Text Alignment

Local alignment is the process of finding taking two documents and finding the best subset of each document that aligns with one another. A commonly used local alignment algorithm for genetics is the Smith-Waterman algorithm. This package offers a version of the Smith-Waterman algorithm intended to be used for natural language processing.

Consider these two documents. The first is part of Shakespeare's Measure for Measure. The second is a made-up piece of literary criticism quoting the play, but our imaginary literary critic has bungled the quotation. This is a common class of problems (not bungling literary critics but) documents which contain pieces, often heavily modified, from other documents.

shakespeare <- paste(
  "Haste still pays haste, and leisure answers leisure;",
  "Like doth quit like, and MEASURE still FOR MEASURE.",
  "Then, Angelo, thy fault's thus manifested;",
  "Which, though thou wouldst deny, denies thee vantage.",
  "We do condemn thee to the very block",
  "Where Claudio stoop'd to death, and with like haste.",
  "Away with him!")
critic <- paste(
  "The play comes to its culmination where Duke Vincentio, quoting from",
  "the words of the Sermon on the Mount, says,",
  "'Haste still goes very quickly , and leisure answers leisure;",
  "Like doth cancel like, and measure still for measure.'",
  "These titular words sum up the meaning of the play.")

We can uses the local alignment function to extract the part of the text that was borrowed. Notice that the resulting object shows us the changes that have been made.

align_local(shakespeare, critic)

See the documentation for the function to see how to tune the match: ?align_local. This function works with character vectors or with documents of class TextReuseTextDocument.

Try the textreuse package in your browser

Any scripts or data that you put into this service are public.

textreuse documentation built on July 8, 2020, 6:40 p.m.