Bibliometric Analysis: Algorithms & Interpretations

Construct co-authorship graphs from bibliometric data using author full names as the unit of analysis. Link authors across documents from Scopus exports and avoid duplicate nodes when building the graph.

Construct a country-based co-authorship graph by extracting affiliation country names from Scopus data, harmonizing organizations, and illustrating connections between countries such as Canada and Egypt.

Construct a citation graph from a three-article data set, using the unit of analysis as the sources, focusing on source titles and journal connections, with directed versus undirected graphs.

Construct a citation graph of documents from a Scopus-exported data set, linking articles by references and displaying author, year, title, source, and doi in VOSviewer.

Explore how a co-occurrence graph represents author keywords and index keywords as nodes, with links showing co-occurrence, using a dataset and upcoming sections on link thickness and node size.

Explains bibliographic coupling as an undirected graph where articles connect when they share references, and emphasizes data cleaning and harmonization for reliable connections.

Build the co-citation graph by representing each reference as a node and linking nodes that appear together in an article's reference list, treating the graph as undirected with harmonization required.

Explore the mathematical foundations of bibliometric graphs, covering normalization, mapping, node placement in two-dimensional space, and the co-occurrence matrix that explains node size, links, and color-based clusters.

Expose how node size in bibliometric networks reflects document counts and citations within a chosen unit of analysis, using occurrence matrices and degree centrality to assess influence.

Construct an occurrence matrix for authors as the unit of analysis, detailing seven articles’ document-based and citation-based counts and exploring degree centrality in co-authorship networks.

Explore how Vosviewer generates the occurrence matrix from documents and citations in a co-authorship network, and how threshold changes affect weighted degree centrality while leaving document-based matrices unchanged.

Determine the thickness of a link by counting how often pairs of items co-occur. Build a co-occurrence matrix using units of analysis like authors or keywords to measure link strength.

Construct a co-occurrence matrix for co-authorship analysis using an n by n authors matrix, counting how many documents two authors published together, with a symmetric matrix and the diagonal disregarded.

See how VOSviewer uses the co-occurrence matrix to set link thickness between nodes. Increase size variation and scale to highlight weights, with hover showing exact link strengths.

Normalize the co-occurrence and occurrence matrices to derive the similarity (association strength) matrix, enabling 2D placement of authors for bibliometric co-authorship analysis.

Apply the distance-based mapping technique to place nodes on a 2d map based on their similarity and association strength, using optimization concepts like objective function and constraints.

Apply the mapping technique to the symmetric co-occurrence and similarity matrices, formulating the optimization problem and defining decision variables as the coordinates (x, y) of each node.

Learn how to use Excel Solver to minimize an objective function by adjusting node coordinates as non-negative decision variables under constraints, exploring solver algorithms and how initial guesses affect outcomes.

Explore how the Excel solver navigates a constrained solution space to minimize or maximize an objective function, revealing feasible regions, local versus global optima, and the impact of initial solutions.

Explore the mathematical foundation of clustering, focusing on maximizing modularity to group highly connected nodes within the same cluster while minimizing links between clusters in hard clustering.

Apply Excel solver to maximize the modularity function for clustering nodes in a network, using binary decision variables and co-occurrence similarity matrices, then interpret clusters consistent with Vosviewer results.

Explore how degree, closeness, and betweenness centrality analyze bibliometric networks, identify influential nodes, bridges, and quick influencers using a Marvel Cinematic Universe case study.

This lesson presents the MCU dataset for bibliometric analysis, detailing attributes such as author names, movie title, year, director, IMDb rating, and keywords, with IMDb, ChatGPT, and Wikipedia as sources.

Explore co-occurrence analysis to understand bibliometric data using Vosviewer and Scopus data, examining all keywords, threshold settings, and the resulting network with weighted degree centrality.

Clean the co-occurrence data by merging similar keywords and creating a replacement mapping, then export and apply it to reduce 234 keywords to 219.

Explore co-occurrence analysis of networks by adjusting node size, and identify degree, closeness, and betweenness centrality to reveal superhero themes and relationships with the Guardians of the Galaxy.

Explore co-occurrence analysis within a large network to identify clusters around Tony Stark and Iron Man, AI themes, and the billionaire inventor motif, using centrality measures to reveal bridges.

Explore co-occurrence analysis of network clusters to interpret context, identify bridge nodes like scientists and Nick Fury, and map connections among characters such as Captain America, Thor, and Avengers.

Examine co-occurrence analysis of a dataset by tracing clusters of nodes such as Doctor Strange, Spider-Man, and Wakanda, highlighting multiverse, magic, and time travel connections.

Explore co-occurrence analysis of a small network and its clusters using modularity. Apply overlay visualization to trace keyword timelines from 2008 to 2023, noting terms like ant-man and quantum realm.

Organize and update the co-occurrence network from a superhero dataset, highlighting Tony Stark–Iron Man and Thanos, and map Asgard, Loki, Thor, Hulk, Nick Fury, Spider-Man, Captain America connections.

Update co-authorship insights by analyzing country-based networks and thresholds to reveal key locations like Earth, Asgard, and the multiverse, tied to superhero themes.

Perform co-authorship analysis with authors as the unit, cleaning names and including all movies, then analyze degree and betweenness centrality, and apply weighted degree centrality via total link strength.

Identify bridges in the co-authorship network by analyzing clusters and key connecting authors, such as Thor and Loki, revealing their cross-cluster research links.

Identify bridges between network clusters by analyzing inter-cluster connections and node distance, highlighting Scott Lang as a primary bridge and Doctor Strange as a linking node.

Analyze a detached cluster (cluster 11) with an overlay visualization to see if it signals a new Marvel movie type, via a 2021 average publication year, guided by bibliometric analysis.

Explore other network types in bibliometric analysis, including co-authorship, co-occurrence, citation analysis, bibliographic coupling, and co-citation, with guidance on data cleaning, visualization, and interpretation.

Identify visualization caveats in Vosviewer, such as missing links and threshold effects, and emphasize data cleaning and harmonization for reliable bibliometric citation analysis.

Conclude by explaining how to conduct different types of bibliometric analysis and examine the mathematical foundations and objective analysis of bibliometric networks, while encouraging course rating and social media connections.