feat(doc_network): update

doc_network/bso.bib

@@ -105,3 +105,33 @@ @misc{jeangirard:hal-04598201
   HAL_ID = {hal-04598201},
   HAL_VERSION = {v1},
 }
+
+@article{10.1371/journal.pone.0098679,
+    doi = {10.1371/journal.pone.0098679},
+    author = {Jacomy, Mathieu AND Venturini, Tommaso AND Heymann, Sebastien AND Bastian, Mathieu},
+    journal = {PLOS ONE},
+    publisher = {Public Library of Science},
+    title = {ForceAtlas2, a Continuous Graph Layout Algorithm for Handy Network Visualization Designed for the Gephi Software},
+    year = {2014},
+    month = {06},
+    volume = {9},
+    url = {https://doi.org/10.1371/journal.pone.0098679},
+    pages = {1-12},
+    abstract = {Gephi is a network visualization software used in various disciplines (social network analysis, biology, genomics…). One of its key features is the ability to display the spatialization process, aiming at transforming the network into a map, and ForceAtlas2 is its default layout algorithm. The latter is developed by the Gephi team as an all-around solution to Gephi users’ typical networks (scale-free, 10 to 10,000 nodes). We present here for the first time its functioning and settings. ForceAtlas2 is a force-directed layout close to other algorithms used for network spatialization. We do not claim a theoretical advance but an attempt to integrate different techniques such as the Barnes Hut simulation, degree-dependent repulsive force, and local and global adaptive temperatures. It is designed for the Gephi user experience (it is a continuous algorithm), and we explain which constraints it implies. The algorithm benefits from much feedback and is developed in order to provide many possibilities through its settings. We lay out its complete functioning for the users who need a precise understanding of its behaviour, from the formulas to graphic illustration of the result. We propose a benchmark for our compromise between performance and quality. We also explain why we integrated its various features and discuss our design choices.},
+    number = {6},
+}
+
+@article{Blondel_2008,
+doi = {10.1088/1742-5468/2008/10/P10008},
+url = {https://dx.doi.org/10.1088/1742-5468/2008/10/P10008},
+year = {2008},
+month = {oct},
+publisher = {},
+volume = {2008},
+number = {10},
+pages = {P10008},
+author = {Blondel, Vincent D and Guillaume, Jean-Loup and Lambiotte, Renaud and Lefebvre, Etienne},
+title = {Fast unfolding of communities in large networks},
+journal = {Journal of Statistical Mechanics: Theory and Experiment},
+abstract = {We propose a simple method to extract the community structure of large networks. Our method is a heuristic method that is based on modularity optimization. It is shown to outperform all other known community detection methods in terms of computation time. Moreover, the quality of the communities detected is very good, as measured by the so-called modularity. This is shown first by identifying language communities in a Belgian mobile phone network of 2 million customers and by analysing a web graph of 118 million nodes and more than one billion links. The accuracy of our algorithm is also verified on ad hoc modular networks.}
+}

doc_network/bso.md

@@ -99,15 +99,42 @@ In practice, a PID is also stored (the wikidata for topics, for example) to disa
                 },
 ```
 
-## 2.3 VOSviewer implementation
+## 2.3 Network creation
 
-We use the open source VOSviewer online tool for network visualization [https://github.com/neesjanvaneck/VOSviewer-Online](https://github.com/neesjanvaneck/VOSviewer-Online). It is based on the VOSviewer tool which is very popular for network analysis in bibliometric studies [@DBLP:journals/corr/abs-1006-1032].
+The network creation process involves several key steps: transforming Elasticsearch results into a graph using Graphology, filtering the network to focus on the most interesting nodes, applying spatialization algorithms for visualization, and detecting communities within the network. Below, we detail each of these steps.
 
-In graph theory, a community corresponds to a set of nodes in a graph that are strongly interconnected with each other, while being less connected with nodes outside this community. Communities can be identified in order to understand the underlying structure and patterns of the graph, as well as to analyze the relationships and interactions between the entities that make it up.
-To identify communities, we use the Louvain method. This algorithm works by optimizing a modularity measure that evaluates the strength of communities in a graph. More precisely, Louvain seeks to maximize modularity by progressively moving the nodes of a graph into different communities, in an iterative fashion.
-At each stage, he merges neighboring communities if this leads to an improvement in the overall modularity of the graph. This iterative process continues until no further moves can increase modularity.
-Clusters are computed with the Louvain algorithm, from the open source javascript library graphology-communities-louvain. 
+The network creation process begins with the results obtained from Elasticsearch, utilizing the open-source JavaScript library Graphology [https://github.com/graphology/graphology](https://github.com/graphology/graphology) to construct and manipulate the network. Each link result from Elasticsearch is transformed into nodes and edges, with the edge strength corresponding to the interaction intensity derived from the Elasticsearch aggregations.
 
+To ensure that the network remains manageable and focuses on the most interesting nodes, we employ a strategy that prioritizes the best-connected nodes rather than the largest nodes. By default, the maximum number of nodes is set to 300. This threshold helps in maintaining the computational efficiency and interpretability of the network.
+
+In graph theory, a component refers to a subgraph in which any two nodes are connected to each other by paths, and which is connected to no additional nodes in the larger graph. Using Graphology, we filter the network components by iteratively removing the smallest components until the number of nodes falls below the threshold or only one component remains. This largest component is then subjected to further filtering if it still exceeds the node threshold. In this second filtering step, we utilize the betweenness centrality metric to retain the best-connected nodes. Betweenness centrality measures the extent to which a node lies on the shortest path between other nodes, thereby identifying nodes that act as bridges within the network.
+
+Once the filtering process is complete, we apply a spatialization algorithm to position the nodes in a 2D space. For this purpose, we use the ForceAtlas2 algorithm, which is designed to produce aesthetically pleasing and informative layouts by simulating a physical system where nodes repel each other and edges act as springs pulling connected nodes together. This results in a clear and intuitive visual representation of the network [@10.1371/journal.pone.0098679].  
+Thanks to Graphology the settings of the ForceAtlas2 algorithm are automatically infered from our network order (number of nodes) as below:
+```
+barnesHutOptimize: order > 2000,
+strongGravityMode: true,
+gravity: 0.05,
+scalingRatio: 10,
+slowDown: 1 + Math.log(order)
+```
+
+In graph theory, a community corresponds to a set of nodes in a graph that are strongly interconnected with each other, while being less connected with nodes outside this community. Communities can be identified in order to understand the underlying structure and patterns of the graph, as well as to analyze the relationships and interactions between the entities that make it up. To identify and visualize communities within the network, we apply the Louvain algorithm using Graphology. This algorithm works by optimizing a modularity measure that evaluates the strength of communities in a graph [@Blondel_2008]. More precisely, Louvain seeks to maximize modularity by progressively moving the nodes of a graph into different communities, in an iterative fashion. At each stage, he merges neighboring communities if this leads to an improvement in the overall modularity of the graph. This iterative process continues until no further moves can increase modularity.  
+This step helps in revealing the underlying structure and communities within the scientific network, providing valuable insights into the interactions and collaborations within the bibliometric data.
+
+## 2.4 VOSviewer implementation
+
+To display the network within our application, we use the open source VOSviewer online tool for network visualization [https://github.com/neesjanvaneck/VOSviewer-Online](https://github.com/neesjanvaneck/VOSviewer-Online). It is based on the VOSviewer software which is very popular for network analysis in bibliometric studies [@DBLP:journals/corr/abs-1006-1032].
+
+VOSviewer accepts JSON files formatted according to a specific template [https://app.vosviewer.com/docs/file-types/json-file-type](https://app.vosviewer.com/docs/file-types/json-file-type). This template includes essential attributes for nodes and edges, such as the node ID, name, position, and additional metadata. To ensure compatibility, we transform our Graphology object into a JSON file that adheres to VOSviewer's required format.
+
+Once the JSON file is generated, VOSviewer renders the network, displaying nodes and edges in an interactive and visually appealing manner. The nodes are colorized based on the communities identified through the clustering process performed using the Louvain algorithm. This colorization helps in visually distinguishing different communities within the network, making it easier to analyze and interpret the underlying structure and interactions.
+
+VOSviewer includes its own spatialization algorithm and parameters for layout customization. However, after testing these options, we found them to be visually less intuitive and informative. Consequently, we chose to use the ForceAtlas2 algorithm for spatialization, as described in the previous section, which offers a more aesthetically pleasing and informative layout by being automatically set for our network.
+
+![Visualization of a network with VOSviewer.  
+*(a) Using ForceAltlas2 spatialization  
+(b) Using VOSviewer default spatialization*](https://raw.githubusercontent.com/dataesr/scanr-ui/refs/heads/staging/doc_network/images/vosviewer-spatialization-comparison.jpg)
 
 # 3. Making insightful maps
 

doc_network/mapping_at_scale.tex

@@ -81,6 +81,19 @@
 \newcommand{\VariableTok}[1]{\textcolor[rgb]{0.10,0.09,0.49}{#1}}
 \newcommand{\VerbatimStringTok}[1]{\textcolor[rgb]{0.25,0.44,0.63}{#1}}
 \newcommand{\WarningTok}[1]{\textcolor[rgb]{0.38,0.63,0.69}{\textbf{\textit{#1}}}}
+\usepackage{graphicx}
+\makeatletter
+\def\maxwidth{\ifdim\Gin@nat@width>\linewidth\linewidth\else\Gin@nat@width\fi}
+\def\maxheight{\ifdim\Gin@nat@height>\textheight\textheight\else\Gin@nat@height\fi}
+\makeatother
+% Scale images if necessary, so that they will not overflow the page
+% margins by default, and it is still possible to overwrite the defaults
+% using explicit options in \includegraphics[width, height, ...]{}
+\setkeys{Gin}{width=\maxwidth,height=\maxheight,keepaspectratio}
+% Set default figure placement to htbp
+\makeatletter
+\def\fps@figure{htbp}
+\makeatother
 \setlength{\emergencystretch}{3em} % prevent overfull lines
 \providecommand{\tightlist}{%
   \setlength{\itemsep}{0pt}\setlength{\parskip}{0pt}}
@@ -331,30 +344,117 @@ \subsection{2.2 Elasticsearch
 \end{Highlighting}
 \end{Shaded}
 
-\hypertarget{vosviewer-implementation}{%
-\subsection{2.3 VOSviewer
-implementation}\label{vosviewer-implementation}}
-
-We use the open source VOSviewer online tool for network visualization
-\url{https://github.com/neesjanvaneck/VOSviewer-Online}. It is based on
-the VOSviewer tool which is very popular for network analysis in
-bibliometric studies (Waltman, Eck, and Noyons 2010).
+\hypertarget{network-creation}{%
+\subsection{2.3 Network creation}\label{network-creation}}
+
+The network creation process involves several key steps: transforming
+Elasticsearch results into a graph using Graphology, filtering the
+network to focus on the most interesting nodes, applying spatialization
+algorithms for visualization, and detecting communities within the
+network. Below, we detail each of these steps.
+
+The network creation process begins with the results obtained from
+Elasticsearch, utilizing the open-source JavaScript library Graphology
+\url{https://github.com/graphology/graphology} to construct and
+manipulate the network. Each link result from Elasticsearch is
+transformed into nodes and edges, with the edge strength corresponding
+to the interaction intensity derived from the Elasticsearch
+aggregations.
+
+To ensure that the network remains manageable and focuses on the most
+interesting nodes, we employ a strategy that prioritizes the
+best-connected nodes rather than the largest nodes. By default, the
+maximum number of nodes is set to 300. This threshold helps in
+maintaining the computational efficiency and interpretability of the
+network.
+
+In graph theory, a component refers to a subgraph in which any two nodes
+are connected to each other by paths, and which is connected to no
+additional nodes in the larger graph. Using Graphology, we filter the
+network components by iteratively removing the smallest components until
+the number of nodes falls below the threshold or only one component
+remains. This largest component is then subjected to further filtering
+if it still exceeds the node threshold. In this second filtering step,
+we utilize the betweenness centrality metric to retain the
+best-connected nodes. Betweenness centrality measures the extent to
+which a node lies on the shortest path between other nodes, thereby
+identifying nodes that act as bridges within the network.
+
+Once the filtering process is complete, we apply a spatialization
+algorithm to position the nodes in a 2D space. For this purpose, we use
+the ForceAtlas2 algorithm, which is designed to produce aesthetically
+pleasing and informative layouts by simulating a physical system where
+nodes repel each other and edges act as springs pulling connected nodes
+together. This results in a clear and intuitive visual representation of
+the network (Jacomy 2014).\\
+Thanks to Graphology the settings of the ForceAtlas2 algorithm are
+automatically infered from our network order (number of nodes) as below:
+
+\begin{verbatim}
+barnesHutOptimize: order > 2000,
+strongGravityMode: true,
+gravity: 0.05,
+scalingRatio: 10,
+slowDown: 1 + Math.log(order)
+\end{verbatim}
 
 In graph theory, a community corresponds to a set of nodes in a graph
 that are strongly interconnected with each other, while being less
 connected with nodes outside this community. Communities can be
 identified in order to understand the underlying structure and patterns
 of the graph, as well as to analyze the relationships and interactions
-between the entities that make it up. To identify communities, we use
-the Louvain method. This algorithm works by optimizing a modularity
-measure that evaluates the strength of communities in a graph. More
-precisely, Louvain seeks to maximize modularity by progressively moving
-the nodes of a graph into different communities, in an iterative
+between the entities that make it up. To identify and visualize
+communities within the network, we apply the Louvain algorithm using
+Graphology. This algorithm works by optimizing a modularity measure that
+evaluates the strength of communities in a graph (Blondel et al. 2008).
+More precisely, Louvain seeks to maximize modularity by progressively
+moving the nodes of a graph into different communities, in an iterative
 fashion. At each stage, he merges neighboring communities if this leads
 to an improvement in the overall modularity of the graph. This iterative
-process continues until no further moves can increase modularity.
-Clusters are computed with the Louvain algorithm, from the open source
-javascript library graphology-communities-louvain.
+process continues until no further moves can increase modularity.\\
+This step helps in revealing the underlying structure and communities
+within the scientific network, providing valuable insights into the
+interactions and collaborations within the bibliometric data.
+
+\hypertarget{vosviewer-implementation}{%
+\subsection{2.4 VOSviewer
+implementation}\label{vosviewer-implementation}}
+
+To display the network within our application, we use the open source
+VOSviewer online tool for network visualization
+\url{https://github.com/neesjanvaneck/VOSviewer-Online}. It is based on
+the VOSviewer software which is very popular for network analysis in
+bibliometric studies (Waltman, Eck, and Noyons 2010).
+
+VOSviewer accepts JSON files formatted according to a specific template
+\url{https://app.vosviewer.com/docs/file-types/json-file-type}. This
+template includes essential attributes for nodes and edges, such as the
+node ID, name, position, and additional metadata. To ensure
+compatibility, we transform our Graphology object into a JSON file that
+adheres to VOSviewer's required format.
+
+Once the JSON file is generated, VOSviewer renders the network,
+displaying nodes and edges in an interactive and visually appealing
+manner. The nodes are colorized based on the communities identified
+through the clustering process performed using the Louvain algorithm.
+This colorization helps in visually distinguishing different communities
+within the network, making it easier to analyze and interpret the
+underlying structure and interactions.
+
+VOSviewer includes its own spatialization algorithm and parameters for
+layout customization. However, after testing these options, we found
+them to be visually less intuitive and informative. Consequently, we
+chose to use the ForceAtlas2 algorithm for spatialization, as described
+in the previous section, which offers a more aesthetically pleasing and
+informative layout by being automatically set for our network.
+
+\begin{figure}
+\centering
+\includegraphics{https://raw.githubusercontent.com/dataesr/scanr-ui/refs/heads/staging/doc_network/images/vosviewer-spatialization-comparison.jpg}
+\caption{Visualization of a network with VOSviewer.\\
+\emph{(a) Using ForceAltlas2 spatialization\\
+(b) Using VOSviewer default spatialization}}
+\end{figure}
 
 \hypertarget{making-insightful-maps}{%
 \section{3. Making insightful maps}\label{making-insightful-maps}}
@@ -432,6 +532,12 @@ \section*{References}\label{references}}
 Stefanie Haustein. 2024. ``An Analysis of the Suitability of Openalex
 for Bibliometric Analyses.'' \url{https://arxiv.org/abs/2404.17663}.
 
+\leavevmode\hypertarget{ref-Blondel_2008}{}%
+Blondel, Vincent D, Jean-Loup Guillaume, Renaud Lambiotte, and Etienne
+Lefebvre. 2008. ``Fast Unfolding of Communities in Large Networks.''
+\emph{Journal of Statistical Mechanics: Theory and Experiment} 2008
+(10): P10008. \url{https://doi.org/10.1088/1742-5468/2008/10/P10008}.
+
 \leavevmode\hypertarget{ref-10.1162ux2fqss_a_00179}{}%
 Chaignon, Lauranne, and Daniel Egret. 2022. ``Identifying Scientific
 Publications Countrywide and Measuring Their Open Access: The Case of
@@ -443,6 +549,12 @@ \section*{References}\label{references}}
 Entity Recognition and Disambiguation Service.'' \emph{Journal of the
 Japanese Association for Digital Humanities} 5 (1): 22--60.
 
+\leavevmode\hypertarget{ref-10.1371ux2fjournal.pone.0098679}{}%
+Jacomy, Tommaso AND Heymann, Mathieu AND Venturini. 2014. ``ForceAtlas2,
+a Continuous Graph Layout Algorithm for Handy Network Visualization
+Designed for the Gephi Software.'' \emph{PLOS ONE} 9 (6): 1--12.
+\url{https://doi.org/10.1371/journal.pone.0098679}.
+
 \leavevmode\hypertarget{ref-jeangirard:hal-04813230}{}%
 Jeangirard, Eric. 2024. ``scanR - Explore public data on French research
 and innovation.'' In \emph{euroCRIS SMM 2024}. Paris, France: euroCRIS.