The InVID TV Fake Video Corpus is a small collection of verified fake videos. It was developed in... more The InVID TV Fake Video Corpus is a small collection of verified fake videos. It was developed in the context of the InVID project with the aim of gaining a perspective of the types of fake video that can be encountered in the real world. Currently the Corpus consists of 59 videos. For each video, information is provided describing the fake, its original source, and the evidence proving it is a fake. As we do not own the videos, the dataset only provides the video URLs and metadata, in the form of a tab-separated value (TSV) file. See the README file for more information.
A comprehensive study prepared for the Broadband Commission for Sustainable Development (UNESCO a... more A comprehensive study prepared for the Broadband Commission for Sustainable Development (UNESCO and ITU). 'Balancing Act: Countering Digital Disinformation while respecting Freedom of Expression' surveys issues with a global scope, includes an action-oriented suite of sector-specific actionable recommendations, and presents a 23-point framework to test disinformation responses. Targeted analyses and recommendations to address the life cycle of online disinformation from production to transmission, reception and reproduction focus on: *Legislators and policy makers (counter-disinformation campaigns, electoral-specific responses, the Freedom of Expression Assessment Framework) (Ch5, Ch8, Executive Summary) *Internet companies, producers and distributors (content curation, technical and algorithmic, advertisement policy, demonetisation responses) (Ch6, Executive Summary) *Journalists, investigative researchers and fact checkers (Ch4, Ch3, Ch2, Executive Summary) *Universities and applied and empirical researchers (Ch3, Executive Summary) Target audiences (educational, ethical and normative, empowerment and credibility labeling responses) [Ch7, Executive Summary) The findings are organized into a typology of 11 different categories of responses to disinformation – ranging from identification and investigatory responses, through to policy and legislative measures, technological steps, and educational approaches. For each category of response, the report includes a description of work being done around the world, by which actors, how it is funded and who or what is targeted. The report further analyses the underlying assumptions and theories of change behind these responses, while weighing the challenges and opportunities. Each category of response is also assessed in terms of its intersections with the universal human right of freedom of expression, with a particular focus on press freedom and access to information. Case studies of responses to COVID-19 disinformation are also presented within each category. At the heart of this report is the need to balance responses to disinformation with respect for freedom of expression. The research shows us that this can be done. This research of the Broadband Commission study is edited by Professor Kalina Bontcheva (University of Sheffield, UK) and Dr. Julie Posetti (International Center for Journalists,U.S.; Centre for Freedom of the Media, University of Sheffield/ Reuters Institute for the Study of Journalism, University of Oxford, UK). The other contributing authors are Denis Teyssou (Agence France Presse, France); Dr. Trisha Meyer (Vrije Universiteit Brussel, Belgium); Sam Gregory (WITNESS, U.S.); Clara Hanot (EU Disinfo Lab, Belgium); and Dr. Diana Maynard (University of Sheffield, UK).
Proceedings of the 1st International Workshop on Multimedia AI against Disinformation
Enabled by recent improvements in generation methodologies, DeepFakes have become mainstream due ... more Enabled by recent improvements in generation methodologies, DeepFakes have become mainstream due to their increasingly better visual quality, the increase in easy-to-use generation tools and the rapid dissemination through social media. This fact poses a severe threat to our societies with the potential to erode social cohesion and influence our democracies. To mitigate the threat, numerous DeepFake detection schemes have been introduced in the literature but very few provide a web service that can be used in the wild. In this paper, we introduce the MeVer DeepFake detection service, a web service detecting deep learning manipulations in images and video. We present the design and implementation of the proposed processing pipeline that involves a model ensemble scheme, and we endow the service with a model card for transparency. Experimental results show that our service performs robustly on the three benchmark datasets while being vulnerable to Adversarial Attacks. Finally, we outline our experience and lessons learned when deploying a research system into production in the hopes that it will be useful to other academic and industry teams. CCS CONCEPTS • Information systems → Multimedia information systems; Web services; Data analytics; • Security and privacy → Human and societal aspects of security and privacy.
2020 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)
This paper presents an overview of the WeVerify H2020 EU project, which develops intelligent huma... more This paper presents an overview of the WeVerify H2020 EU project, which develops intelligent human-in-the-loop content verification and disinformation analysis methods, tools and services. Social media and web content are analysed and contextualised within the broader online ecosystem, in order to expose fabricated content, through cross-modal content verification, social network analysis, micro-targeted debunking, and a blockchain-based public database of known fakes.
The production and spreading of manipulated videos have been on the rise over the past years, and... more The production and spreading of manipulated videos have been on the rise over the past years, and is expected to continue and increase further. Manipulating videos have become easier from a technological perspective, and can be done with freely available tools that require less expert knowledge and fewer resources than in the past. All this poses new challenges for those who aim to tackle the spreading of false, manipulated or misleading video content. This chapter covers many of the aspects raised above. It deals with the motivations of those involved in video verification, showcases respective requirements and highlights the importance and relevance of tackling disinformation on social networks. Furthermore, an overview of the state of the art of available techniques and technologies is provided. The chapter then describes the emergence of new threats like so-called ‘deep fakes’ created with the help of artificial intelligence. Finally, we formulate an empirical typology of false videos spreading online.
The InVID TV Fake Video Corpus was developed in the context of the InVID project with the aim of ... more The InVID TV Fake Video Corpus was developed in the context of the InVID project with the aim of gaining a perspective of the types of fake video that can be encountered in the real world. The dataset does not aspire to serve as an exhaustive list of all forgeries that have circulated the Web in the past, but we intend to maintain and extend it throughout the course of the project as new cases arise. The collection is a collaborative effort between AFP and CERTH-ITI. This is the second version of the dataset, containing 117 fake videos and 110 real videos, alongside annotations and descriptions. As we do not own the rights to the videos, the dataset only contains the video URLs and annotations.
Abstract. This paper presents the approach of the Glocal European funded project towards the co-p... more Abstract. This paper presents the approach of the Glocal European funded project towards the co-production of news on World events, on an information marketplace involving both amateurs and professionals. It discuss the rise of user-generated content amongst worldwide media and how event modelling and technology usage may help to foster this pro-am collaboration. Glocal: Event-based Retrieval of Networked Media [1] is a large scale integrating and collaborative project which started in December 2009 for three years and involves ten partners.
This chapter discusses measures to tackle disinformation through content curation or moderation w... more This chapter discusses measures to tackle disinformation through content curation or moderation within internet communication companies and journalism processes. The effect of such measures affects inter alia what content is allowed on the service; if it is allowed to remain up, if it is fact-checked; its prominence and visibility; whether advertising appears next to it; the degree to which it is automatically recommended or limited in terms of distribution; whether it is labelled, as well as what kinds of paid content appear and how. The issues involved relate to the policy provisions, their enforcement, and the issue of redress. These implicate all online content including information and disinformation. Curatorial responses within the internet companies are primarily addressed via their policies, which we analyse in this chapter. These responses often result in technical or algorithmic measures, which are covered in depth in Chapter 6.2 Technical/Algorithmic responses. These responses also involve normative and ethical elements, which are addressed in chapter 7.1. News organisations, journalists and other publishers of public interest information also respond curatorially to the problem of disinformation. Such functions can include reporting based on collaborative fact-checking, editorial curation of knowledge and resources, collaborative fact-checking partnerships, curation of sources and resources, audience curation (e.g. User Generated Content), and comment moderation. Chapters 4.1, 4.2 and 5.3 deal with editorial curation efforts associated with fact-checking and investigative reporting. Ethical and normative issues associated with editorial curation are addressed in chapter 7.1, and training initiatives related to curation of disinformation within media institutions are addressed in Chapter 7.2 which deals with educational responses. Below, the terms of service, community guidelines and editorial policies of 11 internet communications companies (Facebook, Instagram 181 , WhatsApp 182 , Google, YouTube 183 , Twitter, VK, Weibo, WeChat, LINE and Snapchat) are examined to gain an in-depth understanding of how these companies expressly or indirectly address the problem of disinformation. These actions tend to seek to curb manipulative actors, deceptive behaviours, and what is perceived to be potentially harmful content (François, 2019). Attention is paid to how decisions on content curation/moderation are made, whether/ how users or third parties are enlisted to help with content monitoring, and which appeal mechanisms are available. Actions undertaken by these companies may be efficient and dynamic, but questions are also raised by various actors regarding the regulatory purview granted through this process to private commercial actors. Concerns about the somewhat random application of self-regulatory measures-for example, emphasising responses in the U.S. environment while abrogating responsibilities in high risk countries in the Global South (Ingram, 2018
This chapter describes the methodology used to develop and release a browser extension which has ... more This chapter describes the methodology used to develop and release a browser extension which has become one of the major tools to debunk disinformation and verify videos and images, in a period of less than 18 months. It has attracted more than 12,000 users from media newsrooms, fact-checkers, the media literacy community, human rights defenders, and emergency response workers dealing with false rumors and content.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this p... more The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.
To review the copyright scope of the reuse for journalistic purposes of User Generated Videos, us... more To review the copyright scope of the reuse for journalistic purposes of User Generated Videos, usually found in social media, the starting point is the analysis of current practices in the news industry. Based on this analysis, we provide a set of recommendations for social media reuse under copyright law and social networks' terms of use. Moreover, we describe how these recommendations have been used to guide the development of the InVID Rights Management module, focusing on EU copyright law given the context of the project and the involved partners.
De l’école primaire à l’université, Wikipédia est désormais un acteur incontournable de l’univers... more De l’école primaire à l’université, Wikipédia est désormais un acteur incontournable de l’univers des connaissances. Mais pour s’en emparer, il faut préalablement comprendre cet outil d’un abord complexe et pourtant moins révolutionnaire qu’il n’y parait. À l’occasion des 20 ans de Wikipédia, chercheurs et enseignants ont réuni leurs contributions dans ce second volume qui fait suite à Wikipédia, objet scientifique non identifié, publié en 2015. Cet ouvrage a pour vocation de constituer un guide à destination des enseignants, des chercheurs et des professionnels souhaitant mener à bien des initiatives autour de l’encyclopédie libre : recherches, projets collaboratifs, projets culturels, etc., ou désirant simplement contribuer eux-mêmes à Wikipédia et aux autres projets de l’univers Wikimédia pour participer à leur amélioration
Companion Proceedings of The 2019 World Wide Web Conference, 2019
News agencies produce thousands of multimedia stories describing events happening in the world th... more News agencies produce thousands of multimedia stories describing events happening in the world that are either scheduled such as sports competitions, political summits and elections, or breaking events such as military conflicts, terrorist attacks, natural disasters, etc. When writing up those stories, journalists refer to contextual background and to compare with past similar events. However, searching for precise facts described in stories is hard. In this paper, we propose a general method that leverages the Wikidata knowledge base to produce semantic annotations of news articles. Next, we describe a semantic search engine that supports both keyword based search in news articles and structured data search providing filters for properties belonging to specific event schemas that are automatically inferred.
Proceedings of the Conference for Truth and Trust Online 2019, 2019
This paper presents an overview of the WeVerify H2020 EU project, which is developing intelligent... more This paper presents an overview of the WeVerify H2020 EU project, which is developing intelligent human-in-the-loop content verification and disinformation analysis methods. Social media and web content are analysed and contextualised within the broader online ecosystem, in order to expose fabricated content, through cross-modal content verification, social network analysis, microtargeted debunking and a blockchain-based public database of known fakes. In particular, we introduce the following already developed tools: cross-modal content verification; blockchain-based database of known fakes; an open source content verification browser plugin; a collaborative verification workbench.
Nous présentons une expérience appliquant l'ingénierie dirigée par les modèles à l'analyse et à l... more Nous présentons une expérience appliquant l'ingénierie dirigée par les modèles à l'analyse et à la mise en oeuvre d'une chaîne complète de gestion de dépêches d'agence traitant de manifestations dans les domaines de la culture et des loisirs. La prise en compte d'événements, structurellement complexes, porteurs d'information spatiale et surtout temporelle, associée à la nécessité de gérer l'interopérabilité d'applications hétérogènes, justifient pleinement le recours à l'IDM. Nous donnons une vue d'ensemble de notre démarche et présentons les métamodèles utilisés, notamment le modèle d'événements servant de pivot entre l'interface avec les applications de traitement de la langue naturelle en amont et les environnements de visualisation d'information temporelle du côté de l'utilisateur final. ABSTRACT. This paper presents an application of Model Driven Engineering to the design of a complete news processing line within the scope of culture and leisure information. MDE is the only technology actually enabling to master the management of complex composite events with spatial and temporal properties, and to cope with making heterogeneous applications interoperate. We provide an overview of our approach and outline the salient points of our metamodels. The stress is put on the event metamodel, standing as a pivot between the beforehand Natural Language Processing, and temporal information visualization frameworks on the end user side.
Numerique, reseaux et mobiles ont bouscule irreversiblement la production, la diffusion et la con... more Numerique, reseaux et mobiles ont bouscule irreversiblement la production, la diffusion et la consommation des media d'actualite. L'Observatoire Transmedia est une plateforme de description, d'unification et d'analyse des actualites diffusees par la radio, la television, le web et l'AFP, sur la periode mi 2011/fin 2013. La collaboration des chercheurs en informatique, SHS et professionnels de l'information a permis la conception, la validation et l'amelioration de la plateforme. La realisation d'etudes -analyse de propagation, chronologies d'evenements, ou encore analyse des taux de reprise des depeches AFP- nous a permis de repondre partiellement aux questions initiales du projet: Comment l'actualite aborde t'elle un sujet ? Qui produit l'information ? La multiplication des supports garantit-elle la diversite?
Cet article aborde le phénomène de l'incomplétude des ressources lexicales, c'est-à-dire la probl... more Cet article aborde le phénomène de l'incomplétude des ressources lexicales, c'est-à-dire la problématique des inconnus, dans un contexte de traitement automatique. Nous proposons tout d'abord une définition opérationnelle de la notion d'inconnu. Nous décrivons ensuite une typologie des différentes classes d'inconnus, motivée par des considérations linguistiques et applicatives ainsi que par l'annotation des inconnus d'un petit corpus selon notre typologie. Cette typologie sera mise en oeuvre et validée par l'annotation d'un corpus important de l'Agence France-Presse dans le cadre du projet EDyLex.
The InVID TV Fake Video Corpus is a small collection of verified fake videos. It was developed in... more The InVID TV Fake Video Corpus is a small collection of verified fake videos. It was developed in the context of the InVID project with the aim of gaining a perspective of the types of fake video that can be encountered in the real world. Currently the Corpus consists of 59 videos. For each video, information is provided describing the fake, its original source, and the evidence proving it is a fake. As we do not own the videos, the dataset only provides the video URLs and metadata, in the form of a tab-separated value (TSV) file. See the README file for more information.
A comprehensive study prepared for the Broadband Commission for Sustainable Development (UNESCO a... more A comprehensive study prepared for the Broadband Commission for Sustainable Development (UNESCO and ITU). 'Balancing Act: Countering Digital Disinformation while respecting Freedom of Expression' surveys issues with a global scope, includes an action-oriented suite of sector-specific actionable recommendations, and presents a 23-point framework to test disinformation responses. Targeted analyses and recommendations to address the life cycle of online disinformation from production to transmission, reception and reproduction focus on: *Legislators and policy makers (counter-disinformation campaigns, electoral-specific responses, the Freedom of Expression Assessment Framework) (Ch5, Ch8, Executive Summary) *Internet companies, producers and distributors (content curation, technical and algorithmic, advertisement policy, demonetisation responses) (Ch6, Executive Summary) *Journalists, investigative researchers and fact checkers (Ch4, Ch3, Ch2, Executive Summary) *Universities and applied and empirical researchers (Ch3, Executive Summary) Target audiences (educational, ethical and normative, empowerment and credibility labeling responses) [Ch7, Executive Summary) The findings are organized into a typology of 11 different categories of responses to disinformation – ranging from identification and investigatory responses, through to policy and legislative measures, technological steps, and educational approaches. For each category of response, the report includes a description of work being done around the world, by which actors, how it is funded and who or what is targeted. The report further analyses the underlying assumptions and theories of change behind these responses, while weighing the challenges and opportunities. Each category of response is also assessed in terms of its intersections with the universal human right of freedom of expression, with a particular focus on press freedom and access to information. Case studies of responses to COVID-19 disinformation are also presented within each category. At the heart of this report is the need to balance responses to disinformation with respect for freedom of expression. The research shows us that this can be done. This research of the Broadband Commission study is edited by Professor Kalina Bontcheva (University of Sheffield, UK) and Dr. Julie Posetti (International Center for Journalists,U.S.; Centre for Freedom of the Media, University of Sheffield/ Reuters Institute for the Study of Journalism, University of Oxford, UK). The other contributing authors are Denis Teyssou (Agence France Presse, France); Dr. Trisha Meyer (Vrije Universiteit Brussel, Belgium); Sam Gregory (WITNESS, U.S.); Clara Hanot (EU Disinfo Lab, Belgium); and Dr. Diana Maynard (University of Sheffield, UK).
Proceedings of the 1st International Workshop on Multimedia AI against Disinformation
Enabled by recent improvements in generation methodologies, DeepFakes have become mainstream due ... more Enabled by recent improvements in generation methodologies, DeepFakes have become mainstream due to their increasingly better visual quality, the increase in easy-to-use generation tools and the rapid dissemination through social media. This fact poses a severe threat to our societies with the potential to erode social cohesion and influence our democracies. To mitigate the threat, numerous DeepFake detection schemes have been introduced in the literature but very few provide a web service that can be used in the wild. In this paper, we introduce the MeVer DeepFake detection service, a web service detecting deep learning manipulations in images and video. We present the design and implementation of the proposed processing pipeline that involves a model ensemble scheme, and we endow the service with a model card for transparency. Experimental results show that our service performs robustly on the three benchmark datasets while being vulnerable to Adversarial Attacks. Finally, we outline our experience and lessons learned when deploying a research system into production in the hopes that it will be useful to other academic and industry teams. CCS CONCEPTS • Information systems → Multimedia information systems; Web services; Data analytics; • Security and privacy → Human and societal aspects of security and privacy.
2020 IEEE International Conference on Multimedia & Expo Workshops (ICMEW)
This paper presents an overview of the WeVerify H2020 EU project, which develops intelligent huma... more This paper presents an overview of the WeVerify H2020 EU project, which develops intelligent human-in-the-loop content verification and disinformation analysis methods, tools and services. Social media and web content are analysed and contextualised within the broader online ecosystem, in order to expose fabricated content, through cross-modal content verification, social network analysis, micro-targeted debunking, and a blockchain-based public database of known fakes.
The production and spreading of manipulated videos have been on the rise over the past years, and... more The production and spreading of manipulated videos have been on the rise over the past years, and is expected to continue and increase further. Manipulating videos have become easier from a technological perspective, and can be done with freely available tools that require less expert knowledge and fewer resources than in the past. All this poses new challenges for those who aim to tackle the spreading of false, manipulated or misleading video content. This chapter covers many of the aspects raised above. It deals with the motivations of those involved in video verification, showcases respective requirements and highlights the importance and relevance of tackling disinformation on social networks. Furthermore, an overview of the state of the art of available techniques and technologies is provided. The chapter then describes the emergence of new threats like so-called ‘deep fakes’ created with the help of artificial intelligence. Finally, we formulate an empirical typology of false videos spreading online.
The InVID TV Fake Video Corpus was developed in the context of the InVID project with the aim of ... more The InVID TV Fake Video Corpus was developed in the context of the InVID project with the aim of gaining a perspective of the types of fake video that can be encountered in the real world. The dataset does not aspire to serve as an exhaustive list of all forgeries that have circulated the Web in the past, but we intend to maintain and extend it throughout the course of the project as new cases arise. The collection is a collaborative effort between AFP and CERTH-ITI. This is the second version of the dataset, containing 117 fake videos and 110 real videos, alongside annotations and descriptions. As we do not own the rights to the videos, the dataset only contains the video URLs and annotations.
Abstract. This paper presents the approach of the Glocal European funded project towards the co-p... more Abstract. This paper presents the approach of the Glocal European funded project towards the co-production of news on World events, on an information marketplace involving both amateurs and professionals. It discuss the rise of user-generated content amongst worldwide media and how event modelling and technology usage may help to foster this pro-am collaboration. Glocal: Event-based Retrieval of Networked Media [1] is a large scale integrating and collaborative project which started in December 2009 for three years and involves ten partners.
This chapter discusses measures to tackle disinformation through content curation or moderation w... more This chapter discusses measures to tackle disinformation through content curation or moderation within internet communication companies and journalism processes. The effect of such measures affects inter alia what content is allowed on the service; if it is allowed to remain up, if it is fact-checked; its prominence and visibility; whether advertising appears next to it; the degree to which it is automatically recommended or limited in terms of distribution; whether it is labelled, as well as what kinds of paid content appear and how. The issues involved relate to the policy provisions, their enforcement, and the issue of redress. These implicate all online content including information and disinformation. Curatorial responses within the internet companies are primarily addressed via their policies, which we analyse in this chapter. These responses often result in technical or algorithmic measures, which are covered in depth in Chapter 6.2 Technical/Algorithmic responses. These responses also involve normative and ethical elements, which are addressed in chapter 7.1. News organisations, journalists and other publishers of public interest information also respond curatorially to the problem of disinformation. Such functions can include reporting based on collaborative fact-checking, editorial curation of knowledge and resources, collaborative fact-checking partnerships, curation of sources and resources, audience curation (e.g. User Generated Content), and comment moderation. Chapters 4.1, 4.2 and 5.3 deal with editorial curation efforts associated with fact-checking and investigative reporting. Ethical and normative issues associated with editorial curation are addressed in chapter 7.1, and training initiatives related to curation of disinformation within media institutions are addressed in Chapter 7.2 which deals with educational responses. Below, the terms of service, community guidelines and editorial policies of 11 internet communications companies (Facebook, Instagram 181 , WhatsApp 182 , Google, YouTube 183 , Twitter, VK, Weibo, WeChat, LINE and Snapchat) are examined to gain an in-depth understanding of how these companies expressly or indirectly address the problem of disinformation. These actions tend to seek to curb manipulative actors, deceptive behaviours, and what is perceived to be potentially harmful content (François, 2019). Attention is paid to how decisions on content curation/moderation are made, whether/ how users or third parties are enlisted to help with content monitoring, and which appeal mechanisms are available. Actions undertaken by these companies may be efficient and dynamic, but questions are also raised by various actors regarding the regulatory purview granted through this process to private commercial actors. Concerns about the somewhat random application of self-regulatory measures-for example, emphasising responses in the U.S. environment while abrogating responsibilities in high risk countries in the Global South (Ingram, 2018
This chapter describes the methodology used to develop and release a browser extension which has ... more This chapter describes the methodology used to develop and release a browser extension which has become one of the major tools to debunk disinformation and verify videos and images, in a period of less than 18 months. It has attracted more than 12,000 users from media newsrooms, fact-checkers, the media literacy community, human rights defenders, and emergency response workers dealing with false rumors and content.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this p... more The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.
To review the copyright scope of the reuse for journalistic purposes of User Generated Videos, us... more To review the copyright scope of the reuse for journalistic purposes of User Generated Videos, usually found in social media, the starting point is the analysis of current practices in the news industry. Based on this analysis, we provide a set of recommendations for social media reuse under copyright law and social networks' terms of use. Moreover, we describe how these recommendations have been used to guide the development of the InVID Rights Management module, focusing on EU copyright law given the context of the project and the involved partners.
De l’école primaire à l’université, Wikipédia est désormais un acteur incontournable de l’univers... more De l’école primaire à l’université, Wikipédia est désormais un acteur incontournable de l’univers des connaissances. Mais pour s’en emparer, il faut préalablement comprendre cet outil d’un abord complexe et pourtant moins révolutionnaire qu’il n’y parait. À l’occasion des 20 ans de Wikipédia, chercheurs et enseignants ont réuni leurs contributions dans ce second volume qui fait suite à Wikipédia, objet scientifique non identifié, publié en 2015. Cet ouvrage a pour vocation de constituer un guide à destination des enseignants, des chercheurs et des professionnels souhaitant mener à bien des initiatives autour de l’encyclopédie libre : recherches, projets collaboratifs, projets culturels, etc., ou désirant simplement contribuer eux-mêmes à Wikipédia et aux autres projets de l’univers Wikimédia pour participer à leur amélioration
Companion Proceedings of The 2019 World Wide Web Conference, 2019
News agencies produce thousands of multimedia stories describing events happening in the world th... more News agencies produce thousands of multimedia stories describing events happening in the world that are either scheduled such as sports competitions, political summits and elections, or breaking events such as military conflicts, terrorist attacks, natural disasters, etc. When writing up those stories, journalists refer to contextual background and to compare with past similar events. However, searching for precise facts described in stories is hard. In this paper, we propose a general method that leverages the Wikidata knowledge base to produce semantic annotations of news articles. Next, we describe a semantic search engine that supports both keyword based search in news articles and structured data search providing filters for properties belonging to specific event schemas that are automatically inferred.
Proceedings of the Conference for Truth and Trust Online 2019, 2019
This paper presents an overview of the WeVerify H2020 EU project, which is developing intelligent... more This paper presents an overview of the WeVerify H2020 EU project, which is developing intelligent human-in-the-loop content verification and disinformation analysis methods. Social media and web content are analysed and contextualised within the broader online ecosystem, in order to expose fabricated content, through cross-modal content verification, social network analysis, microtargeted debunking and a blockchain-based public database of known fakes. In particular, we introduce the following already developed tools: cross-modal content verification; blockchain-based database of known fakes; an open source content verification browser plugin; a collaborative verification workbench.
Nous présentons une expérience appliquant l'ingénierie dirigée par les modèles à l'analyse et à l... more Nous présentons une expérience appliquant l'ingénierie dirigée par les modèles à l'analyse et à la mise en oeuvre d'une chaîne complète de gestion de dépêches d'agence traitant de manifestations dans les domaines de la culture et des loisirs. La prise en compte d'événements, structurellement complexes, porteurs d'information spatiale et surtout temporelle, associée à la nécessité de gérer l'interopérabilité d'applications hétérogènes, justifient pleinement le recours à l'IDM. Nous donnons une vue d'ensemble de notre démarche et présentons les métamodèles utilisés, notamment le modèle d'événements servant de pivot entre l'interface avec les applications de traitement de la langue naturelle en amont et les environnements de visualisation d'information temporelle du côté de l'utilisateur final. ABSTRACT. This paper presents an application of Model Driven Engineering to the design of a complete news processing line within the scope of culture and leisure information. MDE is the only technology actually enabling to master the management of complex composite events with spatial and temporal properties, and to cope with making heterogeneous applications interoperate. We provide an overview of our approach and outline the salient points of our metamodels. The stress is put on the event metamodel, standing as a pivot between the beforehand Natural Language Processing, and temporal information visualization frameworks on the end user side.
Numerique, reseaux et mobiles ont bouscule irreversiblement la production, la diffusion et la con... more Numerique, reseaux et mobiles ont bouscule irreversiblement la production, la diffusion et la consommation des media d'actualite. L'Observatoire Transmedia est une plateforme de description, d'unification et d'analyse des actualites diffusees par la radio, la television, le web et l'AFP, sur la periode mi 2011/fin 2013. La collaboration des chercheurs en informatique, SHS et professionnels de l'information a permis la conception, la validation et l'amelioration de la plateforme. La realisation d'etudes -analyse de propagation, chronologies d'evenements, ou encore analyse des taux de reprise des depeches AFP- nous a permis de repondre partiellement aux questions initiales du projet: Comment l'actualite aborde t'elle un sujet ? Qui produit l'information ? La multiplication des supports garantit-elle la diversite?
Cet article aborde le phénomène de l'incomplétude des ressources lexicales, c'est-à-dire la probl... more Cet article aborde le phénomène de l'incomplétude des ressources lexicales, c'est-à-dire la problématique des inconnus, dans un contexte de traitement automatique. Nous proposons tout d'abord une définition opérationnelle de la notion d'inconnu. Nous décrivons ensuite une typologie des différentes classes d'inconnus, motivée par des considérations linguistiques et applicatives ainsi que par l'annotation des inconnus d'un petit corpus selon notre typologie. Cette typologie sera mise en oeuvre et validée par l'annotation d'un corpus important de l'Agence France-Presse dans le cadre du projet EDyLex.
Uploads
Papers by Denis Teyssou