/fingerprint/man/feat2bin.Rd
Unknown | 42 lines | 38 code | 4 blank | 0 comment | 0 complexity | 1674c7179912857a1fc76dd81392d361 MD5 | raw file
- \name{featvec.to.binaryfp}
- \alias{featvec.to.binaryfp}
- \title{
- Convert a Set of Feature Fingerprints to Binary Fingerprints
- }
- \description{
- Most feature vector style fingerprints (such as circular fingerprints) are
- of variable length, with the features being pulled from an implicit (and
- very large) universe of features.
- While it is possible to convert a single feature fingerprint to a binary vector
- via a hashing procedure, one cannot convert the feature representation into 1:1
- binary representation.
- However, for a collection of feature fingerprints, one can define a "local" universe
- as the union of the features contained in the set of fingerprints. With this definition
- each feature can be mapped to a single bit and thus each fingerprint can be converted
- to a fixed length, keyed fingerprint.
- Since circular fingerprints can lead to a very large binary fingerprint, the initial
- fingerprint is usually folded multiple times to achieve a desired bit density.
- Since the fingerprint folding procedure divides a fingerprint into two equal halves and
- OR's the two halves together, the starting fingerprint must be of even length. For circular
- fingerprints that may not lead to an even-length initial bit string, an extra bit is added
- at the top of the bit string.
- }
- \usage{
- featvec.to.binaryfp(fps, bit.length = 256)
- }
- \arguments{
- \item{fps}{
- A list of \code{featvec} objects
- }
- \item{bit.length}{The length of the bit string to stop folding at}
- }
- \value{
- A list of objects of class \code{fingerprint}
- }
- \seealso{\code{\link{fold}}}
- \keyword{logic}
- \author{Rajarshi Guha \email{rajarshi.guha@gmail.com}}