sourCEntral - mobile manpages

pdf

PYNLPL-COMPUTEPMI

NAME

computepmi − manual page for pynlpl-computepmi 0.7.7

DESCRIPTION

usage: pynlpl−computepmi [−h] −f INPUTTEXT [−s] [−t THRESHOLD] [−a] [−A]

[−−pmi] [−−npmi] [−−jaccard] [−−dice]

Simple cooccurence computation

optional arguments:
−h
, −−help

show this help message and exit

−f INPUTTEXT, −−inputtext INPUTTEXT

Input file (plaintext, tokenised, utf−8, one sentence per line) (default: )

−s, −−sorted

Output sorted by co−occurrence score (default: False)

−t THRESHOLD, −−threshold THRESHOLD

Joined occurrence threshold, do not consider words occurring less than this (default: 1)

−a, −−adjacency

Compute the adjacency fraction (how many co−occurrence are immediate bigrams) (default: False)

−A, −−discountadjacency

Do not take immediately adjacent fragments (bigrams) into account when computing mutual information (requires −a) (default: False)

−−pmi

Compute pointwise mutual information (default: False)

−−npmi

Compute normalised pointwise mutual information (default: False)

−−jaccard

Compute jaccard similarity coefficient (default: False)

−−dice

Compute dice coefficient (default: False)

pdf