Please enable JavaScript.
Coggle requires JavaScript to display documents.
Improved fidelity of Semantic Lenses and Answers (Data analysis &…
Improved fidelity of Semantic Lenses and Answers
Answers ranking inside lenses
Corpus preparation routines
Corpus labeling convention development
Corpus labeling: to label answers with S-M-U-B-E-L labels
S label (pure synonyms)
Dicts
Ontology
PPDB
Text2integers (two=2)
M label (answers to merge)
U label (uninformative answers)
Manual labeling:
B label (bad unexpected answer)
L label (long answer)
E label (excellent answer)
Manual check of set label, semi-automated correction
Data download and preprocessing (from SynGFR output to structured corpus)
Model development
Feature design
ML Model launching
Evaluation
Lenses choice and ranking
query category detection
Data analysis & problem formulation
EWB query log analysis
QA lenses analysis for query
GFR query log analysis
Another query logs and data sets (if any)
started
not started
finished