Researchers commonly need to summarize scientific information, a process known as 'evidence synthesis'. The first stage of a synthesis process (such as a systematic review or meta-analysis) is to download a list of references from academic search engines such as 'Web of Knowledge' or 'Scopus'. This information can be sorted manually (the traditional approach to systematic review), or the user can draw on tools from machine learning to help them visualise patterns in the corpus. revtools uses topic models to render ordinations of text drawn from article titles, keywords and abstracts, and allows the user to interactively select or exclude individual references, words or topics. revtools does not currently provide tools for analysis of data drawn from those references, features that are available in other packages such as metagear or metafor.
Import & export
read_bibliography Import bibliographic data
write_bibliography Export bibliographic data
Data storage and manipulation
bibliography-class Format for storing bibliographic data
bibliography-methods Print, summary, as.bibliography, as.data.frame and [ methods for class 'bibliography'
review_info-class Format for storing data from start_review_window
review_info-methods summary methods for class review_info
Duplicate detection
find_duplicates Locate potentially duplicated references
extract_unique_references return a data.frame with only 'unique' references
Topic modelling and visualisation
make_DTM Construct a Document-Term Matrix from bibliographic data
run_LDA Wrapper function for topic models
start_review_window Launch a Shiny app for reference sorting