How to filter meta data by user-defined statements in R?

问题

There is a function called sFilter in R to filter meta data. However, the function is an old (Version: 0.5-10) tm package. Is there any function instead of it in a new version?

My code block is;

query <- "LEWISSPLIT == 'TRAIN'"
trainData <- tm_filter(Corpus, FUN = sFilter, query)

It means, get documents which have "TRAIN" value in their LEWISSPLIT attribute.

<REUTERS TOPICS=?? LEWISSPLIT=?? CGISPLIT=?? OLDID=?? NEWID=??>

回答1:

Just write your own filtering function:

trainData <- tm_filter(Corpus, FUN = function(x, qry) any(meta(x)["lewissplit"] == qry), "TRAIN")

This was adapted from example(tm_filter). There is an example using grep() for more flexible search.

来源：https://stackoverflow.com/questions/34399093/how-to-filter-meta-data-by-user-defined-statements-in-r