Hfst documentation

Hfst is a compiler for finite state transducers. The best documentation for writing such transducers is still Beesley and Karttunen (2003): Finite State Morphology. There still are some important differences between Hfst and the compilers described in the B&K book.

The Hfst subprograms

Hfst consists of a large number of smaller programs, with different functions:

• hfst-calculate
  • Compiles SFST files into HFST transducers
• hfst-compare
  • Compares two transducers, checking for equivalence
• hfst-compose
  • Composes two transducers
• hfst-compose-intersect
  • Perform intersecting composition on two transducers (typically the morphotactic transducer/lexicon and the morphophonological transducer)
• hfst-concatenate
  • Concatenates two transducers
• hfst-conjunct
  • Conjuncts two transducers
• hfst-determinize
  • Determinize a transducer, i.e. create an equivalent, epsilon-free transducer that has no state with two or more transitions that have the same input and output symbols.
• [ hfst-diff-test]
• hfst-disjunct
  • Disjuncts two transducers
• [ hfst-duplicate]
• [ hfst-foma-wrapper.sh]
• hfst-format
  • Determine HFST transducer format and print it to output.
• hfst-fst2fst
  • Converts between Hfst, OpenFst, SFST and Foma transducers
• [ hfst-fst2pairstrings]
• hfst-fst2strings
  • Display the string pairs recognized by a transducer, i.e. paths that lead from the initial state to a final state.
• hfst-fst2txt
  • Prints transducers in AT&T tabular format
• hfst-head
  • Get N first transducers from an archive.
• hfst-invert
  • Turn a transducer upside down.
• hfst-lexc
  • Compile a lexc file into a finite-state transducer
• hfst-lexc2fst
• hfst-lookup
  • lookup, gives lemma+analysis of wordforms
• [ hfst-lookup-optimize]
• hfst-minimize
  • Minimize a transducer, i.e. create an equivalent, epsilon-free, deterministic transducer that has as few states as possible.
• hfst-name
• [ hfst-omor-evaluate]
• hfst-pair-test
• [ hfst-preprocess-for-optimized-lookup-format]
• hfst-proc
• hfst-project
• hfst-push-weights
• hfst-regexp2fst
• hfst-remove-epsilons
• hfst-repeat
• hfst-reverse
• hfst-split
• hfst-strings2fst
  • Compiles string-pairs and pair-strings into transducers
• [ hfst-strip-header]
• hfst-substitute
• hfst-subtract
• hfst-summarise
  • Calculates the properties of a transducer
• [ hfst-symbols]
• hfst-tail
• hfst-twolc
  • Compiles a twol (two-level morphophonology) file into an Hfst transducer
• [ hfst-twolc-loc]
• hfst-txt2fst
  • Converts AT&T tabular format into binary transducers
• hfst-xfst
  • Compiles xfst files into Hfst transducers

See also

• Starting a new language with HFST