An Ontology for Social Networking Sites Interoperability

Konstantinos Tserpes, George Papadakis, Magdalini Kardara, Athanasios Papaoikonomou,

Fotis Aisopos, Emmanuel Sardis and Theodora Varvarigou

Department of Electrical and Computer Engineering, National Technical University of Athens,

9, Heroon Polytechniou Str, 15773 Athens, Greece

Keywords: Social Web, Semantic Web, Ontology, API, Social Networking Sites.

Abstract: The Social Networking Sites (SNS) comprise a pool from which developers can pump functionality and

data. Usable REST APIs are providing access to two valuable business assets: the users' generated content

and social graph. The lack of standards and the antagonistic nature of the SNSs have resulted in the use of

proprietary API specifications and -in turn- data models. Each SNS uses a different method to access and a

different way to describe notions which are largely similar, e.g. “friends” or shared “multimedia items”. The

conceptual similarity between entities “living” in the SNS, creates a remarkable opportunity: The

aggregation of the social functionality and data can provide the basis for a unique platform on top of which

third parties can deploy new added value services, seamlessly using the underlying SNSs APIs. This paper

presents an attempt to implement this concept by means of building a core ontology for supporting SNS

interoperability.

1 INTRODUCTION

The advent of mashups and web APIs enabled the

routinely use of third parties' platform tools so as to

build new applications by leveraging on their

content and core functionality. The advantage is

immense: without any significant cost for

development (just for maintaining the API) on the

platform owner's side and by taking advantage of the

creativity of external developers, the platform's

functionality is extended by added value

applications.

Coping with this concept, the Social Networking

Sites comprise a typical pool of functionality and

data. Their REST APIs are providing access to two

valuable assets: the user generated content and social

graph. The lack of standards and the antagonistic

nature of the SNS have resulted in the use of

proprietary API specifications and -in turn- data

models. Each SNS uses different ways to access and

describe notions which are largely similar, e.g.

“friends” or shared "media items”. The conceptual

similarity between entities “living” in the SNS,

creates a remarkable opportunity: The aggregation

of the social functionality and data can provide the

basis for a unique platform on top of which third

parties can deploy new added value services,

seamlessly using the underlying SNSs APIs.

Providing developers with a cross-platform tool

that enables them to manage the dynamically

generated content and complex social interactions by

allowing them to build, deploy and potentially sell

services that combine data and functionality from

two or more different SNS services disregarding the

underlying API implementation, will create an agile

and profitable market of services and will bring the

Internet of Services concept a step closer to

realization.

The proposed solution aims directly at the heart

of this challenge: Develop an SNS-like API that will

manage to aggregate data and functionality from

underlying SNSs that deliver their functionality

through web APIs. We call this API "SocIoS API"

and it is the core component of the SocIoS integrated

solution (http://www.sociosproject.eu).

The SocIoS API is essentially an object model in

which on one hand unifies the domain of SNS APIs

in order to expose certain concepts that appear

repeatedly in the SNS API worlds, while on the

other hand, leverages on these concepts and allows

services to use them in added value application

scenarios. This twofold task presents some

interesting characteristics that need to be considered

in order to define this specification.

245

Tserpes K., Papadakis G., Kardara M., Papaoikonomou A., Aisopos F., Sardis E. and Varvarigou T..

An Ontology for Social Networking Sites Interoperability.

DOI: 10.5220/0004142202450250

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2012), pages 245-250

ISBN: 978-989-8565-30-3

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

Our method fosters ideas from both approaches:

it defines the core objects of the SocIoS API domain

through an ontology which helps to modeling these

objects to the underlying SNS APIs using software

wrappers (formally called SocIoS adaptors). To

build this ontology we start with the conceptual

analysis of the application scenario in order to

capture the main domain concepts. These include the

notions of User, Activity, Group, Multimedia item,

Event, Location and Message. Next, we study a

number of SNSs in order to identify the API objects

that are semantically equivalent to the domain

concepts. This helps us to place the domain concepts

in the SNS context and examine their relationships

and position in the object model hierarchy. The

combination of these approaches leads to a

conceptualization of the SocIoS domain that is

formally captured in the presented SocIoS ontology.

The objective of this ontology is to provide the

necessary flexibility of the SocIoS object model so

as to be extended as necessary when new SNS APIs

or application domains are to be included later in its

lifetime. The model is captured using the ontology

specification language (OWL) (McGuinness, 2004)

that only includes the information that is necessary

for an object model. We directly derive the initial

SocIoS object model from this ontology. When in

the future the SocIoS object model needs to be

expanded, the ontology will have to be expanded

first, adding the logical relationships between classes

that provide a real ontological definition. This

ensures that the new definitions are logically correct,

i.e. satisfy the conceptualization of the SocIoS

ontology. Then, a new SocIoS object model is

derived by simplifying from the new ontology.

The document is structured as such: Section 2

provides the related work in specifications and APIs

that have similar goals as ours as well as ontologies

from which our approach can adopt concepts so as to

be extendible. Section 3 presents the domain

concepts of SocIoS as the conclusion of an analysis

of three real-world application scenarios. In Section

4 the derived domain concepts are projected to

existing objects of a group of an appropriate

selection of SNS APIs. Finally, Section 5 gives the

SocIoS ontology as a result of this analysis and

discusses the lower level parameters of the concepts.

2 RELATED WORK

There have been a number of approaches to build a

meta-API so as to implement functionality

federation and data marshaling from numerous

SNSs. However, as stated above, the majority

comprise proprietary solutions which limits the

extensibility of the platforms to non-supported APIs.

Such examples are GNIP®, hootsuite® and

Stroodle®.

Even though we study a number of SNS APIs so

as to identify the dominant objects that they

commonly use, our solution uses mainly OpenSocial

(OpenSocial, 2010) as a reference specification.

OpenSocial defines a common API for social

applications across multiple websites. As it is a

product of collaboration of a broad set of members

of the web community it naturally encompasses the

most dominant SNS concepts. However, OpenSocial

is not meant to be used as a meta-API or an

aggregator. This results in two reasons for which

OpenSocial cannot meet our requirements: a) it is

not meant to be adaptive, but rather requires from

the underlying APIs to adapt it, and; b) it is

overdetailed, providing an extreme number of

second-class objects that cannot be mapped to

existing API concepts. The issue of adaptability, is

the one that mainly led us into seeking an ontology

solution.

By defining the SocIoS universe of discourse,

we assume that we will manage to capture the

common denominator of the core SNS applications.

This universe can be used at a higher layer by any

service. However, here, we need to make a clear

separation. With this work we do not want to model

the real world concepts that appear in the SNS as a

result of their use. E.g. we are not interested in

concepts that derive from tagging multimedia

content since this involves practically everything.

Various efforts have been put on that aspect with the

most prominent work being the one of (Mika, 2007).

It is obvious that in the frame of an SNS a user can

essentially import and define anything using any

type of description, essentially capturing the whole

world (or a new world for what is worth). Various

crowdsourcing techniques have been proposed for

that purpose and most of the SNS encompass

functionality for meeting this particular objective

(e.g. tagging mechanisms).

Instead we are focusing on conceptual entities

that are manifested in the object models of the

existing SNSs. By using agreed-upon Semantic Web

formats to describe people, content objects, and the

connections that bind them together, SNSs can

interoperate by appealing to common semantics

(Breslin, 2007).

To this end we analyze the existing work in the

area of formal specifications of the SNS domain so

as to investigate if and how there are ontologies that

already capture notions that our approach would like

KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

246

to define. As such, perhaps the most relevant

ontology is the Friend of a Friend (FOAF) (http://

www.foaf-project.org/) project the puporse of which

is to allow people define their profile, content and

relationship to other people and contents so as to

interconnect all web resources in a distributed

manner.

Another relevant approach is the Semantically-

Interlinked Online Communities (SIOC) Core

Ontology (Bojārs, 2010) which provides the main

concepts and properties required to describe

information from online communities (e.g., message

boards, wikis, weblogs, etc.) on the Semantic Web

(Berrueta, 2010).

Finally, the Linking Open Descriptions of Events

(LODE) is an ontology for publishing descriptions

of historical events as Linked Data, and for mapping

between other event-related vocabularies and

ontologies (Shaw, 2010). Thus, the compatibility to

SocIoS revolves around the notion of Event and the

interplay between real world and SNS world events.

3 APPLICATION SCENARIOS

DOMAIN CONCEPTS

As a starting point to our analysis, we employ three

application scenarios that stem from real end-user

requirements. One scenario is related to journalism

and the use of SNS users as a source of information

to a multimedia dossier article and two are related to

TV commercial production and particularly to the

use of social media for location scouting and casting

extras. All three applications relate to the

exploitation of abstract social networks created on

the fly based on demographic and social criteria (e.g.

credibility, influences, etc), for finding and

retrieving information in a reliable, timely and cost-

efficient manner. We analyze these applications so

as to define the SocIoS domain of discourse so as to

have a reference conceptual classification of our

universe.

The conceptual analysis of the abovementioned

scenarios leads to the following conclusions:

The SNS user is central to the scenario. Their

activity is important to the end-users who want to

monitor it. SNS users can be classified in groups so

as to allow the end-users better monitor their

activity. However, not all users are equally reliable.

Their reliability is assessed through a rating that is

assigned by other SNS users. Furthermore, the basic

functionality allows journalists to look for

multimedia items which are owned by SNS users.

Multimedia items can steer end-users' interest if they

directly fit to their criteria or if they emerge as

central objects in occasions of intense and unusual

activity, i.e. an event. Typically, an event is taking

place at a location and this affects greatly the

definition of events of interest of the end-users.

Finally, communication between SNS users and end-

users is achieved through the basic communication

channel, which is the message.

Based on this short summary, SocIoS concluded

to the fact that the domain concepts are the

following: User, Activity, Group, Multimedia item,

Event, Location, Reliability and Message.

In what follows, we explain how we formally

capture and extend these notions based on the actual

vocabulary used by the object models of the

underlying SNSs.

4 SNS API ANALYSIS

Conceptually, SocIoS, sees the SNS universe as a

single social network, in which people are socially

connected to each other and the basic social

interaction is the sharing of content (text, visual and

audio). In order to achieve that, it needs to consider

the section of the groups of all the common basic

entities of the underlying SNS. This task by itself is

rather complex given the highly diverse nature of

SNS applications. The entities "living" in them are

generally different with the exception of few basic

ones (like the notion of people).

This Section explains in a practical way what the

main objects in the SNS API domain are which fit to

the descriptions of the domain concepts as they

generated from the SocIoS scenarios. This analysis

is pivotal in the definition of the SocIoS API and the

intermediate ontology.

The objective is to identify the common entities

that appear in the SNS APIs which follow the

semantics of the domain concepts presented above,

define them and understand their relationships. For

that, we used a list of specific SNS APIs. The

selection of these SNS APIs is made according to

criteria such as: SNS Popularity, API Popularity,

Scope, Openness and Maturity.

Following these criteria, we conducted an

analysis of the following APIs: OpenSocial,

Facebook, Twitter, FlickR and Youtube. In these

SNS APIs, there are objects that express notions that

are equivalent to the SocIoS domain concepts,

however, they differ in the naming as well as in the

hierarchy in the formal description (others are first-

class objects, while others are mere object

properties). Therefore the relationship is not always

AnOntologyforSocialNetworkingSitesInteroperability

247

one-to-one, nor the objects are the exact semantic

match.

The analysis resulted to a number of objects in

the selected SNSs that are semantically equivalent to

the abovementioned concepts. These are depicted in

Table 1 (in Appendix) and are: Person, Group,

MediaItem, Message, Activity, Event, Location and

Rating.

A number of conclusions can be extracted from

this analysis. Firstly, not all the concepts are

supported by all the SNS APIs (marked as "N/S").

Secondly, some objects are implicitly related to the

concepts, e.g. a Tweet object is semantically

equivalent to a Message. Furthermore, not all

conceptually similar objects are of the same class

(Location is second-class in OpenSocial, Facebook

and Youtube and first-class in all the rest). Finally,

the notion of Events in the way it is defined in this

document is only existent in Twitter.

Having defined and examined the domain

concepts under an SNS context we can move to the

designing of the SocIoS Core Ontology.

5 SOCIOS CORE ONTOLOGY

From the abovementioned analysis, it is clear that

the method to conceptually model the domain, is a

combination of two efforts: map the underlying SNS

object models into a single point of reference and

model the SocIoS scenarios universe. As seen in

Section 4, other domain concepts can be promoted to

classes, whereas others remain as properties.

Furthermore, the analysis of the objects and their

members also assisted in the identification of the

most prevalent data properties.

Figure 1: Overview of simplified SocIoS ontology.

Given the above we drafted an ontology that

captures the basic context and rules behind the

SocIoS domain that we attempted to record. The

ontology starts by putting Person and MediaItem as

the central entities. Then, it analyzes the relationship

that bind these two entities with the rest of the core

entities. For practical reasons, a new entity is

introduced so as to record the SNS in which a

Person is active, publishing and sharing MediaItems.

The overview of this ontology is presented in Figure

In what follows we present the classes depicted

above along with their properties.

5.1 Person

Properties

 connects_to: Other person entities with whom

the person in question shares some kind of

relationship (is a friend, follows, etc).

 owns: Photos that are published and owned by

this person.

 has_rating: A person has a rating assigned by

other people

 belongs_to: The SNS in which the Person

exists

 birthday: The birthday of this person.

 displayName: The name of this Person,

suitable for display to end-users.

 ethnicity: Person's ethnicity.

 gender: The gender of this person.

 location: Description of a person's location.

 name: The complete person's real name.

 tags: A user-defined category label for this

person.

5.2 MediaItems

Properties

 has_rating: Average rating of the media item

on a scale of 0-10.

 created: Creation datetime associated with the

media item.

 description: Description of the media item.

 duration: For audio/video clips - playtime

length.

 last_updated: Update datetime associated with

the media item.

 location: Location corresponding to the media

item.

 num_comments: Number of comments on the

media item.

 num_views: Number of views for the media

item.

 num_votes: Number of votes received for

voting.

 start_time: For streaming/live content,

datetime when the content is available.

KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

248

 tagged_people: Array of string (IDs) of people

tagged in the media item.

 tags: Tags associated with this media item.

 title: Describing the media item.

 type: The type of media.

5.3 Activity

Properties

 relates_to: MediaItem that may accompany the

activity

 has_generator: An activity may be generated or

related to an event

 refers_to: The person who triggered the activity

 body: Specifying an optional expanded version

of an activity.

 publishes: Any photos, videos, or images that

should be associated with the activity.

 postedTime: Specifying the time at which this

activity took place.

 title: Specifying the primary text of an activity.

5.4 Events

Properties

 refers_to: a (physical, social, or mental) object

involved in an event. In SocIoS this is related

to MediaItems or Activity.

 atPlace: a named or relatively specified place

that is where an event happened.

 atTime: an abstract instant or interval of time

that is when an event happened.

 circa: an interval of time that can be precisely

described using calendar dates and clock times.

 illustrates: an event illustrated by some thing

(typically a MediaItem object)

 location: an abstract region of space (e.g. a

geospatial point or region) that is where an

event happened.

5.5 Messages

Properties

 has_recipients: Person entities who receive the

message.

 has_sender: The person who sent the message.

 body: The main text of the message.

 collections: Identifies the messages collection

this message is contained in.

 inReplyTo: Use for threaded

comments/messages.

 status: Status of the message. (NEW, READ,

DELETED)

 timeSent: The time that the message was sent.

 subject: The subject of the message.

5.6 Rating

Properties

 score: The score that depicts the rating

 range: The score range (min, max).

5.7 Group

Properties

 has_member: Person entities who are members

of the group.

 title: Title of the group.

 description: Description of group.

5.8 SNS

The SNS concept captures the notion of the

container in which a typical social network is

manifested for a person and the mediaitems are

shared. Typical examples are Facebook, Twitter,

MySpace, YouTube, FlickR.

Properties

 container: The platform SNS, e.g. Facebook,

Twitter, etc.

Relevant Definitions

No equivalent notions in other ontologies.

6 CONCLUSIONS

This document summarizes the procedure and the

results of the conceptualization of the SocIoS

domain as an intermediate step to define the SocIoS

Object Model (API). The analysis starts from the

domain concepts defined by the SocIoS scenarios

and continues with the identification of conceptually

common data specifications from various APIs that

semantically fit to the domain concepts. The result

of this work is captured in the SocIoS core ontology

that is expected to assist the developers to better

comprehend the notions behind the SocIoS object

model and extend it as appropriate.

The ontology permits the extension of the

SocIoS API towards any direction, either for

including new applications or for supporting further

SNS APIs. This flexibility is also crucial in cases of

strong dependence on third parties because it allows

for quick cope with possible changes. Furthermore,

we present how the ontology is linked to semantic

AnOntologyforSocialNetworkingSitesInteroperability

249

web specifications, a fact that adds to the

extendibility of our proposed solution.

ACKNOWLEDGEMENTS

This work has been supported by the SocIoS project

and has been partly funded by the European

Commission's 7th Framework Programme through

theme ICT-2009.1.2: Internet of Services, Software

and Virtualisation under contract no.257774

REFERENCES

OpenSocial Social Data Specification 1.1, available at:

http://opensocial-resources.googlecode.com/svn/spec/

1.1/Social-Data.xml

McGuinness D. L. and van Harmelen F. eds. "OWL Web

Ontology Language Overview," World Wide Web

Consortium (W3C) recommendation, Feb. 2004;

www.w3.org/TR/owl-features.

Mika, P. Ontologies are us: A unified model of social

networks and semantics, Journal of Web Semantics,

Elsevier, Volume 5, Number 1, p.5-15 (2007)

Breslin, J., Decker, S., "The Future of Social Networks on

the Internet: The Need for Semantics," Internet

Computing, IEEE, vol.11, no.6, pp.86-90, Nov.-Dec.

2007

Bojārs U., Breslin J. G., SIOC Core Ontology

Specification, http://rdfs.org/sioc/spec/

Berrueta D., Brickley D., Decker S., Fernández S., Görn

C., Harth A., Heath T., Idehen K., Kjernsmo K., Miles

A., Passant A., Polleres A., Polo L., SIOC Core

Ontology Specification, Uldis Bojārs, John G. Breslin

(eds.), March 25, 2010, available at: http://sioc-

project.org/ontology#term_Item

Shaw R., Troncy R. and Hardman L., LODE: An ontology

for Linking Open Descriptions of Events, July 7, 2010,

available at: http://linkedevents.org/ontology/

Sabater J. and Sierra C., Regret: A reputation model for

gregarious societies. In Proceedings of the Fourth

Workshop on Deception, Fraud and Trust in Agent

Societies,Montreal, Canada, pages 61—69, 2001.

Schillo M., Funk P., and Rovatsos M. Using trust for

detecting deceitful agents in artificial societites. In

Applied Artificial Intelligence, Special Issue on Trust,

Deception and Fraud in Agent Societies, 2000.

Sabater J. and Sierra C., Reputation and social network

analysis in multi-agent systems. In Proceedings of the

first international joint conference on Autonomous

agents and multiagent systems: part 1 (AAMAS '02).

ACM, New York, NY, USA, 475-482, 2002.

APPENDIX

Table 1: Domain concepts and semantically equivalent objects in selected SNS API (N/S stands for "not supported").

Concept

SNS API objects

OpenSocial

Facebook

Twitter

FlickR

Youtube

User

Person

User

Users

people

users

Group

Group and

Friendslist

Lists

groups:members

N/S

MediaItem

MediaItems

Photo and Video

Tweets

photos

media

Message

Tweets and

direct_messages

act

ty:user

ments

gd:comments

Activity

N/S

Timeline

activity

Event

N/S

Trends

N/S

Reputation

N/S

Review:Rating

Users:suggestions

stats

ratings

Location

Person/

MediaItem:Location

Album:Location

geo

places

yt:location

KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

250