Purpose of addition of this task:
To investigate precision and recall in Inuktitut search using 3 tf-idf
matrices:
(1) no encoding, no grammar (baseline)
(2) yes encoding, no grammar
(3) yes encoding, yes grammar
note:
with/without encoding = with/without transcorder; due to non-standardized
encoding, words with "&" would be cut in the middle of the word >low recall
When reviewing task, please focus on:
We predict that recall increases in (2) compared to (1); precision will
increase in (3)
would these predictions be borne out?
After the review, please
add results (percentages of recall and precision for each matrices)
to InuktituteSearch wiki page.
After the review, the expected next step is to:
combine scripts to tag words (based on roots) and find suffixes
write script to find words and truncate suffixes
Original issue reported on code.google.com by
hisako...@gmail.comon 12 Nov 2011 at 1:21