Module morph

source code

An application of Data-Oriented Parsing to Esperanto. Combines a syntax and a morphology corpus.

Functions

[hide private]

chapelitoj(word)

source code

malchapelitoj(word)

source code

cnf(tree)
make sure all terminals have POS tags; invent one if necessary ("parent_word")

source code

stripfunc(tree)
strip all function labels from a tree with labels of the form "function:form", eg.

source code

dos(words)
`Data-Oriented Segmentation 1': given a sequence of segmented words (ie., a sequence of morphemes), produce a dictionary with extrapolated segmentations (mapping words to sequences of morphemes).

source code

dos1(words)
`Data-Oriented Segmentation 2': given a sequence of segmented words (ie., a sequence of morphemes), produce a dictionary with extrapolated segmentations (mapping words to sequences of morphemes).

source code

dos2(words)

source code

dos3(words)

source code

segmentor(segmentd)
wrap a segmentation dictionary in a naive unknown word segmentation function with some heuristics (phonological rules could probably improve this further)

source code

morphmerge(tree, md, segmented)
merge morphology into phrase structure tree

source code

morphology(train)
an interactive interface to the toy corpus

source code

toy()

source code

interface()

source code

monato()
produce the goodman reduction of the full monato corpus

source code

Variables

[hide private]

__package__ = None

Function Details

[hide private]

stripfunc(tree)

source code

strip all function labels from a tree with labels of the form "function:form", eg. S:np for subject, np.

dos(words)

source code

`Data-Oriented Segmentation 1': given a sequence of segmented words (ie., a sequence of morphemes), produce a dictionary with extrapolated segmentations (mapping words to sequences of morphemes). Assumes non-ambiguity. Method: cartesian product of all possible morphemes at position 0..n, where n is maximum word length.

dos1(words)

source code

`Data-Oriented Segmentation 2': given a sequence of segmented words (ie., a sequence of morphemes), produce a dictionary with extrapolated segmentations (mapping words to sequences of morphemes). Discards ambiguous results. Method: cartesian product of all words with the same number of morphemes.