Proceedings of the International AAAI Conference on Web and Social Media
The French presidential election was one of the main political event of 2017, and triggered a lot... more The French presidential election was one of the main political event of 2017, and triggered a lot of activity on Twitter. The campaign was highly unpredictable and led to the rise of 5 main parties instead of the historical bipartite (left-right) confrontation, ranging from far-left to far-right. This dataset paper proposes #Élysée2017fr, a large and complex dataset of 22853 Twitter profiles active during the campaign (from November 2016 to May 2017), and their corresponding tweets and retweets, plus the retweet and mention networks related to these profiles. The profiles were manually annotated with their political affiliations (up to 2 political parties per profile), their nature (individual or collective), and the sex of the profile's owner when available. This is one of the rare datasets that considers a non-binary stance classification and, to our knowledge, the first one with a large number of profiles, and the first one proposing overlapping political communities. This da...
Présentation6ème édition des journées d'étude « Big Data Mining and Visualization » de l'... more Présentation6ème édition des journées d'étude « Big Data Mining and Visualization » de l'association EGC, Université de LilleLe but de cette présentation est de présenter un nouveau modèle de détection de points de vue politique basé sur les communautés d'utilisateurs-trices sur les réseaux sociaux, en utilisant 2 cas d'études de 2014 : les élections législatives aux États-Unis et le référendum pour l'indépendance écossaise
International audienceThe French presidential election was one of the main political event of 201... more International audienceThe French presidential election was one of the main political event of 2017. The campaign was highly unpredictable, with moving allegiances shifting around the five main parties, instead of the historical left-right confrontation. We propose an analysis of Twitters users participation in this campaign, thanks to more than 20000 manually annotated profiles, with a focus on the differences between male and female participation on the platform. The first analyses show that the use of Twitter varies wildly according to the party. A large part of the profiles seem to have been created specifically for the campaign (19% of the profiles were created after September 2016). The majority of the participants are male, with approximately 10000 profiles owned by men and 4000 by women. Women tend to publish less tweets than men, but to retweet more. The main peaks of activity correlate with campaign events, particularly the debates and rounds of the election. La France Inso...
This collection allowing to analyse the spatio-temporel peculiarities of small-scale events on Tw... more This collection allowing to analyse the spatio-temporel peculiarities of small-scale events on Twitter. We release a dataset related to New York City drawn from 2 collections of tweets:<strong>NY2014:</strong> Released by Zhang et al., SIGIR'16 (doi.org/10.1145/2911451.2911519) from 2014.08.01 to 2014.11.30.<strong>NY2016:</strong> This dataset was collected with Twitter API from 2016.10.01 to 2016.11.30. In accordance with Twitter's Terms of Service, we only provide identifiers of tweets If you use this dataset, please cite our paper:Paul Mousset, Yoann Pitarch, Lynda Tamine. 2018. Studying the Spatio-Temporal Dynamics of Small-Scale Events in Twitter. In Proceedings of the 29th ACM Conference on Hypertext and Social Media (HT '18). ACM, Baltimore, MD, USADOI : doi.org/ 10.1145/3209542.3209561
Leveraging textual and spatial data provided in spatiotextual objects (eg., tweets), has become i... more Leveraging textual and spatial data provided in spatiotextual objects (eg., tweets), has become increasingly important in realworld applications, favoured by the increasing rate of their availability these last decades (eg., through smartphones). In this paper, we propose a spatial retrofitting method of word embeddings that could reveal the localised similarity of word pairs as well as the diversity of their localised meanings. Experiments based on the semantic location prediction task show that our method achieves significant improvement over strong baselines.
Proceedings of the Symposium on Applied Computing, 2017
Numereous digital library projects mine heterogeneous data from different sources to provide expe... more Numereous digital library projects mine heterogeneous data from different sources to provide expert finding services. However, a variety of models seek experts as simple sources of information and neglect authority signals. In this paper we address the issue of modelling the authority of researchers in academic networks. A model, RAC, is proposed that merges several graph representations and incorporate external knowledge about the authority of some major scientific conferences to improve the identification of authoritative researchers. Based on the provided structural model a biased label propagation algorithm aimed to strenghten the scores calculation of the labelled entities and their neighbors is developped. Both quantitative and qualitative analyses validate the effectiveness of the proposal. Indeed, RAC outperforms state-of-the-art models on a real-world graph containing more than 5 million nodes constructed using Microsoft Academic Search, AMiner and Core.edu databases.
Les modeles d'entrepots dits classiques (etoile, etc.) ont emerge et connu un vif succes au s... more Les modeles d'entrepots dits classiques (etoile, etc.) ont emerge et connu un vif succes au sein des entreprises de part leur presentation graphique facile a lire. Ceci est necessaire dans un contexte ou la modelisation multidimen- sionnelle doit etre confrontee a l'avis des decideurs. Dans des travaux anterieurs, nous avons mis en avant un certain manque d'expressivite de ces modeles. Par exemple, dans le cas d'un entrepot de donnees medicales, il n'etait pas possible de modeliser le fait que la tension arterielle d'un patient soit " faible ", " normale " ou " elevee " (hierarchisation de la mesure) depend de son âge et du fait qu'il fume ou non. Ayant developpe une formalisation de hierarchies dites " contextuelles " pour pallier a ce probleme, nous proposons dans ce papier un modele graphique, pour faciliter la lisibilite du modele aupres des decideurs, que nous baptisons le " modele avec satellites ".
Résumé. La qualité des contenus sur les plateformes collaboratives est très hétérogène. Dans la l... more Résumé. La qualité des contenus sur les plateformes collaboratives est très hétérogène. Dans la littérature scientifique, les algorithmes d’analyse structurelle appliqués à la tâche de détection de contenu de qualité reposent généralement sur des graphes définis à partir d’un seul type de nœuds et de relations. Pourtant les graphes sur lesquels reposent ces récentes plateformes présentent de nombreuses sémantiques de nœuds et relations différentes, e.g., producteurs/consommateurs, questions/réponses, etc. Ces solutions souffrent d’un manque de généricité et ne peuvent s’adapter facilement à l’évolution des plateformes. Nous proposons une modélisation générique de ces platformes par les graphes hétérogènes pouvant intégrer automatiquement de nouvelles sémantiques de nœuds et de relations. Un algorithme de prédiction de qualité des contenus reposant sur ce modèle est proposé. Nous montrons qu’il généralise plusieurs travaux de la littérature. Enfin, en intégrant certaines relations in...
Proceedings of the 29th on Hypertext and Social Media, 2018
Numerous domains have interests in studying the viewpoints expressed online, be it for marketing,... more Numerous domains have interests in studying the viewpoints expressed online, be it for marketing, cybersecurity, or research purposes with the rise of computational social sciences. Current stance detection models are usually grounded on the specificities of some social platforms. This rigidity is unfortunate since it does not allow the integration of the multitude of signals informing effective stance detection. We propose the SCSD model, or Sequential Community-based Stance Detection model, a semi-supervised ensemble algorithm which considers these signals by modeling them as a multi-layer graph representing proximities between profiles. We use a handful of seed profiles, for whom we know the stance, to classify the rest of the profiles by exploiting like-minded communities. These communities represent profiles close enough to assume they share a similar stance on a given subject. Using datasets from two different social platforms, containing two to five stances, we show that by combining several types of proximity we can achieve excellent results. Moreover, we compare the proximities to find those which convey useful information in term of stance detection.
Numerous feature-based models have been recently proposed by the information retrieval community.... more Numerous feature-based models have been recently proposed by the information retrieval community. The capability of features to express different relevance facets (query- or document-dependent) can explain such a success story. Such models are most of the time supervised, thus requiring a learning phase. To leverage the advantages of feature-based representations of documents, we propose TournaRank, an unsupervised approach inspired by real-life game and sport competition principles. Documents compete against each other in tournaments using features as evidences of relevance. Tournaments are modeled as a sequence of matches, which involve pairs of documents playing in turn their features. Once a tournament is ended, documents are ranked according to their number of won matches during the tournament. This principle is generic since it can be applied to any collection type. It also provides great flexibility since different alternatives can be considered by changing the tournament typ...
The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitori... more The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitoring the Twitter stream and pushing tweets to users according to given profiles. It includes metrics, files, settings and hypothesis provided by the organizers. In this work, we perform a thorough analysis of each component of the framework used in 2016 and 2017 and found some limitations for the Scenario A of this track. Our main findings point out the weakness of the metrics and give clear recommendations to fairly reuse the collection.
The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitori... more The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitoring the Twitter stream and pushing tweets to users according to given profiles. It includes metrics, files, settings and hypothesis provided by the organizers. In this work, we perform a thorough analysis of each component of the framework used for batch evaluation of scenario A in 2016 and 2017. We found some weaknesses of the metrics and took advantage of these limitations to submit our official run in the 2017 edition. The good evaluation results validate our findings. This paper also gives clear recommendations to fairly reuse the collection.
The CL-SCisumm track provides a framework to evaluate systems summarising scientific papers. It i... more The CL-SCisumm track provides a framework to evaluate systems summarising scientific papers. It includes datasets and metrics provided by the organisers. The track comprises three tasks: (1a) identifying the spans of text in the referred document reflecting citing ext spans i.e., citances), (1b) classifying discourse facets ofthe cited text spans, and (2) generating a short structured summary. For the 2019 edi-tion, we focused our work on the task 1a. This report presents our proposed approachfor this task. We submitted 15 runs corresponding to different configurations of theparameters involved in our approach.
This article describes and studies information exchanges on Twitter during the 2017 French presid... more This article describes and studies information exchanges on Twitter during the 2017 French presidential election campaign. Grounded in the theoretical framework of social representations, it investigates the discourses of the support communities of the five main candidates, using methods drawn from network analysis and the statistical analysis of text data. The findings highlight the weight of the psycho-sociological processes that regulate the way social groups relate to their political environment, and the way in which this relationship materializes through the structuring of communication on Twitter. This article thus shows how the ideological background of the different political communities can regulate their perception of certain objects of social representation connected to the presidential campaign.
Proceedings of the 29th on Hypertext and Social Media, 2018
Small-scale events are emerging as attractive objects of research. On Twitter, small-scale events... more Small-scale events are emerging as attractive objects of research. On Twitter, small-scale events represent weak sensors that report things happening in specific times and places. While previous work addressed the issue of detecting such events, very little is known so far about their inherent properties. In this paper, our main objective was to analyse the spatio-temporal peculiarities of small-scale events w.r.t different levels of location granularity, and to understand the general trend of their propagation along their lifetimes. Our findings suggest that (1) users involved in small-scale events mostly gravitate not significantly far from the geographical focus; (2) events do not exhibit major peaks; and (3) there exists distinct events that we can identify from users' posts that significantly differ from topic distribution, focus concentration and propagation distance perspectives across time.
Motivation Metabolomics studies aim at reporting a metabolic signature (list of metabolites) rela... more Motivation Metabolomics studies aim at reporting a metabolic signature (list of metabolites) related to a particular experimental condition. These signatures are instrumental in the identification of biomarkers or classification of individuals, however their biological and physiological interpretation remains a challenge. To support this task, we introduce FORUM: a Knowledge Graph (KG) providing a semantic representation of relations between chemicals and biomedical concepts, built from a federation of life science databases and scientific literature repositories. Results The use of a Semantic Web framework on biological data allows us to apply ontological-based reasoning to infer new relations between entities. We show that these new relations provide different levels of abstraction and could open the path to new hypotheses. We estimate the statistical relevance of each extracted relation, explicit or inferred, using an enrichment analysis, and instantiate them as new knowledge in ...
Biomedical research increasingly relies on computational approaches to extract relevant informati... more Biomedical research increasingly relies on computational approaches to extract relevant information from large corpora of publications.
Proceedings of the ACM SIGIR International Conference on Theory of Information Retrieval, 2017
Stance detection systems often integrate social clues in their algorithms. While the in uence of ... more Stance detection systems often integrate social clues in their algorithms. While the in uence of social groups on stance is known, there is no evaluation of how well state-of-the-art community detection algorithms perform in terms of detecting like-minded communities, i.e. communities that share the same stance on a given subject. We used Twitter's social interactions to compare the results of community detection algorithms on datasets on the Scottish Independence Referendum and US Midterm Elections. Our results show that algorithms relying on information di usion perform better for this task and con rm previous observations about retweets being better vectors of stance than mentions.
Success of Wikipedia is decidedly due to the free availability of high quality articles across ma... more Success of Wikipedia is decidedly due to the free availability of high quality articles across many different expertise areas. If most of these resolute collaborations between authoritative users might constitute referenceable sources, Wikipedia is not sheltered from well-identified problems regarding articles quality, e.g., reputability of third-party sources and vandalism. Because of the huge number of articles and the intensive edit rate, it is not reasonable to even consider the manual evaluation of the content quality of each article. In this paper, we tackle the problem of modeling and predicting the quality of articles in collaborative platforms. We propose a quality model integrating both temporal and structural features captured from the implicit peer review process enabled by Wikipedia. A generic HITS-like framework is developed and able to capture both the quality of the content and the authority of the associated authors. Notably, a mutual reinforcement principle held between articles quality and author’s authority is exploited in order to take advantage of the collaborative graph generated by the users. Experiments conducted on a set of representative data from Wikipedia show the effectiveness of the computed indicators both in an unsupervised and supervised scenario.
Proceedings of the International AAAI Conference on Web and Social Media
The French presidential election was one of the main political event of 2017, and triggered a lot... more The French presidential election was one of the main political event of 2017, and triggered a lot of activity on Twitter. The campaign was highly unpredictable and led to the rise of 5 main parties instead of the historical bipartite (left-right) confrontation, ranging from far-left to far-right. This dataset paper proposes #Élysée2017fr, a large and complex dataset of 22853 Twitter profiles active during the campaign (from November 2016 to May 2017), and their corresponding tweets and retweets, plus the retweet and mention networks related to these profiles. The profiles were manually annotated with their political affiliations (up to 2 political parties per profile), their nature (individual or collective), and the sex of the profile's owner when available. This is one of the rare datasets that considers a non-binary stance classification and, to our knowledge, the first one with a large number of profiles, and the first one proposing overlapping political communities. This da...
Présentation6ème édition des journées d'étude « Big Data Mining and Visualization » de l'... more Présentation6ème édition des journées d'étude « Big Data Mining and Visualization » de l'association EGC, Université de LilleLe but de cette présentation est de présenter un nouveau modèle de détection de points de vue politique basé sur les communautés d'utilisateurs-trices sur les réseaux sociaux, en utilisant 2 cas d'études de 2014 : les élections législatives aux États-Unis et le référendum pour l'indépendance écossaise
International audienceThe French presidential election was one of the main political event of 201... more International audienceThe French presidential election was one of the main political event of 2017. The campaign was highly unpredictable, with moving allegiances shifting around the five main parties, instead of the historical left-right confrontation. We propose an analysis of Twitters users participation in this campaign, thanks to more than 20000 manually annotated profiles, with a focus on the differences between male and female participation on the platform. The first analyses show that the use of Twitter varies wildly according to the party. A large part of the profiles seem to have been created specifically for the campaign (19% of the profiles were created after September 2016). The majority of the participants are male, with approximately 10000 profiles owned by men and 4000 by women. Women tend to publish less tweets than men, but to retweet more. The main peaks of activity correlate with campaign events, particularly the debates and rounds of the election. La France Inso...
This collection allowing to analyse the spatio-temporel peculiarities of small-scale events on Tw... more This collection allowing to analyse the spatio-temporel peculiarities of small-scale events on Twitter. We release a dataset related to New York City drawn from 2 collections of tweets:<strong>NY2014:</strong> Released by Zhang et al., SIGIR'16 (doi.org/10.1145/2911451.2911519) from 2014.08.01 to 2014.11.30.<strong>NY2016:</strong> This dataset was collected with Twitter API from 2016.10.01 to 2016.11.30. In accordance with Twitter's Terms of Service, we only provide identifiers of tweets If you use this dataset, please cite our paper:Paul Mousset, Yoann Pitarch, Lynda Tamine. 2018. Studying the Spatio-Temporal Dynamics of Small-Scale Events in Twitter. In Proceedings of the 29th ACM Conference on Hypertext and Social Media (HT '18). ACM, Baltimore, MD, USADOI : doi.org/ 10.1145/3209542.3209561
Leveraging textual and spatial data provided in spatiotextual objects (eg., tweets), has become i... more Leveraging textual and spatial data provided in spatiotextual objects (eg., tweets), has become increasingly important in realworld applications, favoured by the increasing rate of their availability these last decades (eg., through smartphones). In this paper, we propose a spatial retrofitting method of word embeddings that could reveal the localised similarity of word pairs as well as the diversity of their localised meanings. Experiments based on the semantic location prediction task show that our method achieves significant improvement over strong baselines.
Proceedings of the Symposium on Applied Computing, 2017
Numereous digital library projects mine heterogeneous data from different sources to provide expe... more Numereous digital library projects mine heterogeneous data from different sources to provide expert finding services. However, a variety of models seek experts as simple sources of information and neglect authority signals. In this paper we address the issue of modelling the authority of researchers in academic networks. A model, RAC, is proposed that merges several graph representations and incorporate external knowledge about the authority of some major scientific conferences to improve the identification of authoritative researchers. Based on the provided structural model a biased label propagation algorithm aimed to strenghten the scores calculation of the labelled entities and their neighbors is developped. Both quantitative and qualitative analyses validate the effectiveness of the proposal. Indeed, RAC outperforms state-of-the-art models on a real-world graph containing more than 5 million nodes constructed using Microsoft Academic Search, AMiner and Core.edu databases.
Les modeles d'entrepots dits classiques (etoile, etc.) ont emerge et connu un vif succes au s... more Les modeles d'entrepots dits classiques (etoile, etc.) ont emerge et connu un vif succes au sein des entreprises de part leur presentation graphique facile a lire. Ceci est necessaire dans un contexte ou la modelisation multidimen- sionnelle doit etre confrontee a l'avis des decideurs. Dans des travaux anterieurs, nous avons mis en avant un certain manque d'expressivite de ces modeles. Par exemple, dans le cas d'un entrepot de donnees medicales, il n'etait pas possible de modeliser le fait que la tension arterielle d'un patient soit " faible ", " normale " ou " elevee " (hierarchisation de la mesure) depend de son âge et du fait qu'il fume ou non. Ayant developpe une formalisation de hierarchies dites " contextuelles " pour pallier a ce probleme, nous proposons dans ce papier un modele graphique, pour faciliter la lisibilite du modele aupres des decideurs, que nous baptisons le " modele avec satellites ".
Résumé. La qualité des contenus sur les plateformes collaboratives est très hétérogène. Dans la l... more Résumé. La qualité des contenus sur les plateformes collaboratives est très hétérogène. Dans la littérature scientifique, les algorithmes d’analyse structurelle appliqués à la tâche de détection de contenu de qualité reposent généralement sur des graphes définis à partir d’un seul type de nœuds et de relations. Pourtant les graphes sur lesquels reposent ces récentes plateformes présentent de nombreuses sémantiques de nœuds et relations différentes, e.g., producteurs/consommateurs, questions/réponses, etc. Ces solutions souffrent d’un manque de généricité et ne peuvent s’adapter facilement à l’évolution des plateformes. Nous proposons une modélisation générique de ces platformes par les graphes hétérogènes pouvant intégrer automatiquement de nouvelles sémantiques de nœuds et de relations. Un algorithme de prédiction de qualité des contenus reposant sur ce modèle est proposé. Nous montrons qu’il généralise plusieurs travaux de la littérature. Enfin, en intégrant certaines relations in...
Proceedings of the 29th on Hypertext and Social Media, 2018
Numerous domains have interests in studying the viewpoints expressed online, be it for marketing,... more Numerous domains have interests in studying the viewpoints expressed online, be it for marketing, cybersecurity, or research purposes with the rise of computational social sciences. Current stance detection models are usually grounded on the specificities of some social platforms. This rigidity is unfortunate since it does not allow the integration of the multitude of signals informing effective stance detection. We propose the SCSD model, or Sequential Community-based Stance Detection model, a semi-supervised ensemble algorithm which considers these signals by modeling them as a multi-layer graph representing proximities between profiles. We use a handful of seed profiles, for whom we know the stance, to classify the rest of the profiles by exploiting like-minded communities. These communities represent profiles close enough to assume they share a similar stance on a given subject. Using datasets from two different social platforms, containing two to five stances, we show that by combining several types of proximity we can achieve excellent results. Moreover, we compare the proximities to find those which convey useful information in term of stance detection.
Numerous feature-based models have been recently proposed by the information retrieval community.... more Numerous feature-based models have been recently proposed by the information retrieval community. The capability of features to express different relevance facets (query- or document-dependent) can explain such a success story. Such models are most of the time supervised, thus requiring a learning phase. To leverage the advantages of feature-based representations of documents, we propose TournaRank, an unsupervised approach inspired by real-life game and sport competition principles. Documents compete against each other in tournaments using features as evidences of relevance. Tournaments are modeled as a sequence of matches, which involve pairs of documents playing in turn their features. Once a tournament is ended, documents are ranked according to their number of won matches during the tournament. This principle is generic since it can be applied to any collection type. It also provides great flexibility since different alternatives can be considered by changing the tournament typ...
The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitori... more The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitoring the Twitter stream and pushing tweets to users according to given profiles. It includes metrics, files, settings and hypothesis provided by the organizers. In this work, we perform a thorough analysis of each component of the framework used in 2016 and 2017 and found some limitations for the Scenario A of this track. Our main findings point out the weakness of the metrics and give clear recommendations to fairly reuse the collection.
The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitori... more The TREC Real-Time Summarization (RTS) track provides a framework for evaluating systems monitoring the Twitter stream and pushing tweets to users according to given profiles. It includes metrics, files, settings and hypothesis provided by the organizers. In this work, we perform a thorough analysis of each component of the framework used for batch evaluation of scenario A in 2016 and 2017. We found some weaknesses of the metrics and took advantage of these limitations to submit our official run in the 2017 edition. The good evaluation results validate our findings. This paper also gives clear recommendations to fairly reuse the collection.
The CL-SCisumm track provides a framework to evaluate systems summarising scientific papers. It i... more The CL-SCisumm track provides a framework to evaluate systems summarising scientific papers. It includes datasets and metrics provided by the organisers. The track comprises three tasks: (1a) identifying the spans of text in the referred document reflecting citing ext spans i.e., citances), (1b) classifying discourse facets ofthe cited text spans, and (2) generating a short structured summary. For the 2019 edi-tion, we focused our work on the task 1a. This report presents our proposed approachfor this task. We submitted 15 runs corresponding to different configurations of theparameters involved in our approach.
This article describes and studies information exchanges on Twitter during the 2017 French presid... more This article describes and studies information exchanges on Twitter during the 2017 French presidential election campaign. Grounded in the theoretical framework of social representations, it investigates the discourses of the support communities of the five main candidates, using methods drawn from network analysis and the statistical analysis of text data. The findings highlight the weight of the psycho-sociological processes that regulate the way social groups relate to their political environment, and the way in which this relationship materializes through the structuring of communication on Twitter. This article thus shows how the ideological background of the different political communities can regulate their perception of certain objects of social representation connected to the presidential campaign.
Proceedings of the 29th on Hypertext and Social Media, 2018
Small-scale events are emerging as attractive objects of research. On Twitter, small-scale events... more Small-scale events are emerging as attractive objects of research. On Twitter, small-scale events represent weak sensors that report things happening in specific times and places. While previous work addressed the issue of detecting such events, very little is known so far about their inherent properties. In this paper, our main objective was to analyse the spatio-temporal peculiarities of small-scale events w.r.t different levels of location granularity, and to understand the general trend of their propagation along their lifetimes. Our findings suggest that (1) users involved in small-scale events mostly gravitate not significantly far from the geographical focus; (2) events do not exhibit major peaks; and (3) there exists distinct events that we can identify from users' posts that significantly differ from topic distribution, focus concentration and propagation distance perspectives across time.
Motivation Metabolomics studies aim at reporting a metabolic signature (list of metabolites) rela... more Motivation Metabolomics studies aim at reporting a metabolic signature (list of metabolites) related to a particular experimental condition. These signatures are instrumental in the identification of biomarkers or classification of individuals, however their biological and physiological interpretation remains a challenge. To support this task, we introduce FORUM: a Knowledge Graph (KG) providing a semantic representation of relations between chemicals and biomedical concepts, built from a federation of life science databases and scientific literature repositories. Results The use of a Semantic Web framework on biological data allows us to apply ontological-based reasoning to infer new relations between entities. We show that these new relations provide different levels of abstraction and could open the path to new hypotheses. We estimate the statistical relevance of each extracted relation, explicit or inferred, using an enrichment analysis, and instantiate them as new knowledge in ...
Biomedical research increasingly relies on computational approaches to extract relevant informati... more Biomedical research increasingly relies on computational approaches to extract relevant information from large corpora of publications.
Proceedings of the ACM SIGIR International Conference on Theory of Information Retrieval, 2017
Stance detection systems often integrate social clues in their algorithms. While the in uence of ... more Stance detection systems often integrate social clues in their algorithms. While the in uence of social groups on stance is known, there is no evaluation of how well state-of-the-art community detection algorithms perform in terms of detecting like-minded communities, i.e. communities that share the same stance on a given subject. We used Twitter's social interactions to compare the results of community detection algorithms on datasets on the Scottish Independence Referendum and US Midterm Elections. Our results show that algorithms relying on information di usion perform better for this task and con rm previous observations about retweets being better vectors of stance than mentions.
Success of Wikipedia is decidedly due to the free availability of high quality articles across ma... more Success of Wikipedia is decidedly due to the free availability of high quality articles across many different expertise areas. If most of these resolute collaborations between authoritative users might constitute referenceable sources, Wikipedia is not sheltered from well-identified problems regarding articles quality, e.g., reputability of third-party sources and vandalism. Because of the huge number of articles and the intensive edit rate, it is not reasonable to even consider the manual evaluation of the content quality of each article. In this paper, we tackle the problem of modeling and predicting the quality of articles in collaborative platforms. We propose a quality model integrating both temporal and structural features captured from the implicit peer review process enabled by Wikipedia. A generic HITS-like framework is developed and able to capture both the quality of the content and the authority of the associated authors. Notably, a mutual reinforcement principle held between articles quality and author’s authority is exploited in order to take advantage of the collaborative graph generated by the users. Experiments conducted on a set of representative data from Wikipedia show the effectiveness of the computed indicators both in an unsupervised and supervised scenario.
Uploads
Papers by Yoann PITARCH