An Ontology-based Approach to Social Networks Mining

Viacheslav Lanin

, Lyudmila Lyadova

, Elena Zamyatina

and Nikita Vostroknutov

Department of Information Technologies in Business, HSE University,

Perm, Russian Federation

Keywords: Social Network, Social Media, Multifaceted Ontology, Event Classification, Event Logs, Process Mining.

Abstract: The article presents an approach to the analysis of processes in social networks based on using multifaceted

ontologies. An overview of existing tools for analyzing social networks is provided and the results of studying

social networks are presented. A multifaceted ontology describing social networks has been developed based

on the research findings. The main result for this study is the ontology of events, which can be used to pre-

process data, extracted from social networks, to generate event logs in a form suitable for export to Process

Mining tools to analyze networks (identification of user behavior patterns, analysis of information distribution,

etc.). Examples illustrating the proposed approach are given.

1 INTRODUCTION

From the user's point of view any social network is an

interactive multi-user website, the content of which is

filled by the network participants themselves. It is an

automated social environment that allows a group of

users to communicate about their common interests.

Communication is carried out via different tools (for

example a web service of internal mail or instant

messaging, and so on).

On the one hand, social networks help solve many

tasks, but on the other hand, they might become a

source of problems. This fact has motivated a large

number of researchers to study social networks. The

greatest interest is caused by the problems of

dissemination of information on the network, issues

of community formation, etc. The methods of static

analysis of networks using graph models, statistical

methods and machine learning methods are better

developed.

When analyzing social networks, economists

receive information about transactions, the influence

of others on human behavior, while political scientists

investigate the formation of political preferences.

Static and dynamic methods may be used for these

purposes.

This article focuses on the methods that allow

https://orcid.org/0000-0002-0650-2314

https://orcid.org/0000-0001-5643-747X

https://orcid.org/0000-0001-8123-5984

researchers to study the dynamic behavior of users in

networks. The most promising methods are based on

the analysis of event logs. Logs can be formed based

on data obtained from real social networks. But

studies can also be carried out using simulation

modeling tools. The ontological approach expands

the possibilities of studying social networks with

existing tools. It is proposed to use multifaceted

ontology (Abrosimova, 2018; Shalyaeva, 2016;

Shalyaeva, 2017) to retrieve information on processes

in social networks, to prepare event logs used by

Process Mining tools to solve many tasks of

analyzing various systems.

2 RELATED WORKS

Consider approaches to solving the most interesting

problems of social network analysis according to goal

of the research presented in this paper.

Previous research works provide many examples

of using social networks to distribute content among

users. This is how marketers try to spread information

about products in order to make profit (Kang, 2015;

Yang, 2018; Bindi,2017; Zhao, 2018). On the other

hand, attackers try to spread malicious or fake

information (Dang-Pham, 2020, Dmitriev, 2020,

234

Lanin, V., Lyadova, L., Zamyatina, E. and Vostroknutov, N.

An Ontology-based Approach to Social Networks Mining.

DOI: 10.5220/0010716600003064

In Proceedings of the 13th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2021) - Volume 2: KEOD, pages 234-239

ISBN: 978-989-758-533-3; ISSN: 2184-3228

Ilieva, 2018; Tumbinskay, 2017). Thus, it is

necessary to develop algorithms or strategies that

either contribute to the fastest possible dissemination

of information, or, conversely, hinder dissemination.

In the article (Van der Aalst, 2005), event logs are

used to identify social connections between

employees of a company, to build a sociogram that

reflects the structure of social connections for users

of the CRM system of the company. These tools are

useful to study connections between users of mass

media too. In order to use the Process Mining tools, it

is needed to get an event log.

An overview of the technologies for extracting

information on events is given in the paper (Zhan,

2019).Event logs can be built on the base of data

stored in databases (Calvanese, 2016).

Event logs can be constructed with using data of

network messages (Carrasquel, 2021) and event data

of different applications (Mukala, 2015). Methods of

extracting data to generate event logs from

unstructured or semistructured sources are also

discussed. News events can be extracted from social

media (Peña-Araya, 2015).

The article (Zavarella, 2014) describes a system

that automatically identifies certain types of global

events, such as natural disasters, epidemics, and

military conflicts through the analysis of news sites

and social networks. The article describes several

experimental approaches to semantic integration of

user content published in social networks with

existing information systems.

The development of a system used to

automatically receive and categorize events based on

user posts on Twitter is described in the article (Ritter,

2012). The author uses machine learning algorithms

to work with publications and identifies various event

groups from comments, including entities, events,

dates, and categories.

Semantic technologies extend data pre-processing

capabilities at preparing event logs for the analysis

with process mining tools. These technologies have

become the kernel for process analysis software

developing (De Medeiros, 2008) as they allow solve

the tasks of information retrieval, data extraction and

analysis, in particular, when searching for and

analyzing facts (Vokhmintsev, 2013).

Simulation methods, their capabilities and

advantages are discussed in the articles (Mikov, 2013;

Zamyatina, 2020; Dmitriev, 2020). The results of

application of various methods and tools of

simulation modeling to solve problems of social

networks analysis are presented.

3 SOCIAL NETWORKS

ONTOLOGY DEVELOPMENT

In this research, developed ontology should be based

on the combination of two aspects: functional and

structural (Kietzmann, 2011) views to social media.

According to the purpose of this research, the key

concept must be an Event.

It is necessary to define possible events happening

in social media via their functional and structural

elements. So, event-based approach to the definition

of social media can be discussed.

According to the selected approach possible types

of events in social media should be defined first and

generalized social media types could be explored then

via event logs.

3.1 Social Networks Studying

The main properties of social networks to be

identified to prepare information for modeling.

All social networks have a property called user-

oriented design. In the article (Dawot, 2014) the basic

principles of user-oriented design are described:

interaction between users; recognition (individuality)

of community members. These properties allow

solve the tasks of generating event logs, where it is

necessary to identify not only events, but also objects

and processes and their cases. In the article

(Kietzmann, 2011), the authors highlight seven main

structural elements, characterizing social networks

based on the principles of user-oriented design:

 Identity (possibility of self-expression of the

social network user).

 Conversations (ways of communication between

social network users, both personal and group).

 Sharing (methods of obtaining, distributing and

modifying public content.

 Presence (the ability of users to have the

location of other users).

 Relationships (different connections between

social network users).

 Reputation (the ability of users to confirm their

popularity or competence).

 Communities (user interaction with

communities (groups) and sub-communities).

The identification of functional components for

the listed above structural components leads to

information extraction need for analyzing events and

their description. The event model of a social network

is formed on the basis of the received information on

the functional components relevant to the structural

components of social networks.

An Ontology-based Approach to Social Networks Mining

235

3.2 Designing an Event Model of a

Social Network

The main event elements of the social network are

listed in Table 1.

Table 1: Main event elements of the social network.

Functional element Event element

Personal profile Adding updating or deleting personal

information

Content profile

(combining various

publications from

users or groups into

one structure,

Facebook feed for

example)

Changing of user content profile

Activity log Changing a history of user activity

Social content Generation of discussions, comments

and personal messages

Discussion topic Creating or changing a topic of

discussion

Notification Notification about post or social

connection

Content creation Publishing text, audio, video post. Rate

content (like), sending sticker or emoji.

Editing text, audio, video message

Content sharing of

(personal or third-party)

Rate content (like), repost.

Community Creating a community

Deleting a community

Social connection Applying for social connection

Changing of social connection

Location status Changing of location status

Content popularity Changing of content popularity metric

Subscription Adding or removing subscription

This general events description is the base for

event ontology development.

3.3 Development of the Social Network

Event Ontology

The first part of ontology deals with events. The

bottom-up method was used for developing an

ontology part describing social media events.

Respectively the first stage was a class definition for

elementary events. The total number of classes of

elementary events is more than thirty. After that, at

the second stage, elementary events were generalized

to appropriate super classes. The top of the event

hierarchy is the generalized class named

Social_Media_Event.

The ontology of a social network includes the

general class Social_Network, which comprises

certain subclasses highlighted on the basis of the

analysis of structural elements listed above (Section

3.1) of social networks (Figure 1).

Figure 1: Functional Element Class Hierarchy.

The identified events are presented in ontology as

subclasses of functional class (Figure 2).

Figure 2: Example of Event Mapping Structure.

Typically, in social media, several elementary

events are closely related to each other. For example,

the event of sending a message by one user is

associated with the event of creating a notification for

other users to whom the message is addressed.

The Complex_Event class has been added to the

ontology to model such related events that are

reflective of each other (a “reflected” event is a

consequence of the first event). It allows to perform a

joint analysis of several related events more

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

236

effectively than considering the events independently

of each other. Instances of classes can be used to

model such complex events. The instances can be

adapted for representing common case behavior in a

particular social network. For example, when a user

clicks the Subscribe button on Youtube, two different

events are triggered within the system. The first one

is adding the channel to the user’s subscription list,

and the other is a notification of the channel owner

about a new subscriber. The following processes can

be described using two classes stored inside the

ontology (Figure 3): Update_User_Popularity_Event

and Add_Subscription_Event. To represent such a

complex event within the ontology, it is necessary to

create instances of classes for the existing events.

Figure 3: User_Subscribed object.

Several more complex events were introduced to

show how it is possible to model processes within

social networks with the created ontology.

The resulting ontology, describing social network

and identified events, is presented using the full

ontograph in Figure 4.

4 CREATING EVENT LOGS

WITH USING ONTOLOGY

To study social networks a variety of methods of

Process Mining can be used. But the event log can be

considered the main concept. The event data

described in the ontology can be used to

automatically generate event logs. To generate a log,

it is needed to solve two tasks:

1) to extract event data from social networks

using descriptions available in the ontology;

2) to generate a log according to a given form

using queries to select the desired events, the

data about which is stored in the ontology.

Applicability of the developed tools is illustrated

by examples of journals generated with the system

(fragments of the event records contained in them).

Figure 4: Social network ontology.

An Ontology-based Approach to Social Networks Mining

237

4.1 Event Logs Generation

To create an event log using the API, a request to the

API to obtain data must contain the following

information: event type (corresponds to events

classified in ontology); event description; resource

(event performer) and timestamp.

To achieve this goal, when working with the VK

API, the functions of obtaining information on the

content profile and “likes” on publications of the

content profile (“identity” block), on notifications

(“conversations”), on the distribution of content in the

community (“distribution”), on obtaining user status

(“presence”), on obtaining applications for friends

(“relationship”) were used.

After filling the ontology with the events received

from a social network, the user should be able to

unload event logs from the ontology using queries.

The generated event logs should also be stored in

the ontology for further usage in social network

mining with Process Mining tools (for example, event

logs can be exported to ProM).

4.2 Event Logs Examples

Event logs are generated in XML format. When

generating, it is possible to select events that interest

the user: you can select records by user ID, dates, or

event types, for example.

The short fragment of the event log generated by

user ID (only events for 330355947 ID are selected):

<entry>

<event>

new_ontology.

send_connection_event2

</event>

</entry>

The short fragment of the event log generated by

selected event type (only event records of type

Send_Connection_Event are uploaded):

<entry>

<event>

new_ontology.

send_connection_event1

</event>

</entry>

<entry>

<event>

new_ontology.

send_connection_event2

</event>

</entry>

Only single entries (fragments of logs that can

contain many records when analyzing real networks)

are shown above.

Users, by studying social networks, can combine

data from different journals formed independently of

each other, for example, containing information about

events in several groups.

5 CONCLUSIONS

The main result of the study at this stage is the

ontology of social media events. The presence of the

described information in the event logs allows to

analyze the connections between various events in

social networks, to identify patterns of user behavior,

to evaluate the “intensity of work” and “workload” of

users. Event logs are prepared in a standard format for

export to analyze with external applications (Process

Mining tools, for example, ProM). However, it is

possible to extend protocols with new attributes,

which enable the development of analysis tools that

take into account values of these attributes.

The next stage of the study involves

experimenting with the information, extracted from

social networks, with simulation tools (Dmitriev,

2020). This research method allows generating event

logs based on models and comparing results with real

data obtained with network mining.

REFERENCES

Abrosimova, P., Shalyaeva, I., Lyadova, L. (2018). The

Ontology-Based Event Mining Tools for Monitoring

Global Processes. In: Proceedings of the IEEE 12th

International Conference on Application of Information

and Communication Technologies (AICT 2018).

Almaty, 2018. Pp. 108-113.

Bindu, P.V., Thilagam, P.S., Ahuja, D. (2017). Discovering

suspicious behavior in multilayer social networks. In:

Computers in Human Behavior. Vol. 73, pp. 568–582.

Calvanese D., Montali M., Syamsiyah A., van der Aalst

W.M.P. (2016). Ontology-Driven Extraction of Event

Logs from Relational Databases. In: Business Process

Management Workshops. BPM 2016. Lecture Notes in

Business Information Processing, vol 256. Springer,

Cham.

KEOD 2021 - 13th International Conference on Knowledge Engineering and Ontology Development

238

Carrasquel, J.C., Chuburov, S.A., Lomazova, I.A. (2021).

Pre-processing Network Messages of Trading Systems

into Event Logs for Process Mining. In: Tools and

Methods of Program Analysis. TMPA 2019.

Communications in Computer and Information

Science, vol 1288. Springer, Cham. Pp. 88-100.

Dang-Pham, D., Pittayachawan, S., Bruno, V. (2020).

Applications of social network analysis in behavioural

information security research: Concepts and empirical

analysis. In: Computers & Security, 2017. Vol. 68, pp.

1–15.

Dawot N.I.M. (2014). A review of features and functional

building blocks of social media. In: Proceedings of the

8th. Malaysian Software Engineering Conference

(MySEC). IEEE, 2014. Pp. 177-182.

De Medeiros, A.K.A., Van der Aalst, W., Pedrinaci, C.

(2008). Semantic process mining tools: Core building

blocks. In: Proceedings of the 16th European

Conference on Information Systems. Galway, Ireland,

2008. Pp. 15-23.

Dmitriev, I, Zamyatina, E. (2020). How to Prevent Harmful

Information Spreading in Social Networks Using

Simulation Tools. In: Analysis of Images, Social

Networks and Texts. AIST 2019. Communications in

Computer and Information Science. Vol. 1086, pp.

201–213. Springer, Cham.

Ilieva, D. (2018). Fake News, Telecommunications And

Information Security. In: International Journal

“Information Theories and Applications”. Vol. 25(2),

pp. 174–181.

Kang, H., Munoz, D. (2015). A dynamic network analysis

approach for evaluating knowledge dissemination in a

multi-disciplinary collaboration network in obesity

research. In: 2015 Winter Simulation Conference

(WSC). Pp. 1319-1330.

Kietzmann J.H. (2011). Social media? Get serious!

Understanding the functional building blocks of social

media. In: Business horizons. 2011, vol. 54, №. 3. Pp.

241-251.

Mikov A., Zamyatina E., Kozlov A., Ermakov S. (2013).

Some Problems of the Simulation Model Efficiency and

Flexibility. In: Proceedings of the 8-th EUROSIM

Congress on Modelling and Simulation EUROSIM,

2013, Cardiff, Wales, United Kingdom. Pp. 532–538.

Mukala, P., Buijs, J., Leemans, M., van der Aalst, W.

(2015). Learning analytics on coursera event data: a

process mining approach. In Proceedings of the 5th

International Symposium on Data-driven Process

Discovery and Analysis (SIMPDA 2015), Vienna,

Austria, December 9-11, 2015. CEUR Workshop

Proceedings; vol. 1527. Pp. 18-32.

Peña-Araya, V. (2015). Galean: Visualization of

Geolocated News Events from Social Media. In:

Proceedings of the 38th International ACM SIGIR

Conference on Research and Development in

Information Retrieval (SIGIR '15). ACM New York.

2015. Pp. 1041-1042.

Ritter A. (2012). Open domain event extraction from twitter

In: Proceedings of the 18th ACM SIGKDD

international conference on Knowledge discovery and

data mining. 2012. Pp. 1104-1112.

Shalyaeva, I., Lyadova, L., Lanin, V. (2016). Events

Analysis Based on Internet Information Retrieval and

Process Mining Tools. In: Proceedings of 10th

International Conference on Application of Information

and Communication Technologies (AICT2016)

. Baku,

2016, pp. 168-172.

Shalyaeva, I., Lyadova, L., Lanin, V. (2017). Ontology-

Driven System for Monitoring Global Processes on

Basis of Internet News. In: Proceedings of IEEE 11th

International Conference on Application of Information

and Communication Technologies (AICT2017).

Moscow, 2017, pp. 385-389.

Tumbinskaya, M.V. (2017). Protection of information in

social networks from social engineering attacks of the

attacker. In: Journal of Applied Informatics. Vol.

12(3(69)), pp. 88–102.

Van der Aalst, W.M.P., Reijers, H. ., Song M. (2005).

Discovering social networks from event logs. In:

Computer Supported Cooperative Work (CSCW). 2005,

vol. 14, № 6. Pp. 549-593.

Vokhmintsev, A., Melnikov, A. (2013). The Knowledge on

the Basis of Fact Analysis in Business Intelligence. In:

Digital Product and Process Development Systems.

NEW PROLAMAT 2013. IFIP Advances in Information

and Communication Technology, vol 411. Springer,

Berlin, Heidelberg. Pp.134-141.

Yang, D., Liao, X., Shen, H., Cheng, X., Chen, G. (2018).

Dynamic node immunization for restraint of harmful

information diffusion in social networks. In: Physica A:

Statistical Mechanics and its Applications. Vol. 503,

pp. 640–649. DOI: 10.1016/j.physa.2018.02.128.

Zamyatina, E., Dmitriev, I. (2020). How to Prevent

Harmful Information Spreading in Social Networks

Using Simulation Tools. In: Analysis of Images, Social

Networks and Texts. AIST 2019. Communications in

Computer and Information Science. Vol. 1086, pp.

201–213. Springer, Cham.

Zavarella V. (2014). An Ontology-Based Approach to

Social Media Mining for Crisis Management. In: SSA-

SMILE@ ESWC, 2014. Pp. 55-66.

Zhan, L., Jiang, X. (2019). Survey on Event Extraction

Technology in Information Extraction Research Area.

In: Proceedings of the IEEE 3rd Information

Technology, Networking, Electronic and Automation

Control Conference (ITNEC). Chengdu, China, 2019.

Pp. 2121-2126.

Zhao, N., Cheng, X., Guo, X. (2018). Impact of information

spread and investment behavior on the diffusion of

internet investment products. In: Physica A: Statistical

Mechanics and its Applications. Vol. 512, pp. 427–436.

An Ontology-based Approach to Social Networks Mining

239