In the field of scientometrics, methods derived from Social Network analysis (SNA) and Natural language processing (NLP) are among common techniques used to analyze and visualize graphs. These methods focus on both the structural and morphological aspects of the social networks investigated, whether or not their actors are localized. SNA and NLP approaches are not specifically interested in the spatial component (i.e. localization, interactions, geovisualization) of social networks. Their complementarity with gravitational approaches, combining analysis of actors' positions and separations (distance, proximity, neighborhood) has nevertheless been widely used in the field of spatial analysis in geography. This workshop aims to present a scientometric co-authorship' analysis on a preselected topic (e.g., low carbon initiatives, climate change, AI in transportation), using Cortext Manager [1] and Arabesque [2]: two web applications respectively mobilized to geocode authors' affiliations addresses at several geographical scales and filtering and exploration spatial networks for thematic mapping purposes. Emphasis will be placed on examining the contributions of different countries or groups of countries to scientific advancements in the selected field and the collaboration patterns that emerge. This hands-on session will guide participants through spatial data analysis and network analysis enabling them to identify thematic and territorial patterns within scientific communities. For doing that, participants will learn how to classify documents by lexical extraction and semantic clustering, or by tagging the textual content of scientific articles with the corresponding Sustainable Development Goals (SDG) categories. This will be followed by the geocoding of authors' affiliation addresses to pinpoint their exact geographic locations. Next, geographic/spatial aggregation methods will be explored, preparing data at different scales, from the address scale to larger meso-level units of analysis, such as: Eurostat’s NUTS3, OECD’s Functional Urban Areas or NETSCITY’s perimeters (most active urban areas in science production [3]). Finally, the resulting co-author’ spatial networks files will be geovisualize in Arabesque, a cartographic tool based on the paradigm of visualization cartography. Several methods of statistical filtering with options adding contextual geographic information or cartographic (re)projections will be applied to the dataset. A particular attention will be paid to the cartographic design of actors' interrelations at different scales through arrows: to their geometry and their semantics, playing on their graphic semiology. Participants will have time to play with the datasets and tools covered in the workshop, with guidance from the trainers and access to a set of materials: a mini-website giving access to all the resources, including datasets, tutorials and examples of results. [1] https://docs.cortext.net/space/ [2] https://arabesque.univ-eiffel.fr/ [3] https://www.irit.fr/netscity/prod/public/intro/ This workshop is part of RETICULAR (RÉseaux, Territoires en Interactions et interrelations Cartographiques), a collaborative research program funded by Université Gustave Eiffel and supported by the LISIS, Aménagement Mobilités Environnement (AME-splott) and Composants et systèmes(COSYS-grettia) départements, with the collaboration of CNRS (Géographie-cités).

Auto-logistic actor attribute models (ALAAMs) are models for analysing social influence or social contagion for cross-sectional networks, when the outcome of interest is dichotomous. If no dependencies among the outcomes of the nodes are assumed, this model reduces to logistic regression. When dependencies through the network, such as social contagion, are assumed, however, the ALAAM provides testable parameters that capture these processes. The workshop will introduce the R package “balaam”, which provides a range of different parameters for network dependencies and estimates the model using Bayesian inference. The package also provides goodness-of-fit analysis, model selection indices, as well as principled approaches for dealing with missing outcomes. Topics treated are: principles of Bayesian inference; model specification; MCMC estimation for the ALAAM; model selection; missing data analysis. Prerequisites The workshop is intended for participants who have working knowledge of quantitative analysis and experience in empirical network research. Fundamental social network analysis skills are assumed. Literature: Koskinen, J. and Daraganova, G. (2022) Bayesian analysis of social influence. Journal of the Royal Statistical Society Series A: Statistics in Society 185.4, pp. 1855–1881. Daraganova, G. & Robins, G. (2013) Autologistic actor attribute model. In: Lusher, D., Koskinen, J. & Robins,G. (Eds.) Exponential random graph models for social networks: theory, methods and applications. New York: Cambridge University Press. pp. 102–114. Daraganova, G. & Pattison, P. (2013) Autologistic actor attribute model analysis of unemployment: dual importance of who you know and where you live. In: Lusher, D., Koskinen, J. & Robins, G. (Eds.) Exponential random graph models for social networks: theory, methods and applications. New York: Cambridge University Press, pp. 237–247. ALAAM website: https://github.com/johankoskinen/ALAAM

This workshop will provide a tutorial of advanced usage of 'ergm' and extension packages, focusing on binary networks. Topics include specifying complex structural constraints, estimation tuning, use of term operators, and observational (e.g., missing data) structure. Also included is using the new 'ergm.multi' package for modelling multilayer and multimode networks, as well as joint models of ensembles of networks. Prerequisites: Familiarity with R and 'ergm' required. If you are new to ERGMs, the introductory workshop on ERGMs using Statnet is strongly suggested.

The netdiffuseR package provides tools for analyzing and simulating diffusion of innovations and contagion processes on networks. In this workshop, we demonstrate the package’s features by analyzing empirical and simulated data on the diffusion of innovations. The session will include examples of using netdiffuseR jointly with other network analysis packages such as RSiena, statnet, and igraph. NetdiffuseR's main features are computing network exposure models based on weight matrices (direct ties, structural equivalence, attribute-weighted, etc.), thresholds, infectiousness and susceptibility. The package works with both static and dynamic networks. Some other capabilities include handling relatively large graphs, simulating networks and diffusion of innovation processes, and visualizing the diffusion of innovations. While there are no prerequisites, it is suggested that you have a working knowledge of the R programming language. While there are no prerequisites, it is suggested that you have a working knowledge of the R programming language. We will use the latest version of the netdiffuseR R package, which can be found on GitHub here: https://github.com/USCCANA/netdiffuseR. During the workshop day, we will provide access to a cloud version of RStudio with the latest version of netdiffuseR, so do not worry if you cannot install the package before the workshop. Duration: 3 hours Max participants: 30