ P} de-

notes that the user with proﬁle p

has declared in this

proﬁle to have a second proﬁle p

The publishing arcs set Pu ⊆ A = {p

, b

| p

∈

P, b

∈ B} indicates that the user with proﬁle p

has

published in this proﬁle a bundle b

The shared arcs set S ⊆ A = {b

, b

| b

, b

∈ B}

speciﬁes that the bundle b

(published by a user) is

derived from an already published bundle b

The tagging arcs set T ⊆ A = {p

, br

, w | p

∈

P, br

∈ B ∪ R and w is a word}, denotes that the user

with proﬁle p

assigned the word w to describe a bun-

dle or a resource br.

The referencing arcs set Re ⊆ A = {b

, p

| b

∈

B, p

∈ P} models the fact that a bundle b

includes a

reference to the proﬁle p

The like arcs set L ⊆ A = {p

, pbr

| p

∈ P, pbr

∈

B ∪ R ∪ P} describes the information that a user with

the proﬁle p

expressed a preference/appreciation for

a bundle, a resource or another user proﬁle pbr

The containing arcs set Co ⊆ A = {b

, r

| b

∈

B, r

∈ R} indicates that a bundle b

contains the re-

source r

4 QUERY LANGUAGE FOR DATA

EXTRACTION

Starting from the data model described in the previ-

ous section, we deﬁne the syntax of the query lan-

guage which can be adopted by users to easily extract

data from multiple social networks. We describe its

syntax and its realization through a formal grammar.

The structure of our query language is based on the

well-known select statement of the SQL language,

that retrieves zero or more tuples from one or more

tables in a database. The basic idea behind this lan-

guage is to conceptually substitute entities to tuples

and social networks to tables in the standard mean-

ing of the select statement. It means that, after the

select keyword, the user has to specify the kind of

entity he/she is looking for (i.e., proﬁles, resources or

bundles), then he/she speciﬁes the list of social net-

works to query after the from keyword and, at last,

it is possible to add ﬁltering criteria after the where

keyword. Obviously, the language allows for the use

of logical operators in the where clause in order to

enable more complex predicates.

The ﬁnal structure of our query language is re-

ported in the following listing.

se l ect E1 , E2 ...

from S1 , S2 ...

whe r e P1 l o p P2 ...

In the previous listing, Ei represents the kind of

the requested entity. According to the data model de-

scribed in the previous section, it is a keyword among

profile, resource and bundle. Note that the query

language allows selecting different kinds of entities

in a single query. Si is a conventional name of the

social network; the user speciﬁes the set of social net-

works separating their names by commas. At last, Pi

is a logical proposition which can be used to ﬁlter re-

sults; the logical proposition are combined together

by logical operators lop. Following the data model

of the previous section, in the where clause the user

can specify the value of some elements of the tuple

representative of the requesting entity (speciﬁed af-

ter the select keyword). For example, if the user is

querying for resources, he/she can specify an url, a

type, a description and/or a date. Similarly, if he/she

is asking for proﬁles, it is possible to specify the url,

the screen name, personal information and/or picture

name as search criteria.

Among the different searching criteria that can be

used in the where clause, some of them are strictly re-

lated to the speciﬁc social network. For example, the

kind of a resource (e.g., ﬁlm, book, song) strictly de-

pends on the metadata used by the social network it-

self which stores that information. For this reason, our

language implements a property we named aware-

ness. It represents the capability of the language to be

extended in keywords by users that are “aware” of the

metadata used by every single social network to store

its data. Consequently, we can distinguish between

“known” and “unknown” social network. In the ﬁrst

category, the user inserts the social networks he/she

previously analyzed, while in the second he/she puts

the rest of social network he/she wants to use.

The list of social networks the user wants to use

is stored in a conﬁguration ﬁle. In this ﬁle, the user

speciﬁes the list of keywords identifying social net-

works he/she wants to add to the query language in the

from clause. If a social network is “unknown” (i.e.,

the user does not know the metadata related to the

social network), the user is automatically limited by

the language syntax to use a set of general keywords

given by the data model; for each “known” social net-

work, instead, he/she has to deﬁne a speciﬁc conﬁg-

uration ﬁle in which he/she details the (sub-)kinds of

proﬁles, resources and bundles can be requested. It

means that, if a social network, for example, is able to

manage books, ﬁlms, and songs as kinds of resources,

the user is allowed also to ask for a speciﬁc kind of

them. In this case, the user speciﬁes in the conﬁgu-

WEBIST 2019 - 15th International Conference on Web Information Systems and Technologies

368

ration ﬁle of the social network the keywords he/she

want to use in its concrete language and the mapping

with the metadata of the social network. After this

one-time operation, he/she can open the language ed-

itor and use the keywords he/she speciﬁed.

The awareness property allows also for future

easy integration of additional social networks. The

operations needed are: (1) edit the conﬁguration ﬁle

of the social network and add the new social to the list,

(2) possibly create a conﬁguration ﬁle for the new so-

cial network with the mapping between keywords and

metadata.

The described language has been completely re-

alized by means of the Xtext framework (Eysholdt

and Behrens, 2010). Xtext is a complete framework

for developing programming languages and domain-

speciﬁc languages with textual grammars. Within

Xtext it is possible to deﬁne your own language using

a powerful grammar language. As a result, you get

automatically the full infrastructure including parser,

type checker and the entire editing environment for

Eclipse.

To realize our query language in Xtext, it is nec-

essary to code the syntax of the language. It has to

be deﬁned as a set of rules to which the text has to

be compliant. The main rule, named Query checks

for the overall structure of the query. The following

listing reports this grammar rule.

Que r y :

’ sel ec t ’ se lec t io n += Se l ec t ion

’ from ’

so c ial += So c ial ( ’ ,’ s o cial +=

So c ial ) *

( ’ wher e ’ fi l ter += Fi l ter ) ? ;

This rule speciﬁes the sequence of keywords and

textual portions that have to be satisﬁed by a query.

The syntax of each clause has been designed with a

speciﬁc rule. In the following listing, we report the

remaining part of the grammar.

Se lec t io n :

(( ’ pro f ile ’ pr ofil e += Pro fil e ( ’, ’

pr o fil e += Pro file ) *) |

( ’ re s our ce ’

re s ou r ce += Re sou r ce ( ’ ,’

re s ou r ce += Re sou r ce ) *) |

( ’ bun d le ’ b undl e += Bun d le ( ’ , ’

bu n dle += Bu n dle ) *) ) * ;

So c ial :

name = ID ;

Fi l ter :

( c ond iti o n += Co ndi t io n

(( ’ a n d ’| ’ or ’) co ndi t io n +=

Co ndi t io n ) *) ;

Co ndi t io n :

(( p r ofi le += Pro file )? &

( r eso u rce += Res our c e ) ? &

( b u ndle += B und l e ) ?) ;

Pr o fil e :

( ’ pr ofi leN ame ’ name = ID ) ?

(( ’ pe rs on a lI nf o rm at io n ’

pe rs on a lI nf or m at io n = S TR ING ) ?

( ’ sc r ee n Na m e ’

Sc ree nNa me = STRI NG )? &

( ’ UR L pr o fi l e ’ URL = STRI NG ) ?) ;

Re s ou r ce :

(( ’ ty p e ’ type += Type ) ? &

( ’ de sc ri p ti on Re s ou rc e ’

de scr ip t io n = S T RING )? &

( ’ cr ea t io n _d at e ’

cr ea t io n_d at e += Da te )? &

( ’ UR Ire sou rce ’ URI = S T RING ) ?) ;

Type :

name = ID ;

Bu n dle :

(( ’ de sc r ip ti onB un dl e ’

de scr ip t io n = S T RING )? &

( ’ cr ea t io n _d at e ’

cr ea t io n_d at e += Da te )? &

( ’ UR I bu n dle ’ URI = S T RING ) ?) ;

Date :

d = Day ’- ’m = Mont h ’ -’y = Year ;

Day : INT ;

Mon t h : INT ;

Year : I N T ;

The Selection rule mainly checks the usage of the

three keywords (proﬁle, resource, and entity), and

properly calls three speciﬁc rules. The Social rule

is very simple as, at grammar level, it could check

only for the presence of a sequence of social network

identiﬁers. As previously said, the keywords identi-

fying social networks are deﬁned by the user itself in

a proper conﬁguration ﬁle. At last, the Filter rule,

together with an additional rule named Condition,

checks the format of the search ﬁlter and allows to

create complex combinations of conditions by using

logical operators. The ﬁltering mechanisms the user

A Novel Query Language for Data Extraction from Social Networks

369

Figure 1: Examples of usage of our query language: (a) content assist enabled; (b) wrong query (c) query for Facebook; (d)

query for Instagram.

can specify for entities are deﬁned in the three rules

Proﬁle, Resource, and Bundle. These rules check for

the presence of keywords related to the attributes of

these entities which are given by the data model of

the previous section.

The implementation of the awareness property of

the language, that is the veriﬁcation of additional key-

words in conﬁguration ﬁles, is realized by specify-

ing customized validation rules. Speciﬁcally, we de-

ﬁned the behavior of the Validator class that is used

in the Xtext framework to customize the veriﬁcation

activity. Mainly, in this class we inserted the code

necessary to check for the presence of the social net-

work identiﬁer in the conﬁguration ﬁle and the veriﬁ-

cation of the existence of the conﬁguration ﬁle related

to the social network addressed in the query. In this

case, the validation allows using additional keywords

as given by the conﬁguration ﬁle and previously de-

scribed. The same logic has been implemented also

in the ProposalProvider class to enable the content

assist and suggest users the accepted keywords with

respect to the structure of the query.

5 USING THE QUERY

LANGUAGE

This section shows some examples of use in order to

better clarify the objective of the deﬁned query lan-

guage and to demonstrate its effectiveness.

Fig. 1 depicts 4 different usage scenario. Fig. 1(a)

shows the suggestions of the content assist in the from

clause. In this scenario, the editing environment re-

trieved the list of social network from a conﬁguration

ﬁle and suggest them to a user. The user can select a

social network from the shown list, avoiding syntacti-

cal errors.

Fig. 1(b) shows a validation error. In this scenario

the user speciﬁes that he/she is looking for resources

from the Facebook social network and, in the where

clause he/she ﬁlters result on the type Tweet. In this

case, the environment check that this type (expressed

as a keyword) is not speciﬁed in the conﬁguration ﬁle

related to the Facebook and raises an error.

Fig. 1(c) and Fig. 1(d) show two correct usage

examples. In the former example the user asks for

nametags in Instagram, while the latter shows a query

requesting events from Facebook. These last exam-

ples show the concrete support the environment of-

fers to users by enabling the writing of query which

uses speciﬁc keywords related to the social network

metadata.

6 CONCLUSIONS AND FUTURE

WORK

This paper proposes a preliminary version of a novel

query language for the data extraction from Online

Social Networks. The main feature of this language

is the possibility to retrieve data from multiple On-

line Social Networks through a single query. With

respect to existing languages, the proposal also of-

fers the possibility for users to add keywords in the

language which reﬂect the metadata used by social

networks to address its data. This feature, which

we called awareness, enables the possibility to add

knowledge into the language. The proposed language

has been realised by means of the Xtext framework,

a well-known framework for the realisation of textual

Domain Speciﬁc Languages.

WEBIST 2019 - 15th International Conference on Web Information Systems and Technologies

370

This work represents a ﬁrst step in a wide project

which will conduct to the complete development of

a framework for data extraction from multiple social

networks. In fact, future works include the develop-

ment of the needed transformations that could auto-

matically translate the queries into executable code,

whose execution performs the needed operations on

the OSNs to obtain data. We plan to integrate different

existing techniques for the data extraction from web

sources (such as web scraping and APIs call) in order

to design an efﬁcient and effective tool. Of course,

the most critical issue is to cope with the continuous

changes in web presentation layers and in APIs, of-

fered by the OSNs. At last, we plan to build inno-

vative applications which could use the social data to

enforce the trust associated with web proﬁles.

ACKNOWLEDGEMENTS

This work has been partially supported by the

project “Secure Citizen Remote Identiﬁcation (SE-

CRI)” (CUP J88C17000150006), funded by Regione

Calabria - POR Calabria FESR-FSE 2014-2020 -

Asse I

REFERENCES

Resource Description Framework . https://www.w3.org/rdf.

SPARQL Query Language for RDF .

https://www.w3.org/tr/rdf-sparql-query/.

(2008). Facebook Query Language.

http://developers.facebook.com.

Barbieri, D. F., Braga, D., Ceri, S., VALLE, E. D., and

Grossniklaus, M. (2010). C-sparql: a continuous

query language for rdf data streams. International

Journal of Semantic Computing, 4(01):3–25.

Bell, G. (2009). Building social Web applications: Estab-

lishing community at the heart of your site. ” O’Reilly

Media, Inc.”.

Buccafurri, F., Lax, G., Nicolazzo, S., and Nocera, A.

(2016). A model to support design and development

of multiple-social-network applications. Information

Sciences, 331:99–119.

Buccafurri, F., Lax, G., Nocera, A., and Ursino, D. (2015).

A system for extracting structural information from

social network accounts. Software: Practice and Ex-

perience, 45(9):1251–1275.

Buna, S. (2016). Learning GraphQL and Relay. Packt Pub-

lishing Ltd.

Choi, H., Son, J., Cho, Y., Sung, M. K., and Chung, Y. D.

(2009). Spider: a system for scalable, parallel/dis-

tributed evaluation of large-scale rdf data. In Proceed-

ings of the 18th ACM conference on Information and

knowledge management, pages 2087–2088. ACM.

Crockford, D. (2006). The application/json media type for

javascript object notation (json). Technical report.

Diamantini, C., Lo Giudice, P., Musarella, L., Potena, D.,

Storti, E., and Ursino, D. (2018). An approach to

extracting thematic views from highly heterogeneous

sources of a data lake. In Proceedings of the 26th

Italian Symposium on Advanced Database Systems

(SEBD).

Eysholdt, M. and Behrens, H. (2010). Xtext: implement

your language faster than the quick and dirty way.

In Proceedings of the ACM international conference

companion on Object oriented programming systems

languages and applications companion, pages 307–

309. ACM.

Glinz, M. (2002). Statecharts for requirements

speciﬁcation-as simple as possible, as rich as

needed. In Proceedings of the ICSE2002 workshop

on scenarios and state machines: models, algorithms,

and tools.

Lomborg, S. and Bechmann, A. (2014). Using apis for data

collection on social media. The Information Society,

30(4):256–265.

Marres, N. and Weltevrede, E. (2013). Scraping the social?

issues in live social research. Journal of cultural econ-

omy, 6(3):313–335.

Miller, J. J. (2013). Graph database applications and con-

cepts with neo4j. In Proceedings of the Southern Asso-

ciation for Information Systems Conference, Atlanta,

GA, USA, volume 2324.

Prud, E., Seaborne, A., et al. (2006). Sparql query language

for rdf.

Richardson, L., Amundsen, M., and Ruby, S. (2013). REST-

ful Web APIs: Services for a Changing World. ”

O’Reilly Media, Inc.”.

Rotruck, J. T., Pope, A. L., Ganther, H. E., Swanson, A.,

Hafeman, D. G., and Hoekstra, W. (1973). Selenium:

biochemical role as a component of glutathione per-

oxidase. Science, 179(4073):588–590.

San Martın, M., Gutierrez, C., and Wood, P. T. (2011). Snql:

A social networks query and transformation language.

cities, 5:r5.

Sheng, L., Ozsoyoglu, Z. M., and Ozsoyoglu, G. (1999).

A graph query language and its query processing. In

Proceedings 15th International Conference on Data

Engineering (Cat. No. 99CB36337), pages 572–581.

IEEE.

Silva, R. R. C., Leal, B. C., Brito, F. T., Vidal, V. M., and

Machado, J. C. (2017). A differentially private ap-

proach for querying rdf data of social networks. In

Proceedings of the 21st International Database En-

gineering & Applications Symposium, pages 74–81.

ACM.

Thushar, A. and Thilagam, P. S. (2008). An rdf approach for

discovering the relevant semantic associations in a so-

cial network. In Proceedings of the 16th International

Conference on Advanced Computing and Communi-

cations (ADCOM), pages 214–220. IEEE.

A Novel Query Language for Data Extraction from Social Networks

371