search

massimoaria / bibliometrix

An R-tool for comprehensive science mapping analysis. A package for quantitative research in scientometrics and bibliometrics.
https://www.bibliometrix.org
Other
521 stars 150 forks source link

Coupling analysis results change randomly #259

Closed Xavi-GitHub closed 2 years ago

Xavi-GitHub commented 2 years ago

Hello Either using R Studio or Biblioshiny, every time I repeat the “run” command (R Studio), our the “Apply” button (Biblioshiny), without changing anything (keeping the exact same arguments), the returned coupling map and bibliometric results are different. I get no errors, but the results seem to change randomly. Why is this? How these results can be valid for analysis and interpretation? Thank you Xavi

massimoaria commented 2 years ago

Unfortunately, this happens because igraph team changed the "louvain" algorithm that now returns different results at each iteration. You can take a look at this discussion for more info https://github.com/igraph/rigraph/issues/539

To overcome the issue, we decided to change the default community detection algorithm and now it is set as "walktrap". Moreover, we also added the argument "cluster" to the function couplingMap() so a user can choose the community detection algorithm that he prefers. The same we did for all methods based on the network analysis (thematicMap, thematicEvolution, etc.)

Please, update bibliometrix from GitHub.