Registrations

• Registration

Thursday, 6th may

Michela Bertolotto

Dr. Michela Bertolotto received a BSc and a PhD degree in Computer Science from the University of Genova (Italy). Subsequently she worked at the National Center for Geographic Information and Analysis (NCGIA) and the Department of Spatial Information Science and Engineering of the University of Maine (USA) as a postdoctoral research associate. She is currently a faculty member at the School of Computer Science at University College Dublin (Ireland). Her main research interests include spatio-temporal data modelling and mining, open and crowd-sourced spatial data, LiDAR data management, map personalization. She has published over 120 scientific papers in journals and conferences in geographic information science and related disciplines. She is an associate editor of the International Journal of Geographical Information Science and an editorial board member of the Journal of Spatial Information Science and the ISPRS Journal of GeoInformation.

Title: UrbanARK:  Assessment, Risk management, and Knowledge for flood management in Urban areas

UrbanARK is a collaborative research project between University College Dublin, Queen’s University Belfast, and New York University funded under the SFI US-Ireland Research & Development Programme. UrbanARK brings together leading interdisciplinary research expertise across the areas of civil engineering, social science, geomatics and computer science to support urban coastal communities with regard to the increasing threat of coastal flooding.  
The UrbanArk project aims at enhancing flood risk management for urban coastal communities, and improving their resilience and emergency preparedness. Accurately predicting local flooding in urban areas is highly complex, in part because of the multitude of underground spaces that are often not properly accounted for. Within urban centres, underground spaces such as storage areas, transportation corridors, basement car parks, public facilities, retail & office and private spaces (e.g., residential basements) present a priority risk during flood events with respect to timely evacuation. However, often the location, geometry and volume of these underground spaces are not well known.  Incomplete knowledge poses significant logistical challenges in generating highly accurate maps of these priority risk areas. Furthermore, underground spaces are commonly not considered in urban flood prediction models. At the same time, communicating flood risks and enhancing emergency preparedness poses further challenges. The use of urban underground spaces is highly varied ranging from residential to office space to retail and the perception of risk differs from one user community to the next. Thus emergency planners need engaging communication tools to increase community resilience and preparedness to flooding events.
UrbanARK explores the use of high-resolution airborne and mobile ground-based LiDAR scanning to identify and survey high-risk underground spaces. The LiDAR data is then used to refine flood prediction models and to develop immersive Virtual Reality applications as a communication tool to support communities and emergency planners. The presentation will introduce the UrbanARK project and illustrate the advancement in LiDAR data collection and analysis. It will introduce the multitude of remote sensing datasets collected for UrbanARK together with the underlying strategic data acquisition techniques. It will follow with a discussion on the requirement for scalability in management and analysis of increasingly large volumes of high-resolution spatial data (such as LiDAR and other types of data) for interdisciplinary research.

Françoise Gourmelon

Françoise Gourmelon is CNRS research director in Environmental Geography. She is based at the European University Institute of the Sea (Plouzané, 29) from where she heads the UMR Littoral, Environment, Remote Sensing, Geomatics, a multi-site unit in the West of France.

Title: Opening up geographic data: new contexts, new research challenges

By comparing her scientific background and policies for opening up research data, she will present the past and current initiatives of two operational research structures, a laboratory and an observatory of the sciences of the universe, to which she participated. Opening up research data is an international issue which the environmental research domain is currently taking the measure of, whereas other actors (State, communities) are more advanced in the organization of data dissemination via general or thematic geographic data infrastructures. But this organization, motivated by the INSPIRE directive in Europe since the 2000s, remains fragmented and, in many cases, unsuited to the multiple uses to which these socio-technical systems intend to respond. In order to understand their functioning and promote their operationality, a conceptual framework is proposed. Using the empirical results of the GEOBS research project [1], it sets up a methodology borrowed from the analysis of socio-ecosystems and information commons.

Friday, 7 mai

Antoine Doucet

Antoine Doucet is a university professor at the L3i laboratory at La Rochelle University. He holds a doctorate from the University of Helsinki (Finland) obtained in 2005 and an accreditation to supervise research since 2012. Head of the "Images and Content" research team at L3i, he is also director of the STIC department of the University of Science and Technology of Hanoi (USTH), and coordinator of the H2020 NewsEye project which aims to facilitate access to the old European press as well as its cross-lingual analysis. His research is at the intersection of information retrieval, automatic language processing, text data mining and artificial intelligence. In this context, his work aims to develop methods that can scale and use as few external resources as possible, in order to be particularly applicable to any language of drafting and to any type of document, from the online press to social networks, from the native digital document to the digitized document.

Title:Cross-lingual extraction of geographic information from digitized corpora, particularly historical ones

Many documents can only be made accessible for automatic analysis in the form of digitized images. This is particularly the case for any historical or handwritten document, but also for many native digital documents, passed in the form of an image for various reasons (for example: file conversion or passage through the analog to insert a handwritten signature, send by post, etc.).
Being able to analyze the textual content of such digitized documents requires a phase of conversion from the captured image to a textual representation, a key part of which is optical character recognition (OCR). The resulting text is often imperfect, to an extent which is notably correlated with the quality of the initial medium (which may be stained, folded, aged, etc.) and with the quality of the image taken from it.
This invited conference will present recent advances in AI and automatic language processing enabling this type of corpus to be analyzed in a robust way against OCR errors. For example, I will show how we were able, within the framework of the NewsEye project, to create the state of the art in cross-lingual recognition and disambiguation of named entities (names of places, but also of people, and of organizations) in corpora old press releases written in 4 languages ¿¿between 1850 and 1950, despite particularly degraded corpus.
This type of result opens the way to a large-scale geospatial analysis, which can in particular overcome (linguistic) borders.