pointwise_mi

corpustools.mutualinfo.mutual_information.pointwise_mi(corpus_context, query, env_filtered=False, word_boundary='Word-end only', in_word=False, stop_check=None, call_back=None)

Calculate the mutual information for a bigram.

Parameters:

corpus_context : CorpusContext

Context manager for a corpus

query : tuple

Tuple of two strings, each a segment/letter

env_filtered : bool

True if a env filter selected by the user. Defaults to False

word_boundary : str or bool

How to count word boundaries once per word. str if no env filter selected, bool with env filters ‘Word-end only’ counts once, ‘Both sides’ counts twice (word-initial and word-final), and ‘Ignored’ does not count word boundaries. Trueː env filter selected and # can be a part of a bigram. Defaults to ‘Word-end only’ (count word boundary once in word-final position)

in_word : bool

Flag to calculate non-local, non-ordered mutual information, defaults to False

stop_check : callable or None

Optional function to check whether to gracefully terminate early

call_back : callable or None

Optional function to supply progress information during the function

Returns:

float

Mutual information of the bigram