Script for evaluating DBpedia entity types using hierarchical f-measure, precision and recall
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
LICENSE
README.md
computeHmeasures.py

README.md

Hierarchical evaluation measures

Implementation of hierarchical F-measure (hF), hierarchical precision (hP) and hierarchical recall (hR) and exact precision. The script was developed for evaluation of type quality in DBpedia.

#Example execution

python computeHmeasures.py "en.lhd.core.2014.nt" "dbpedia_2014.owl" "gs3-toDBpedia2014.nt" "en.lhd.core.gs3.log"

Input files - Gold standard datasets

The datasets are described in our JWS paper

Additional details can be found here http://ner.vse.cz/datasets/linkedhypernyms/

Output

reading gs
reading predicted
finished reading input datasets
total instances in groundtruth:1033.0
total instances in intersection of groundtruth and prediction:402.0
hP:0.864357864358
hR:0.370553665326
hF:0.518726997186
Precision (exact):0.654228855721