A few things to fix in getpilotref:

• refs are not drawn randomly
• take in object, not a file
• return objects, with option to return files
• test if n parameter value makes sense
• test if input data frame makes sense

Breakdown splitref into three functions:

1. splitref_two
Split references amongst two collaborators (allowing different proportions for splitting).

• change arguments
• try another way of assigning props

2. splitref_k
Split references amongst multiple collaborators (allowing for only an equal proportion of splits).

3. spiltref_kprop (to do later on)
Split references amongst multiple collaborators (allowing different proportions for splitting).

• Add RIS formatting for export to be used in Rayyan (see code from Patrice below)
• Draft function for fast exclusion labelling along PICO statement

# format for RIS export
deduplicated_refs<- mutate(deduplicated_refs, source_type=ifelse(is.na(source_type)=="TRUE", "JOUR", source_type)) # Add source type where missing
deduplicated_refs <- dplyr::select(deduplicated_refs, source_type, title, author, source, abstract, year, doi, volume, issue, start_page, end_page, language, keywords) 

A few things to fix in splitref function:

• single prop value is ignored and not tested if makes sense for given n
• sum of prop value is not tested if makes sense
• take in object, not a file path
• return objects, with option to return files
• allow any sensible n parameter value
• test if input data frame makes sense