Minor Changes

  • Added detection for numeric node attributes and addition to node size control.
  • Added base vertex size controls. Base vertex size gives greater control over starting node size independent from normalization.
  • Enabled vertex size for visNetwork graph.

Bug Fixes:

  • Fixed a problem with categorical attributes having multiple underscores in their names.

Minor Changes

  • Updated standard package documentation, added citation and README.Rmd.

Bug Fixes:

  • Fixed the issue with pruning un-selected failing.

Minor Changes:

  • Added neighborhood select feature to enable selection of ego networks from graphs.
  • Added node labels selection and checkbox for displaying labels for selected nodes only.
  • visNetwork plot now displays multiple edges.

Minor Changes:

  • Added a legend to network plots with vertex categorical attributes.
  • Removed automatic iconv conversion of text for text analysis. So unicode should display properly for plots in most cases.
  • Made Arial Unicode MS the default plot font for macOS systems that have it installed. This is to avoid problems with unicode characters. It can be disabled in the Network Graphs section.
  • Added an option for HTML decoding text.
  • Made removal of URL’s, numbers and punctuation optional for text analysis.
  • Added text analysis word length control.
  • Added some additional wordcloud controls such as re-seed.
  • Cleaned up text corpus, word frequency and text analysis plot functions.
  • Improved speed of sentiment analysis plots. They will also now only update when the Sentiment Analysis tab is open or clicked on.
  • Added convenience functions dtbGraph() and eawGraph(). These load package included “Divided They Blog” and “Enviro Activist Websites 2006” igraph graphs.
  • Cleaned up package dependencies.
  • CRAN version update.

Minor Changes:

  • Set visNetwork plot edge width if weight column present in edges.
  • Removed arrows from visNetwork plot if graph undirected.
  • Fixed a groups parameter warning from dplyr::summarise.
  • Cleaned up package dependencies.

Bug Fixes:

  • Fixed an issue with custom classes assigned to dataframes causing an vctrs error when using dplyr functions. The classes are not required so they are simply removed.

Minor Changes:

  • Removed code support for older vosonSML versions prior to 0.29.
  • Code refactoring and removal of networkD3 plots.
  • Added new hex sticker to package documentation.

Minor Changes:

  • Added the option to inherit node colors from loaded graphml files in plots.
  • Clicking nodes in the visnetwork plot now also selects or deselects them in the data table below it.
  • Added a more simple node label option of node index which is the numeric index of the vertex.
  • Added the niter or “number of iterations” input field for FR and GraphOpt plots.
  • Added the charge, mass, spring.length and spring.constant input fields for GraphOpt plots.


Refer to igraph documentation for further information on layout parameters at https://igraph.org/r/doc/layout_with_fr.html and https://igraph.org/r/doc/layout_with_graphopt.html.

Minor Changes:

  • Moved rtweet from package imports to suggests.

Bug Fixes:

  • Fixed reddit URL formatting to support retrieving additional threads.

Minor Changes:

  • Renamed bimodal networks to twomode.
  • Added percentage sliders to twitter semantic network creation tab.

Major Changes:

  • Supports the new vosonSML version 0.29 changes to the network and graph creation process. It is backwards compatable with version 0.27 albeit without the new features.
  • Added a new tab to the Collect Data controls named Create Network. This allows the creation of the different types of networks that are supported by vosonSML.
Twitter:
  • Networks supported are activity, actor, bimodal and semantic.
  • Both activity and actor have an option to Add Text that adds tweet text as a network node or edge attribute.
  • The actor network also has the option to Lookup User Data. This will retrieve profile information for users that became nodes during network creation who were not authors of tweets. Their profile information was missing most likely because they were referenced in tweets but none of their tweets were collected in the search. In a twitter search profile information is only returned for the authors of tweets captured in the search.
  • Both bimodal and semantic networks have the option to filter out terms using the Remove Terms field. This accepts a comma delimited list of terms that can be actors and hashtags (e.g @climate_person, #climate) for bimodal networks and terms, actors and hashtags (e.g climate, @climate_person, #climate) for semantic networks. The semantic network currently uses the default vosonSML options of only including the 5% most frequent terms and 50% most frequently occuring hashtags in the network.
Youtube:
  • Networks supported are activity and actor.
  • Both activity and actor have an option to Add Text that adds video comment text as a network node or edge attribute.
  • The actor network with Add Text option allows for the further option to Find Replies in Text. This will create reply edges between actors in the network when an @actor_name reference is found at the beginning of a text comment.
  • There is also the Add Video Details option for actor networks that retrieves video information and adds it to the network as a node attributes of VideoID type nodes (e.g title, description and publisher). This also replaces VideoID type nodes with the actors ID who published the video.
  • A further option for Add Video Details is Only replace Video ID's. This option does not supplement the network with the additional video information retrieved but instead only replaces the VideoID type nodes with the actors ID who published the video.
Reddit:
  • Networks supported are activity and actor.
  • Both activity and actor have an option to Add Text that adds thread comment text as a network node or edge attribute.


Refer to vosonSML documentation for further information on network types and options at https://vosonlab.github.io/vosonSML/reference/index.html.

Minor Changes:

  • Added a new Collect data download button named Network to allow the nodes and edges dataframes to be downloaded after Create Network. The data is downloaded as an R object in a .rds file that can be loaded into R using the readRDS() function.

Bug Fixes:

  • Fixed the console scrolling in the Collect section of the interface to scroll to the bottom when there is new text output.

Minor Changes:

  • Changed the required package check to be more efficient by calling required() instead of installed.packages().
  • Changed the app startup messages to use message() instead of cat() so if desired they can be suppressed.
  • Added pkgStartupMsgs parameter to runVOSONDash() to hide the app’s required package startup messages.
  • Added optional isLocal parameter to runVOSONDash() to manually set local or server flag if desired. If not set app will detect if it is running in server mode.
  • Disabled the loading and saving of keys and tokens in server mode.
  • Added a tooltip to Create Web Auth Token to add some further information on process and note.

Minor Changes:

  • Changed the node size controls in Network Graphs to be based on normalized continuous values.
  • Moved the calculation of network metrics to the getNetworkMetrics() package function.
  • Demonstration data select box now hidden when another data source is loaded.
  • Abbreviated long name layouts in the Graph layout control.
  • Added tooltips to various interface controls using the popper.js javascript library (v1.14.3). Documentation at https://getbootstrap.com/docs/4.1/components/popovers/ and https://popper.js.org.

Bug Fixes:

  • Fixed problem with visNetwork graphs rendering slightly off the canvas.

Major Changes:

  • VOSONDash Shiny app has been re-structured into an R package. Have started moving some of the ‘business logic’ away from the app server and into the general package.
  • Social Media API keys are no longer saved to the users working directory. API keys and tokens are now saved to files in the users home directory as specified by the system environment variable HOME. This location can be found in the VOSONDash start up information, in the apps API Keys tab or by using the following R function in RStudio: Sys.getenv("HOME"). The keys and tokens files are named vosondash_keys.rds and vosondash_tokens.rds respectively.
  • Twitter no longer creates an API authentication token for each twitter collection. Instead twitter tokens are now created and managed in the API Keys tab and should be saved and re-used.
  • A twitter web authorization token can now be created and used for API access without the need for a Developer account. This still requires a twitter app and Consumer API keys but means if a user has access to these that the user can authorize the twitter app against their account and use VOSONDash to perform twitter collections. Using this token API rate-limiting will be applied to the users account not the app. Twitter refers to this method of authentication as ‘Application-user authentication: OAuth 1a (access token for user context)’. This feature uses the rtweet packages interactive web authorization as implemented by the rtweet::create_token function. As stated above once a token has been created it should be saved and re-used.

Minor Changes:

  • Demonstration data has been included in the package and is accessible by a new selection box below the graphml file loader in Network Graphs. If there is a file with the same name as the demonstration graphml file but with an additional ‘.txt’ extension in the packages demonstration data folder it will be loaded as the graph description.
  • Added graph summary information for Network Graphs plots as an overlay in the bottom left corner.
  • Added a select box in the top right corner of the graph plots that the user can use to select the render plot height in pixels.
  • Changed selected row and vertex highlighting colors in graph data table and plot.
  • Removed visNetwork plot navigation controls.
  • Removed D3 network plots for the time being.
  • NRC sentiment analysis plots have been re-worked and now split into the 8 emotions, measured and sorted as the proportion of the total of the emotions in the text and the valence which is the measure of how positive or negative the text is.
  • The All categories plot has now been removed from text analysis when single categories are chosen or combined. In that particular use-case the plots are the same. It will only be shown when the All value in the sub-categories select box is selected.
  • Removed keys fields from Collect twitter and it now displays selected auth token.
  • Variable naming changes and a lot of code re-structuring and refactoring.