mjwestgate / starchart

Early-stage package for mapping evidence synthesis in R, and presenting those maps for broader use

starchart

Early-stage package for mapping evidence synthesis in R, and presenting those maps for broader use. Intended to inform development of the 'metaverse' package. Aim is to map the packages and functions available in R for systematic reviews, systematic maps, and meta-analyses.

starchart's Issues

Extract info on external dependencies (software or operating systems) for a set of R packages

It would be handy to know which R packages in our list have either:

1. External dependencies (i.e. software outside R that is required for a package to run), or
2. Dependent on a specific operating system (less common I suspect, but possible)

...because either of these can make it more difficult for the user to install and use them. Conversely, internal dependencies (i.e. those already on CRAN) are not a problem.

Scrape R 'help' files for text mining

Text mining of help files could be a good way to understand the function of individual packages and functions. For example, you could drop help files into a topic model using revtools to classify functions/packages into topics. BUT this first requires a way to import help files to R as raw text. There are at least three ways to do this:

• find the file in your package library and import as text (logical but possibly system-dependent?)
• load help files using help.start() and scrape like a regular website (more circuitous)
• find the repo for that package and download from source (assumes that repo available for all packages)

I don't mind which we choose so long as it works. Maintaining the xml structure of the dataset could be handy as well so that we can restrict text mining to certain parts of the file (e.g., description).