acqdiv / frequent-frames

Code and data for frequent frames analysis

Frequent frames analysis pipeline

Details

For more information about the ACQDIV project, see: http://www.acqdiv.uzh.ch/. The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement no 615988 (PI Sabine Stoll).

Pipeline

1. Generate the bigrams

Run the script: bigrams.py to generate the bigram frequencies. Note this make take a long time.

2. Generate the trigram counts for each analysis

For words, morphemes, dyads, etc. This generates counts for the frequency calculation of the frames and for operationalization.

Call: python3 utils.py

Make sure to set the words or morphemes in the script!

3. For global recall (DB) get the word types:

Call: sh get_types.sh

4. Run the frames processing and PR analysis

This step requires that the bigrams are calculated and pickled first. Then run the frames analysis with frames.py. The script outputs a TSV file to load into R.

Make sure to set words or morphemes (pos or glosses) in the script!

5. Statistical analysis with R scripts

The frames-functions.R file contains various functions for processing and plotting the frequent frames analyses. The words, morphemes and dyads analyses are in words.R, morpheme-gloss-pos.R and dyads.R.

frequent-frames's Issues

Data files too large

Some of the data files are too large for Git (over 100MB). Time to use:

Git Large File Storage - https://git-lfs.github.com
See http://git.io/iEPt8g for more information

Add publication info