--colloc_pmi_thresh

Switch

--colloc_pmi_thresh PMI

Description

The PMI (pointwise mutual information) threshold above which a multigram from --colloc_table is treated as a valid collocation, when extracting features with --use_collocs.

Argument and Default Value

A float. Default: 3.0 (dlac.DEF_PMI, same default as --set_pmi_threshold).

Details

Looks at the feat_colloc_filter column of the collocation table you specify with --colloc_table. Distinct from --set_pmi_threshold, which is used when building a collocation table (--feat_colloc_filter) in the first place, not when consuming one via --use_collocs.

Other Switches

Required Switches:

• --use_collocs

• --colloc_table

Optional Switches:

• --colloc_column

Example Commands

dlatkInterface.py -d dla_tutorial -t msgs -c user_id --use_collocs --colloc_table my_collocs --colloc_pmi_thresh 5.0