Project

RiskTrack - Tracking tool based on social media for risk assessment on radicalisation

Goal: This project aims to help in the prevention of terrorism through the identification of radicalisation. In line with the EU priorities in this matter, the team of experts will identify and tackle the factors or indicators that raise a red flag about which individuals or communities are being radicalised and recruited to commit violent acts of terrorism.

Thus, the project aims to develop a risk assessment methodology studying how to detect signs of radicalisation (e.g., use of language, behavioural patterns in social networks...) and to allow these signs to be compared against other web features (and patterns) extracted from sources such as social networks or social media. These features will be tested and analysed using advanced data mining methods, knowledge representation (semantic and ontology engineering) and multilingual technologies.

This project is supported by the European Commission (grant number: 723180).

The partners in charge of this project are the following:
- Universidad Autónoma de Madrid (UAM, Spain)
- Universite Lyon 1 Claude Bernard (UCBL, France)
- Parc Sanitari Sant Joan De Déu (PSSJDD, Spain)
- Cyprus Neuroscience And Technology Institute (CNTI, Cyprus)

Methods: Data Mining, Big Data, Social Media, Complex Networks, Semantic Web, Profiling, EGO Networks, Risk Assessment

Date: 1 October 2016 - 30 September 2018

Updates
0 new
31
Recommendations
0 new
7
Followers
2 new
35
Reads
9 new
612

Project log

Francisco Javier Torregrosa López
added a research item
The present study analyzed the differences in the language usage between pro-ISIS users and random users on Twitter. Based on the literature, it was expected that, when comparing the tweets from both samples, distinctive patterns would be found on their usage of similar linguistic categories. This observational study compared a dataset of 105 pro-ISIS users with 91 random Twitter users, both collected between 2015 and 2016. The Linguistic Inquiry Word Count (LIWC) software was employed to analyze the terminology used by both groups from a quantitative perspective. Relevant LIWC categories used in previous studies were included in the assessment. ISIS supporters used significantly more third person plural pronouns and less first person singular and second person pronouns. They also used more words related with death, certainty, and anger than the random group, along with more words containing six letters or more. Finally, more negative language and tone was used by the pro-ISIS group. The language used by ISIS supporters on Twitter was discussed, as well as comparisons to relevant studies on other political extremists. Ultimately, our results suggest that broad similarities in language usage exist between ISIS supporters and other extreme ideologies. ARTICLE HISTORY
Francisco Javier Torregrosa López
added a research item
Twitter is one of the most commonly used Online Social Networks in the world and it has consequently attracted considerable attention from different political groups attempting to gain influence. Among these groups is the alt-right; a modern far-right extremist movement that gained notoriety in the 2016 US presidential election and the infamous Charlottesville Unite the Right rally. This article details the process used to create a database on Twitter of users associated with this movement, allowing for empirical research into this extremist group to be undertaken. In short, Twitter accounts belonging to leaders and groups associated with the Unite the Right rally in Charlottesville were used to create this database. After compiling users who followed these leading alt-right figures, an initial sample was gathered (n = 549). This sample was then validated by two researchers. This involved using a scoring method created for this process, in order to eliminate any accounts that were not supportive of the alt-right. Finally, a total amount of 422 accounts were found to belong to followers of this extremist movement, with a total amount of 123.295 tweets.
Francisco Javier Torregrosa López
added a research item
RiskTrack is a project supported by the European Union, with the aim of helping security forces, intelligence services and prosecutors to assess the risk of Jihadi radicalization of an individual (or a group of people). To determine the risk of radicalization of an individual, it uses information extracted from its Twitter account. Specifically, the tool uses a combination of linguistic factors to establish a risk value, in order to help the analyst with the decision making. This article aims to describe the linguistic features used on the first prototype of the RiskTrack tool. These factors, along with the way of calculating them and their contribution to the final risk value, will be presented in this paper. Also, some comments about the tool and the next updates will be suggested at the end of this paper.
Andreas Andreou
added an update
Torregrosa J., Panizo Á. (2018) “RiskTrack: Assessing the Risk of Jihadi Radicalization on Twitter Using Linguistic Factors”. In: Yin H., Camacho D., Novais P., Tallón-Ballesteros A. (eds) Intelligent Data Engineering and Automated Learning – IDEAL 2018. IDEAL 2018. Lecture Notes in Computer Science, vol 11315. Springer, Cham
 
Andreas Andreou
added an update
Torregrosa J., Gilpérez-López I, Lara-Cabrera R., Garriga D., Camacho D. (2017) “Can an Automatic Tool Assess Risk of Radicalization Online? A Case Study on Facebook”. European Intelligence and Security Informatics Conference (EISIC). IEEE.
 
Andreas Andreou
added an update
Thorburn J., Torregrosa J., Panizo Á. (2018) “Measuring Extremism: Validating an Alt-Right Twitter Accounts Dataset”. In: Yin H., Camacho D., Novais P., Tallón-Ballesteros A. (eds) Intelligent Data Engineering and Automated Learning – IDEAL 2018. IDEAL 2018. Lecture Notes in Computer Science, vol 11315. Springer, Cham
 
Andreas Andreou
added an update
Masmoudi A., Barhamgi M., Faci N., Saoud Z., Belhajjame K., Benslimane D., Camacho D. (2018) “An Ontology-Based Approach for Mining Radicalization Indicators from Online Messages”. In the Proceedings of the 32nd International Conference on Advanced Information Networking and Applications (AINA), IEEE.
 
Andreas Andreou
added an update
Lara-Cabrera R., Pardo AG., Benouaret K., Faci N., Benslimane D., Camacho D. (2017d) “Measuring the Radicalisation Risk in Social Networks”. IEEE Access, vol 5, pp 10892-10900
 
Andreas Andreou
added an update
Lara-Cabrera R., Gonzalez-Pardo A., Camacho D. (2017c) “Statistical analysis of risk assessment factors and metrics to evaluate radicalisation in Twitter”. Future Generation Computer Systems. Elsevier B.V.
 
Andreas Andreou
added an update
Lara-Cabrera R., Gonzalez-Pardo A., Camacho D. (2017b) “Linguist markers to early detection of radicalization in Social Networks”. In the Proceedings of the 1st International Caparica Conference in Translational Forensics
 
Andreas Andreou
added an update
Lara-Cabrera R., Gonzalez-Pardo A., Barhamgi M., Camacho D. (2017a) “Extracting radicalisation behavioural patterns from social network data”. In the Proceedings of the 28th International Workshop on Database and Expert Systems Applications. IEEE Computer Society
 
Andreas Andreou
added an update
Gonzalez-Pardo A., Lara-Cabrera R., Camacho D. (2017b) “Representation of information in Social Network to Perform Community Finding Tasks”. In the Proceedings of the 1st International Caparica Conference in Translational Forensics
 
Andreas Andreou
added an update
Gonzalez-Pardo A., Lara-Cabrera R., Camacho D. (2017a) “Estudio de factores para la evaluación del riesgo de radicalización en redes sociales”. In the Proceedings of the V Congreso Nacional de i+d en Defensa y Seguridad (DESEi+d 2017)
 
Andreas Andreou
added an update
Gonzalez-Pardo, A., & Camacho, D. (2018b). Design of an ACO algorithm for Solving Community Finding Problems. In the Proceeding of the XVIII Conferencia de la Asociacion Espanola para la Inteligencia Artificial, pp. 1001-1002.
 
Andreas Andreou
added an update
Gonzalez-Pardo, A., & Camacho, D. (2018a). Design of Japanese Tree Frog Algorithm for Community Finding Problems. In International Conference on Intelligent Data Engineering and Automated Learning (pp. 307-315). Springer, Cham.
 
Andreas Andreou
added an update
Gmati, H., Mouakher, A., Gonzalez-Pardo, A., & Camacho, D. (2018) A new algorithm for communities detection in social networks with node attributes. Journal of Ambient Intelligence and Humanized Computing, 1-13.
 
Andreas Andreou
added an update
Gilpérez-López I., Torregrosa J., Barhamgi M., Camacho D. (2017) “An initial study on radicalization risk factors: Towards an assessment software tool”. In the Proceedings of the 28th International Workshop on Database and Expert Systems Applications, pp. 11-16. IEEE Computer Society
 
Andreas Andreou
added an update
Fang, X. S., Sheng, Q. Z., Wang, X., Barhamgi, M., Yao, L., & Ngu, A. H. (2017) SourceVote: Fusing multi-valued data via inter-source agreements. In International Conference on Conceptual Modeling (pp. 164-172). Springer, Cham.
 
Andreas Andreou
added an update
Camacho, D., Gilpérez-López, I., Gonzalez-Pardo, A., Ortigosa, A., & Urruela, C. (2016). RiskTrack: a new approach for risk assessment of radicalisation based on social media data. In CEUR Workshop Proceedings.
 
Andreas Andreou
added an update
Benouaret K., Benouaret I, Barhamgi M., Djamal D. (2017) “Top-k Cloud Service Plans using Trust and QoS”. In Proceedings of the 14th International Conference on Services Computing, pp. 507-508. IEEE.
 
Andreas Andreou
added an update
Barhamgi, M., Yang, M., Yu, C. M., Yu, Y., Bandara, A. K., Benslimane, D., & Nuseibeh, B. (2017) “Enabling End-Users to Protect their Privacy”. In Proceedings of the 2017 ACM on Asia Conference on Computer and Communications Security, pp. 905-907. ACM.
 
Andreas Andreou
added an update
Program for the 19th International Conference on Intelligent Data Engineering and Automated Learning
 
David Camacho
added 4 research items
The unstoppable growth that Social Networks (SN) have suffered in the last years, has produced that the data stored in those networks grows exponentially. This data appears from the information that the users provide in their corresponding profile, the different connections that they stablished while they are using the SN, and also due to the different interactions that the user perform within the SN. All this data has become in a great opportunity to extract information from the SN and also from the users. One of the most typical information that can be extracted from this data is the different groups, or clusters, of users. The main idea is to gather users in one or more groups in such a way users belonging to the same group are similar, whereas there are several differences among the users of the other groups. This problem is commonly known as Community Finding Problems, and the different groups of users are called "communities" [1]. The data used to perform the community detection task is critical because it will affect to the quality of the communities found by the algorithms. In this way, it is possible to detect the different communities based on data extracted from the network (such as relation between users), or based on the information provided by the users in their profiles [1]. But it is also possible to compute other metrics related to the users behaviour in order to define the different communities [2, 3, 4]. The goal of this paper is to analyse the different approaches that can be used to perform the Community Finding Tasks taking into account the different types of data available in the most popular Social Network.
With the growth of the Islamic terrorism and groups like Islamic State or Al Qaeda, academics have focused on detecting risk factors to understand the radicalisation phenomena and, thus, preventing it. Even though there is not a single profile of an Islamic radical, most of the authors point to several conditions that are partially shared by most of those radicals. Moreover, the development of the Online Social Networks, such as Facebook or Twitter, along with the Internet, has created a new field for the radicals to start their radicalisation paths, but also a new chance to detect the behavioral changes they present there. With this change on the rules in mind, this paper focuses on assessing the behavioral traces that can represent a sign that a person is becoming radicalised on the Online Social Networks. Both theoretical [1] and empirical information [2, 3] will be take into account in order to create a final report of the radicalisation risk factors and their indicators presented on the Online Social Networks. Table 1. Personal risk factors Table 2. Group risk factors
Nowadays, social networks are essential communication tools that produce a large amount of information about their users and their interactions. Spreading propaganda in digital environments is a good way for various extremist groups who want to reach out with their messages, and it is considered to be an important part of the terrorist group Islamic State (IS) success in recruiting supporters from all over the world. Although propaganda is not the sole cause of radicalisation or recruitment to violent extremist ideologies, interactions on social networks can be an important component of a radicalization process due to its easy accessibility and the ability to capture and retain an individual's interest. Even though it is not clear what role Internet and social media plays in radicalization, some previous studies have focused on measuring the risk for individuals to radicalization [1,2] and the possibility to detect individuals or groups that engage in violent extremism [3]. In this work, we focus on identifying a set of linguistic indicators that can be used to measure frustration, the perception of discrimination, and the declaration of negative and positive ideas about the Western society and Violent extremism respectively. The indicators have been tested on three different datasets: tweets by pro-ISIS users, tweets from users flagged as radicals by the Anonymous collective and a random sample of tweets gathered from the public Twitter stream. Figure 1. Density distribution of the ratio of tweets expressing positive ideas about Jihadism according to the studied indicators and their respective metrics. References [1]
David Camacho
added a research item
Nowadays, Social Networks have become an essential communication tools producing a large amount of information about their users and their interactions, which can be analysed with Data Mining methods. In the last years, Social Networks are being used to radicalise people. In this paper, we study the performance of a set of indicators and their respective metrics, devoted to assess the risk of radicalisation of a precise individual on three different datasets. Keyword-based metrics, even though depending on the written language, performs well when measuring frustration, perception of discrimination as well as declaration of negative and positive ideas about Western society and Jihadism, respectively. However, metrics based on frequent habits such as writing ellipses are not well enough to characterise a user in risk of radicalisation. The paper presents a detailed description of both, the set of indicators used to assess the radicalisation in Social Networks and the set of datasets used to evaluate them. Finally, an experimental study over these datasets are carried out to evaluate the performance of the metrics considered.
David Camacho
added a project reference
Andreas Andreou
added an update
The scientific paper "Can an Automatic Tool Assess Risk of Radicalization online? A Case Study on Facebook" produced in the context of the RiskTrack project has been accepted for publication at the proceedings of the 2017 European Intelligence and Security Informatics Conference (EISIC 2017).
 
Andreas Andreou
added an update
On May 3, 2017, Dr. David Camacho gave a lecture on New Applications of Artificial Intelligence at the Universidad Complutense de Madrid.
 
Andreas Andreou
added an update
The Advisory Board of the project has just been established! Follow the link to read the synthesis of the AB constituting of experts in the fields of radicalization and terrorism, psychology, computer science and linguistics.
 
David Camacho
added a research item
Social Networks (SNs) have become a powerful tool for the jihadism as they serve as recruitment assets, live forums, psychological warfare as well as sharing platforms. SNs enable vulnerable individuals to reach radicalised people hence triggering their own radicalisation process. There are many vulnerability factors linked to socio-economic and demographic conditions that make jihadist militants suitable targets for their radicalisation. We focus on these vulnerability factors, studying, understanding and identifying them on the Internet. Here we present a set of radicalisation indicators and a model to assess them using a dataset of tweets published by several Islamic State of Iraq and Sham sympathizers. Results show that there is a strong correlation between the values assigned by the model to the indicators.
David Camacho
added an update
Here you can find a short description of our project recently published in a workshop. The paper provides a gentle introduction to the basic goals of our project.
Do not hesitate to contact me for further information!
David
 
Andreas Andreou
added an update
Andreas Andreou, forensic linguist at the Cyprus Neuroscience & Technology Institute (CNTI), presented the key elements of the project RiskTrack at the seminar "The language of crime" which was co-organised by the Constanteion Institute of Criminology and Forensic Sciences and the CNTI. The presentation entitled "The language as evidence in the investigation of the crime: Introduction to Forensic Linguistics" provided an overview to the audience, comprised of LEAs and lawyers, of the field of Forensic Linguistics emphasizing how linguistic knowledge and methodology can be valuable in the investigation and combating of crime. The seminar was held at the European University Cyprus on November 25, 2016.
 
Irene Gilpérez López
added an update
Today we finished the kick-off meeting of the RiskTrack project.
It has been two really productive days in which we have exchanged different points of view, ideas and knowledge, and have organised our next steps in the project.
In representation of the UAM team, thank you all very much for coming to Madrid, it was a real pleasure to have you here.
 
David Camacho
added an update
Principal Investigator
 
Irene Gilpérez López
added a project goal
This project aims to help in the prevention of terrorism through the identification of radicalisation. In line with the EU priorities in this matter, the team of experts will identify and tackle the factors or indicators that raise a red flag about which individuals or communities are being radicalised and recruited to commit violent acts of terrorism.
Thus, the project aims to develop a risk assessment methodology studying how to detect signs of radicalisation (e.g., use of language, behavioural patterns in social networks...) and to allow these signs to be compared against other web features (and patterns) extracted from sources such as social networks or social media. These features will be tested and analysed using advanced data mining methods, knowledge representation (semantic and ontology engineering) and multilingual technologies.
This project is supported by the European Commission (grant number: 723180).
The partners in charge of this project are the following:
- Universidad Autónoma de Madrid (UAM, Spain)
- Universite Lyon 1 Claude Bernard (UCBL, France)
- Parc Sanitari Sant Joan De Déu (PSSJDD, Spain)
- Cyprus Neuroscience And Technology Institute (CNTI, Cyprus)