| 1. |
|
|
| 2. |
|
|
| 3. |
- Abdulla, Parosh Aziz, et al.
(författare)
-
Bisimulation minimization of tree automata
- 2007
-
Ingår i: International Journal of Foundations of Computer Science. - 0129-0541. ; 18:4, s. 699-713
-
Tidskriftsartikel (refereegranskat)abstract
- We extend an algorithm by Paige and Tarjan that solves the coarsest stable refinement problem to the domain of trees. The algorithm is used to minimize nondeterministic tree automata (NTA) with respect to bisimulation. We show that our algorithm has an overall complexity of $O(\hat{r} m \log n)$, where $\hat{r}$ is the maximum rank of any symbol in the input alphabet, m is the total size of the transition table, and n is the number of states.
|
|
| 4. |
- Abdulla, Parosh Aziz, et al.
(författare)
-
Bisimulation minimization of tree automata
- 2007
-
Ingår i: International Journal of Foundations of Computer Science. - 0129-0541. ; 18:4, s. 699-713
-
Tidskriftsartikel (refereegranskat)
|
|
| 5. |
- Abdulla, Parosh Aziz, et al.
(författare)
-
Bisimulation Minimization of Tree Automata
- 2006
-
Ingår i: Implementation and Application of Automata. - Berlin : Springer-Verlag. ; , s. 173-185
-
Konferensbidrag (refereegranskat)
|
|
| 6. |
- Abdulla, Parosh Aziz, et al.
(författare)
-
Composed Bisimulation for Tree Automata
- 2008
-
Ingår i: Implementation and Application of Automata. - Berlin, Heidelberg : Springer Berlin/Heidelberg. - 9783540708438 ; , s. 212-222
-
Konferensbidrag (refereegranskat)
|
|
| 7. |
- Abdulla, Parosh Aziz, et al.
(författare)
-
Composed bisimulation for tree automata
- 2009
-
Ingår i: International Journal of Foundations of Computer Science. - 0129-0541. ; 20:4, s. 685-700
-
Tidskriftsartikel (refereegranskat)
|
|
| 8. |
|
|
| 9. |
|
|
| 10. |
|
|
| 11. |
- Ashcroft, Michael, et al.
(författare)
-
A step towards detecting online grooming : Identifying adults pretending to be children
- 2015
-
Ingår i: Proc. 5th European Intelligence and Security Informatics Conference. - : IEEE Computer Society. - 9781479986514 ; , s. 98-104
-
Konferensbidrag (refereegranskat)abstract
- Online grooming is a major problem in todays society where more and more time is spent online. To become friends and establish a relationship with their young victims in online communities, groomers often pretend to be children. In this paper we describe an approach that can be used to detect if an adult is pretending to be a child in a chat room conversation. The approach involves a two step process wherein authors are first classified as being children or adults, and then each child is being examined and false children distinguished from genuine children. Our results show that even if it is hard to separate ordinary adults from children in chat logs it is possible to distinguish real children from adults pretending to be children with a high accuracy. In this paper we will discuss the accuracy of the methods proposed, as well as the features that were important in their success. We believe that this work is an important step towards automated analysis of chat room conversation to detect and possible attempts of grooming. Our approach where we use text analysis to distinguish adults who are pretending to be children from actual children could be used to inform children about the true age of the person that they are communicating. This would be a step towards making the Internet more secure for young children and eliminate grooming.
|
|
| 12. |
- Ashcroft, Michael, et al.
(författare)
-
Are You Really a Child? : A Machine Learning Approach To Protect Children from Online Grooming
- 2015
-
Ingår i: Proc. National Symposium on Technology and Methodology for Security and Crisis Management.
-
Konferensbidrag (refereegranskat)abstract
- Online grooming and sexual abuse of children is a major threat towards the security of todays society where more and more time is spent online. To become friends and establish a relationship with their young victims in online communities, groomers often pretend to be children. In this work we describe an approach that can be used to detect if an adult is pretending to be a child in a chat room conversation. Our results show that even if it is hard to separate ordinary adults from children in chat logs it is possible to distinguish real children from adults pretending to be children with a high accuracy.
|
|
| 13. |
- Ashcroft, Michael, et al.
(författare)
-
Detecting jihadist messages on twitter
- 2015
-
Ingår i: Proc. 5th European Intelligence and Security Informatics Conference. - : IEEE Computer Society. - 9781479986514 ; , s. 161-164
-
Konferensbidrag (refereegranskat)abstract
- Jihadist groups such as ISIS are spreading online propaganda using various forms of social media such as Twitter and YouTube. One of the most common approaches to stop these groups is to suspend accounts that spread propaganda when they are discovered. This approach requires that human analysts manually read and analyze an enormous amount of information on social media. In this work we make a first attempt to automatically detect messages released by jihadist groups on Twitter. We use a machine learning approach that classifies a tweet as containing material that is supporting jihadists groups or not. Even tough our results are preliminary and more tests needs to be carried out we believe that results indicate that an automated approach to aid analysts in their work with detecting radical content on social media is a promising way forward. It should be noted that an automatic approach to detect radical content should only be used as a support tool for human analysts in their work.
|
|
| 14. |
- Ashcroft, Michael, et al.
(författare)
-
Multi-domain alias matching using machine learning
- 2016
-
Ingår i: Proc. 3rd European Network Intelligence Conference. - : IEEE. - 9781509034550 ; , s. 77-84
-
Konferensbidrag (refereegranskat)abstract
- We describe a methodology for linking aliases belonging to the same individual based on a user's writing style (stylometric features extracted from the user generated content) and her time patterns (time-based features extracted from the publishing times of the user generated content). While most previous research on social media identity linkage relies on matching usernames, our methodology can also be used for users who actively try to choose dissimilar usernames when creating their aliases. In our experiments on a discussion forum dataset and a Twitter dataset, we evaluate the performance of three different classifiers. We use the best classifier (AdaBoost) to evaluate how well it works on different datasets using different features. Experiments show that combining stylometric and time based features yield good results on our synthetic datasets and a small-scale evaluation on real-world blog data confirm these results, yielding a precision over 95%. The use of emotion-related and Twitter-related features yield no significant impact on the results.
|
|
| 15. |
- Atig, Mohamed Faouzi, et al.
(författare)
-
Activity profiles in online social media
- 2014
-
Ingår i: Proc. 6th International Conference on Advances in Social Networks Analysis and Mining. - : IEEE Computer Society. - 9781479958771 ; , s. 850-855
-
Konferensbidrag (refereegranskat)
|
|
| 16. |
- Berggren, Mathias, et al.
(författare)
-
The Generalizability of Machine Learning Models of Personality across Two Text Domains
- 2024
-
Ingår i: Personality and Individual Differences. - : Elsevier. - 0191-8869 .- 1873-3549. ; 217
-
Tidskriftsartikel (refereegranskat)abstract
- Machine learning of high-dimensional models have received attention for their ability to predict psychological variables, such as personality. However, it has been less examined to what degree such models are capable of generalizing across domains. Across two text domains (Reddit message and personal essays), compared to low-dimensional- and theoretical models, atheoretical high-dimensional models provided superior predictive accuracy within but poor/non-significant predictive accuracy across domains. Thus, complex models depended more on the specifics of the trained domain. Further, when examining predictors of models, few survived across domains. We argue that theory remains important when conducting prediction-focused studies and that research on both high- and low-dimensional models benefit from establishing conditions under which they generalize.
|
|
| 17. |
- Berglind, Tor, et al.
(författare)
-
Levels of Hate in Online Environments
- 2019
-
Ingår i: Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2019). - New York, NY, USA : Association for Computing Machinery (ACM). - 9781450368681 ; , s. 842-847
-
Konferensbidrag (refereegranskat)abstract
- Hate speech in online environments is a severe problem for many reasons. The space for reasoning and argumentation shrinks, individuals refrain from expressing their opinions, and polarization of views increases. Hate speech contributes to a climate where threats and even violence are increasingly regarded as acceptable. The amount and the intensity of hate expressions vary greatly between different digital environments. To analyze the level of hate in a given online environment, to study the development over time and to compare the level of hate within online environments we have developed the notion of a hate level. The hate level encapsulates the level of hate in a given digital environment. We present methods to automatically determine the hate level, utilizing transfer learning on pre-trained language models with annotated data to create automated hate detectors. We evaluate our approaches on a set of websites and discussion forums.
|
|
| 18. |
- Björklund, Johanna, 1978-, et al.
(författare)
-
Aspects of plan operators in a tree automata framework
- 2012
-
Ingår i: 2012 15th International Conference on Information Fusion (FUSION). - : IEEE. - 9780982443842 - 9781467304177 ; , s. 1462-1467
-
Konferensbidrag (refereegranskat)abstract
- Plan recognition addresses the problem of inferring an agents goals from its action. Applications range from anticipating care-takers’ needs to predicting volatile situations. In this contribution, we describe a prototype plan recognition system that is based on the well-researched theory of (weighted) finite tree automata. To illustrate the system’s capabilities, we use data gathered from matches in the real-time strategy game StarCraft II. Finally, we discuss how more advanced plan operators can be accommodated for in this framework while retaining computational efficiency by taking after the field of formal model checking and over-approximating the target language.
|
|
| 19. |
- Brynielsson, Joel, et al.
(författare)
-
A Vision of a Toolbox for Intelligence Production
- 2008
-
Ingår i: Skövde Workshop on Information Fusion Topics (SWIFT 2008). ; , s. 77-80
-
Konferensbidrag (refereegranskat)abstract
- In this paper, we describe preliminary work on a toolbox aiming to help analysts involved in the intelligence production process. Intelligence analysts are overwhelmed by information, both in the form of sensory data, text stemming from human observations and other sources. In order to make sense of this information and to produce the intelligence reports needed by decision-makers, assisting computer tools are needed. We briefly describe parts of the intelligence process and touch upon the subject of what parts can and cannot be automated. A tool for tagging information semantically that we are currently working on is described, and ideas for two other tools are briefly outlined.
|
|
| 20. |
- Brynielsson, Joel, 1974-, et al.
(författare)
-
Analysis of Weak Signals for Detecting Lone Wolf Terrorists
- 2012
-
Ingår i: Proceedings of the IEEE European Intelligence and Security Informatics Conference 2012 (EISIC 2012). ; , s. 197-204
-
Konferensbidrag (refereegranskat)abstract
- Lone wolf terrorists pose a large threat to modern society. The current ability to identify and stop these kind of terrorists before they commit a terror act is limited since they are very hard to detect using traditional methods. However, these individuals often make use of Internet to spread their beliefs and opinions, and to obtain information and knowledge to plan an attack. Therefore, there is a good possibility that they leave digital traces in the form of weak signals that can be gathered, fused, and analyzed.In this work we present an analysis method that can be used to analyze extremist forums to profile possible lone wolf terrorists. This method is conceptually demonstrated using the FOI Impactorium fusion platform. We also present a number of different technologies that can be used to harvest and analyze information from Internet, serving as weak digital traces that can be fused using the suggested analysis method, in order to discover possible lone wolf terrorists.
|
|
| 21. |
- Brynielsson, Joel, et al.
(författare)
-
Detecting Social Positions Using Simulation
- 2010
-
Ingår i: Proceedings of the 2010 International Conference on Advances in Social Network Analysis and Mining (ASONAM 2010). - : IEEE. - 9780769541389 ; , s. 48-55
-
Konferensbidrag (refereegranskat)abstract
- Describing social positions and roles is an important topic within social network analysis. One approach is to compute a suitable equivalence relation on the nodes of the target network. One relation that is often used for this purpose is regular equivalence, or bisimulation, as it is known within the field of computer science. In this paper we consider a relation from computer science called simulation relation. Simulation creates a partial order on the set of actors in a network and we can use this order to identify actors that have characteristic properties. The simulation relation can also be used to compute simulation equivalence which is a less restrictive equivalence relation than regular equivalence but is still computable in polynomial time. This paper primarily considers weighted directed networks and we present definitions of both weighted simulation equivalence and weighted regular equivalence. Weighted networks can be used to model a number of network domains, including information flow, trust propagation, and communication channels. Many of these domains have applications within homeland security and in the military, where one wants to survey and elicit key roles within an organization. Identifying social positions can be difficult when the target organization lacks a formal structure or is partially hidden.
|
|
| 22. |
- Brynielsson, Joel, 1974-, et al.
(författare)
-
Development of Computerized Support Tools for Intelligence Work
- 2009
-
Ingår i: Proceedings of the 14th International Command and Control Research and Technology Symposium (14th ICCRTS). - Washington, DC.
-
Konferensbidrag (refereegranskat)abstract
- In the tasks facing the armed forces today there is a need for new and improved intelligence analysis tools. The opponents no longer follow strict doctrines that determine their behavior and force-composition. Several different opposing groups must be taken into account, some of which will appear to act friendly towards us. In this paper, we describe a vision for how various information fusion tools can be used to help intelligence analysts and decision-makers achieve situation awareness. We consider intelligence work and propose an analyst-centric toolbox aiming to help analysts involved in the intelligence production process to prepare suitable reports. Intelligence analysts are overwhelmed by information, both in the form of sensory data, text stemming from human observations and other sources. We describe parts of the intelligence process and touch upon the subject of what parts can and cannot be automated. The toolbox is outlined by describing a number of possible tools, e.g., semantic information tagging, a threat model construction assistant, a situation picture construction assistant, social network visualization, a game-theoretic reasoning engine, etc. Some of the tools described have been implemented as concept prototypes whereas others are the subject of ongoing research.
|
|
| 23. |
- Brynielsson, Joel, et al.
(författare)
-
Social Positions and Simulation Relations
- 2012
-
Ingår i: Social Network Analysis and Mining. - Wien : Springer. - 1869-5450 .- 1869-5469. ; 2:1, s. 39-52
-
Tidskriftsartikel (refereegranskat)abstract
- Describing social positions and roles is an important topic within the social network analysis. Identifying social positions can be difficult when the target organization lacks a formal structure or is partially hidden. One approach is to compute a suitable equivalence relation on the nodes of the target network. Several different equivalence relations can be used, all depending on what kind of social positions that are of interest. One relation that is often used for this purpose is regular equivalence, or bisimulation, as it is known within the field of computer science. In this paper we consider a relation from computer science called simulation relation. The simulation relation creates a partial order on the set of actors in a network and we can use this order to identify actors that have characteristic properties. The simulation relation can also be used to compute simulation equivalence which is a related but less restrictive equivalence relation than regular equivalence that is still computable in polynomial time. We tentatively term the equivalence classes determined by simulation equivalence social positions. Which equivalence relation that is interesting to consider depends on the problem at hand. We argue that it is necessary to consider several different equivalence relations for a given network, in order to understand it completely. This paper primarily considers weighted directed networks and we present definitions of both weighted simulation equivalence and weighted regular equivalence. Weighted networks can be used to model a number of network domains, including information flow, trust propagation, and communication channels. Many of these domains have applications within homeland security and in the military, where one wants to survey and elicit key roles within an organization. After social positions have been calculated, they can be used to produce abstractions of the network—smaller versions that retain some of the most important characteristics.
|
|
| 24. |
- Brynolfsson, Joel, et al.
(författare)
-
Abstraction techniques for social networks
- 2010
-
Ingår i: Proceedings of the 2010 International Conference on Advances in Social Networks Analysis and Mining. - : IEEE. - 9780769541389
-
Konferensbidrag (refereegranskat)
|
|
| 25. |
- Cohen, Katie, et al.
(författare)
-
Linguistic markers of a radicalized mind-set among extreme adopters
- 2017
-
Ingår i: Proc. 10th ACM International Conference on Web Search and Data Mining. - New York : ACM Press. - 9781450346757 ; , s. 823-824
-
Konferensbidrag (refereegranskat)abstract
- The words that we use when communicating in social media can reveal how we relate to ourselves and to others. For instance, within many online communities, the degree of adaptation to a community-specific jargon can serve as a marker of identification with the community. In this paper we single out a group of so called extreme adopters of community-specific jargon from the whole group of users of a Swedish discussion forum devoted to the topics immigration and integration. The forum is characterized by a certain xenophobic jargon, and we hypothesize that extreme adopters of this jargon also exhibit certain linguistic features that we view as markers of a radicalized mind-set. We use a Swedish translation of LIWC (linguistic inquiry word count) and find that the group of extreme adopters differs significantly from the whole group of forum users regarding six out of seven linguistic markers of a radicalized mind-set.
|
|
| 26. |
- Dahlin, Johan, et al.
(författare)
-
Combining Entity Matching Techniques for Detecting Extremist Behavior on Discussion Boards
- 2012
-
Ingår i: Advances in Social Networks Analysis and Mining (ASONAM), 2012. - : IEEE. - 9781467324977 - 9780769547992 ; , s. 850-857
-
Konferensbidrag (refereegranskat)abstract
- Many extremist groups and terrorists use the Web for various purposes such as exchanging and reinforcing their beliefs, making monitoring and analysis of discussion boards an important task for intelligence analysts in order to detect individuals that might pose a threat towards society. In this work we focus on how to automatically analyze discussion boards in an effective manner. More specifically, we propose a method for fusing several alias (entity) matching techniques, that can be used to identify authors with multiple aliases. This is one part of a larger system, where the aim is to provide the analyst with a list of potential extremist worth investigating further.
|
|
| 27. |
- Fernquist, Johan, et al.
(författare)
-
IoT Data Profiles : The Routines of Your Life Reveals Who You Are
- 2017
-
Ingår i: 2017 European Intelligence and Security Informatics Conference (EISIC). - : IEEE. - 9781538623855 ; , s. 61-67
-
Konferensbidrag (refereegranskat)abstract
- Preserving privacy is getting more and more important. The new EU general data protection regulation (GDPR) which will apply from May 2018 will introduce developments to some areas of EU data protection law and increase the privacy and personal integrity by strengthen and unify data protection for all individuals in EU. GDPR will most likely have an impact on many organizations and put pressure on many organizations that handle data. In this work, we investigate to what extent data profiles consisting of data from connected things can be used to identify a user. We use time and event profiles that can be created based on when, where and how a user communicates and uses digital devices. Our results show that such data profiles can be used to identify individuals and that collecting and creating data profiles of users can be seen as a serious threat towards privacy and personal integrity.
|
|
| 28. |
- Fernquist, Johan, et al.
(författare)
-
Online Hate A Study on the Feasibility to Detect Hate Speech in Swedish
- 2019
-
Ingår i: 2019 Ieee International Conference On Big Data (Big Data). - : IEEE. - 9781728108582 ; , s. 4724-4729
-
Konferensbidrag (refereegranskat)abstract
- Elate speech in digital environments is becoming a societal challenge. To deal with the problem, techniques that automatically detect hate speech have been developed by social media companies as well as researchers. Hate can be expressed in many different ways, which makes it difficult to detect. automatically using algorithms. Also, how hate is expressed depends heavily on the language. The effectiveness of automatic detection techniques is still to be improved in many languages. In this paper, we attempt to detect hate speech in Swedish using machine learning. We compare different pre-trained language models that are line-tuned on a corpus of hateful comments. To examine how well our models would work in a real scenario, we used a set of randomly selected comments from a Swedish discussion forum. The results showed that using pre-trained language models provides a better result than using a baseline SVM model, but it also reveals that detecting hate speech in the wild is challenge that need more research.
|
|
| 29. |
|
|
| 30. |
- Franke, Ulrik, 1981-, et al.
(författare)
-
IT-beroende och sårbarhet i det moderna samhället
- 2017
-
Ingår i: Sverige – värt att skydda. - Stockholm : Kungl Krigsvetenskapsakademien. - 9789188581006 ; , s. 109-124
-
Bokkapitel (populärvet., debatt m.m.)abstract
- Chapter 5 analyses the increasing digitalization of Sweden's Society's critical infrastructure and its connection to the internet. The increased efficiency brings with it enlarged risks and vulnerability to cyber-attacks and information operations. Vulnerabilities include both civil society and military capabilities, which are directly affected by the threat of cyber warfare and indirectly threatened through a possible collapse of civilian functions on which the armed forces depend. Sweden stands on the threshold of a new digitized society, and requires a digital immune system that reduces the vulnerability of society, through increased risk awareness among government agencies, industry, and politicians and better preparedness to counter threats by planning, exercises and possibly an early warning system.
|
|
| 31. |
- Fängström, Torbjörn, et al.
(författare)
-
Internet of Things and Future Threats Towards our Society
- 2015
-
Ingår i: Proc. National Symposium on Technology and Methodology for Security and Crisis Management.
-
Konferensbidrag (refereegranskat)abstract
- Internet of Things (IoT) is all things around us that are connected to the Internet. New technologies such as small and cheap sensors with wireless communication makes it possible to connect most of the electronic devices we use in our everyday life and according to analyst firm Gartner, close to 26 billion things to be connected to the Internet of Things in 2020. However, with new technology new threats arises. The Internet of Things in combination with the increasing number of internet users globally creates new possibilities for attacks for criminals to exploit. In this work we will investigate Internet of things and possible threats towards the security of the society.
|
|
| 32. |
- Högberg, Johanna, 1978-, et al.
(författare)
-
Weighted unranked tree automata as a framework for plan recognition
- 2010
-
Ingår i: 2010 13th International Conference on Information Fusion. - : IEEE. - 9780982443811 - 9780982443811
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- As the amount of information accessible to military intelligence continues to surge, operator assisted surveillance becomes less tractable. To process the information stream efficiently, automatic systems for threat detection are called for. These systems must be sufficiently robust to process incomplete or noisy data, and capable of dealing with uncertainties and probabilities. For safety reasons and accountability, it is imperative that the surveillance systems are specified in a formal framework that allows for rigorous mathematical verification. To this end, we demonstrate how the unobstructed keyhole plan recognition problem can be modelled within the framework of weighted unranked tree automata, and outline a software system for recognition of hostile behavior.
|
|
| 33. |
- Isbister, Tim, et al.
(författare)
-
Gender Classification with Data Independent Features in Multiple Languages
- 2017
-
Ingår i: 2017 European Intelligence and Security Informatics Conference (EISIC). - : IEEE. - 9781538623855 ; , s. 54-60
-
Konferensbidrag (refereegranskat)abstract
- Gender classification is a well-researched problem, and state-of-the-art implementations achieve an accuracy of over 85%. However, most previous work has focused on gender classification of texts written in the English language, and in many cases, the results cannot be transferred to different datasets since the features used to train the machine learning models are dependent on the data. In this work, we investigate the possibilities to classify the gender of an author on five different languages: English, Swedish, French, Spanish, and Russian. We use features of the word counting program Linguistic Inquiry and Word Count (LIWC) with the benefit that these features are independent of the dataset. Our results show that by using machine learning with features from LIWC, we can obtain an accuracy of 79% and 73% depending on the language. We also, show some interesting differences between the uses of certain categories among the genders in different languages.
|
|
| 34. |
- Johansson, Fredrik, et al.
(författare)
-
Detecting multiple aliases in social media
- 2013
-
Ingår i: Proc. 5th International Conference on Advances in Social Networks Analysis and Mining. - New York : ACM Press. - 9781450322409 ; , s. 1004-1011
-
Konferensbidrag (refereegranskat)
|
|
| 35. |
|
|
| 36. |
|
|
| 37. |
- Kaati, Lisa, 1975-, et al.
(författare)
-
A Machine Learning Approach to Identify Toxic Language in the Online Space
- 2022
-
Ingår i: 2022 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM). - : IEEE (Institute of Electrical and Electronics Engineers). - 9781665456616 ; , s. 396-402
-
Konferensbidrag (refereegranskat)abstract
- In this study, we trained three machine learning models to detect toxic language on social media. These models were trained using data from diverse sources to ensure that the models have a broad understanding of toxic language. Next, we evaluate the performance of our models on a dataset with samples of data from a large number of diverse online forums. The test dataset was annotated by three independent annotators. We also compared the performance of our models with Perspective API - a toxic language detection model created by Jigsaw and Google’s Counter Abuse Technology team. The results showed that our classification models performed well on data from the domains they were trained on (F1 = 0.91, 0.91, & 0.84, for the RoBERTa, BERT, & SVM respectively), but the performance decreased when they were tested on annotated data from new domains (F1 = 0.80, 0.61, 0.49, & 0.77, for the RoBERTa, BERT, SVM, & Google perspective, respectively). Finally, we used the best-performing model on the test data (RoBERTa, ROC = 0.86) to examine the frequency (/proportion) of toxic language in 21 diverse forums. The results of these analyses showed that forums for general discussions with moderation (e.g., Alternate history) had much lower proportions of toxic language compared to those with minimal moderation (e.g., 8Kun). Although highlighting the complexity of detecting toxic language, our results show that model performance can be improved by using a diverse dataset when building new models. We conclude by discussing the implication of our findings and some directions for future research
|
|
| 38. |
|
|
| 39. |
- Kaati, Lisa, et al.
(författare)
-
Author Profiling in the Wild
- 2017
-
Ingår i: 2017 European Intelligence and Security Informatics Conference (EISIC). - : IEEE. - 9781538623855 ; , s. 155-158
-
Konferensbidrag (refereegranskat)abstract
- In this paper, we use machine learning for profiling authors of online textual media. We are interested in determining the gender and age of an author. We use two different approaches, one where the features are learned from raw data and one where features are manually extracted. We are interested in understanding how well author profiling works in the wild and therefore we have tested our models on different domains than they are trained on. Our results show that applying models to a different domain then they were trained on significantly decreases the performance of the models. The results show that more efforts need to be put into making models domain independent if techniques such as author profiling should be used operationally, for example by training on many different datasets and by using domain independent features.
|
|
| 40. |
|
|
| 41. |
|
|
| 42. |
- Kaati, Lisa, et al.
(författare)
-
Detecting multipliers of jihadism on twitter
- 2015
-
Ingår i: Proc. 15th ICDM Workshops. - : IEEE Computer Society. - 9781467384926 ; , s. 954-960
-
Konferensbidrag (refereegranskat)abstract
- Detecting terrorist related content on social media is a problem for law enforcement agency due to the large amount of information that is available. In this paper we describe a first step towards automatically classifying twitter user accounts (tweeps) as supporters of jihadist groups who disseminate propaganda content online. We use a machine learning approach with two set of features: data dependent features and data independent features. The data dependent features are features that are heavily influenced by the specific dataset while the data independent features are independent of the dataset and that can be used on other datasets with similar result. By using this approach we hope that our method can be used as a baseline to classify violent extremist content from different kind of sources since data dependent features from various domains can be added.
|
|
| 43. |
- Kaati, Lisa, et al.
(författare)
-
Digitala diskussioner och de svenska valen 2022
- 2023
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- För att människor ska kunna lita på att deras röster räknas korrekt och att valresultatet speglar det faktiska stödet för de olika kandidaterna och politiska partierna, måste valsystemet vara säkert och pålitligt. Lika viktigt som att valsystemet faktiskt är säkert och pålitligt är att det finns ett förtroende för att valen genomförs korrekt och för de myndigheter som ansvarar för valens genomförande. I den här rapporten beskrivs en analys av diskussioner om valens genomförande i ett urval av sociala medier som kännetecknas av ett fokus på samhällsfrågor och politik. Analyserna bygger på fyra olika teman kopplade till valens genomförande: förtidsröster, valsedlar, röstmottagning och räkning samt valfusk. Resultaten visar att det förekommer ett flertal narrativ om oegentligheter kopplade till allmänna val. Det rör sig om allt från oro för att oegentligheter ska ske, spekulationer om att oegentligheter kommer att ske och upplevda oegentligheter. Det förekommer också diskussioner och rapporter om att genomförandet av valen faktiskt fungerar som det ska och att det svenska valsystemet är robust.
|
|
| 44. |
- Kaati, Lisa, et al.
(författare)
-
En studie i fördom : Om rasistiska stereotyper i digitala miljöer
- 2022
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- I denna rapport studeras uttryck för rasism i digitala miljöer, i första hand fördomar – det vill säga ogrundade negativa värderingar av individer baserade på deras grupptillhörighet – mot de minoritetsgrupper som i Regeringens plan mot rasism, liknande former av fientlighet och hatbrott angivits som speciellt utsatta. Bland de inlägg som nämnt etniska minoriteter och publicerats på diskussionsforumet Reddit från januari till juli 2022, innehöll 16 % fördomar och ytterligare 3% andra former av negativa attityder mot etniska minoriteter. De fördomar som uttrycktes kunde indelas i fyra kategorier: Föreställningar om grupper som (1) skapar oro eller oreda i samhället , (2) har odemokratiska eller föråldrade värderingar, (3) inte kan integreras eller (4) väljer att försörja sig på bidrag eller ekonomisk brottslighet framför arbete. Vår undersökning visar att muslimer och andra personer med ursprung i Mellanöstern och Nordafrika är särskilt utsatta för andras fördomar. Fördomsfulla inlägg innehåller ofta okunskap om och sammanblandning av olika etniska och religiösa minoriteter. Eftersom undersökningen gjordes våren 2022 har vi testat resultaten mot mediebevakade händelser under tidsperioden. En kraftig ökning av islamofobiska uttryck sammanföll med rapporteringen om de så kallade påskupploppen i april 2022.
|
|
| 45. |
- Kaati, Lisa, 1975-, et al.
(författare)
-
General Risk Index : A Measure for Predicting Violent Behavior Through Written Communication
- 2023
-
Ingår i: 2023 IEEE International Conference on Big Data (BigData). - : IEEE (Institute of Electrical and Electronics Engineers). - 9798350324464 ; , s. 4065-4070
-
Konferensbidrag (refereegranskat)abstract
- One of the most challenging threats to the security of society is attacks from violent lone offenders. Identifying potential offenders is difficult since they act alone and do not necessarily communicate with others. However, several targeted violent attacks have been preceded by communication published on social media and the internet. Such communication is a valuable component when conducting risk and threat assessments.In this paper, we introduce a diagnostic measure of the risk of violent behavior based on text analysis. Using automated text analysis, we extract psychological variables and warning indicators from a given text and summarize these in an index that we denote as the general risk index. When developing the general risk index, we analyzed data (text) from 208 288 users on 32 online environments with diverse ideologies/orientations, including 76 previous violent lone offenders. A receiver operating characteristics (ROC) analysis showed that, when using the general risk index, it was possible to correctly classify between 90% and 96% of the cases depending on the comparison sample. These results support the predictive validity of the general risk index, suggesting that the risk index can be used to identify individuals with an increased risk of committing violent attacks that need further investigation.
|
|
| 46. |
- Kaati, Lisa, et al.
(författare)
-
Identifying Warning Behaviors of Violent Lone Offenders in Written Communication
- 2016
-
Ingår i: 2016 IEEE 16Th International Conference On Data Mining Workshops (ICDMW). - New York : IEEE. - 9781509059102 ; , s. 1053-1060
-
Konferensbidrag (refereegranskat)abstract
- Violent lone offenders such as school shooters and lone actor terrorists pose a threat to the modern society but since they act alone or with minimal help form others they are very difficult to detect. Previous research has shown that violent lone offenders show signs of certain psychological warning behaviors that can be viewed as indicators of an increasing or accelerating risk of committing targeted violence. In this work, we use a machine learning approach to identify potential violent lone offenders based on their written communication. The aim of this work is to capture psychological warning behaviors in written text and identify texts written by violent lone offenders. We use a set of features that are psychologically meaningful based on the different categories in the text analysis tool Linguistic Inquiry and Word Count (LIWC). Our study only contains a small number of known perpetrators and their written communication but the results are promising and there are many interesting directions for future work in this area.
|
|
| 47. |
|
|
| 48. |
- Kaati, Lisa, 1975-, et al.
(författare)
-
Predicting Targeted Violence from Social Media Communication
- 2022
-
Ingår i: 2022 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM). - 9781665456616 ; , s. 383-390
-
Konferensbidrag (refereegranskat)abstract
- For decades, threat assessment professionals have used structured professional judgment instruments to make decisions about, for example, the likelihood of violent behavior of an individual. However, with the increased use of social media, most people use online digital platforms to communicate, which is also the case for potential violent offenders. For example, many mass shootings in recent years have been preceded by communication in online forums. In this paper, we introduce methods to identify markers of the warning behaviors Leakage, Fixation, Identification, and Affiliation and examine their discriminant validity. Our results show that violent offenders score higher on these markers and that these markers were present among a significantly higher proportion of violent offenders as compared to the normal population. We argue that our method can be used to predict potential planned, purposeful, or instrumental targeted violence in written communication. Automated methods for detecting warning behavior from written communication can serve as a complement to traditional threat assessment and provides unique opportunities for threat assessment beyond traditional methods.
|
|
| 49. |
- Kaati, Lisa, 1975-
(författare)
-
Reduction Techniques for Finite (Tree) Automata
- 2008
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Finite automata appear in almost every branch of computer science, for example in model checking, in natural language processing and in database theory. In many applications where finite automata occur, it is highly desirable to deal with automata that are as small as possible, in order to save memory as well as excecution time.Deterministic finite automata (DFAs) can be minimized efficiently, i.e., a DFA can be converted to an equivalent DFA that has a minimal number of states. This is not the case for non-deterministic finite automata (NFAs). To minimize an NFA we need to compute the corresponding DFA using subset construction and minimize the resulting automaton. However, subset construction may lead to an exponential blow-up in the size of the automaton and therefore even if the minimal DFA may be small, it might not be feasible to compute it in practice since we need to perform the expensive subset construction.To aviod subset construction we can reduce the size of an NFA using heuristic methods. This can be done by identifying and collapsing states that are equal with respect to some suitable equivalence relation that preserves the language of the automaton. The choice of an equivalence relation is a trade-off between the desired amount of reduction and the computation time since the coarser a relation is, the more expensive it is to compute. This way we obtain a reduction method for NFAs that is useful in practice.In this thesis we address the problem of reducing the size of non-deterministic automata. We consider two different computation models: finite tree automata and finite automata. Finite automata can be seen as a special case of finite tree automata and all of the previously mentioned results concerning finite automata are applicable to tree automata as well. For non-deterministic bottom-up tree automata, we present a broad spectrum of different relations that can be used to reduce their size. The relations differ in their computational complexity and reduction capabilities. We also provide efficient algorithms to compute the relations where we translate the problem of computing a given relation on a tree automaton to the problem of computing the relation on a finite automaton.For finite automata, we have extended and re-formulated two algorithms for computing bisimulation and simulation on transition systems to operate on finite automata with alphabets. In particular, we consider a model of automata where the labels are encoded symbolically and we provide an algorithm for computing bisimulation on this partial symbolic encoding.
|
|
| 50. |
- Kaati, Lisa
(författare)
-
Samtalstonen i sociala medier
- 2023
-
Ingår i: Hot mot det demokratiska samtalet. - : Sveriges Kommuner och Regioner. - 9789180471398 ; , s. 49-56
-
Bokkapitel (övrigt vetenskapligt/konstnärligt)abstract
- Lisa Kaati diskuterar hur man med hjälp av nya tekniker kan identifiera toxiskt språk för att skapa en mer anständig samtalston på sociala medier. Hon konstaterar att internet erbjuder fantastiska möjligheter för alla att delta i diskussioner när som helst och om vad som helst. Men alltför ofta präglas samtalsklimatet av toxiskt språk. Begreppet används för att beskriva kommunikation som förgiftar samtalsklimatet i sociala medier. Det kan vara kommunikation som är förbjuden i lag (hets mot folkgrupp, förtal) men också andra former av kränkningar som nedsättande tilltal, respektlöshet eller integritetskränkningar. Att upprepade gånger utsättas för toxiska kommentarer innebär en oerhörd påfrestning och kan leda till att man väljer att dra sig tillbaka från det offentliga samtalet. Det kan i sin tur innebära att vissa röster tystnar och att de mer lågmälda och diskuterande samtalen försvinner. På många av de stora sociala medieplattformarna finns användarvillkor som förbjuder viss typ av kommunikation men det finns också en stor mängd plattformar som inte har några regler för vad som får publiceras så länge det inte bryter mot någon lag. Eftersom många av dessa plattformar finns i USA är det amerikansk lag som gäller. För att hantera mängden av kommunikation har nya tekniker utvecklats för att identifiera toxiskt språk automatiskt. Dessa tekniska lösningar bygger på olika typer av textanalys. Det sker främst genom maskininlärningsbaserade tekniker där datorn själv lär sig att känna igen toxiskt språk. Det kräver i sin tur tillräckligt många och varierade exempel på vad som är toxiskt språk. Idag använder många sociala medieplattformar automatiserade tekniker för att hitta inlägg i kommentarsfält som inte följer användarreglerna. Det har även tagits fram andra verktyg som gör det möjligt att undvika toxiska kommentarer genom att dölja dem eller att varna/förmana den som skriver genom att markera innehåll som kan uppfattas som toxiskt och vara konfliktdrivande. Lisa Kaati framhåller att automatiserade tekniker är nödvändiga i vårt digitala samhälle för att det ska vara möjligt att identifiera toxiskt språk. Samtidigt finns det ett stort behov av att utveckla och förfina metoderna. Vidare krävs medvetenhet om metodernas begränsningar samt om betydelsen av att använda dem på ett ansvarsfullt och etiskt sätt.
|
|
