TRAIN

NAME
SYNOPSIS
DESCRIPTION
XML DOCUMENTS AND THEIR SYNTAX
SEE ALSO
FILES
AUTHOR

NAME

train − train weighted FSA (Finite State Playground)

SYNOPSIS

train xml-file training-data-file

DESCRIPTION

The train program takes an FSA and a set of words and adjusts the weights of arcs accordingly. The weighting system is the simplest possible, with each arc storing the number of times it is traversed. The FSA must be deterministic, so for each training item, the adjustment is unambiguous.

The second file is a text file containing a column of words. The characters making up the words are the same as those accepted by the re program.

XML DOCUMENTS AND THEIR SYNTAX

The XML syntax is quite simple. An FSA is represented as a net object which is composed of a series of fsanode objects, each of which encodes a separate state in the FSA. Whether the state is a start and/or a final state is encoded with attributes.

Arcs between states are represented as arc objects within the fsanode objects, with attributes indicating what the arc symols are and what the destination states are.

The full syntax is encoded in the fsanet.dtd file.

SEE ALSO

fsp(1)
re(1)
concatenate(1)
dictionary(1)
intersect(1)
kleene(1)
union(1)
winnow(1)

http://www.u.arizona.edu/~hammond/

Author’s home page

FILES

flbi/src/

Distribution source code

flbi/xml/fsanet.dtd

DTD file for XML specification of FSAs

flbi/man/train.1

This man file

AUTHOR

Mike Hammond (hammond@u.arizona.edu)

Copyright (C) 2007. All rights reserved.