The files
	varied-localized-all-AllConf.csv
	varied-localized-all-Coherence.csv
	varied-localized-all-Cosine.csv
	varied-localized-all-IR.csv
	varied-localized-all-Kulczynski.csv
	varied-localized-all-MaxConf.csv
contain the evaluation results for the experiments with the QALD dataset for localized rules that were obained with the following settings:
	min_entities_per_class          => 100
        max_entities_per_class          => 10000
        min_anchor_count                => 10
        min_propertyonegram_length      => 4
        min_propertypattern_count       => 5
        min_propertystring_length       => 5
        max_propertystring_length       => 100
.

The first experiment in each file contains the results when all parameters (min_supA, min_supB, min_supAB, min_condBA, min_condAB, threshold) were set to their respective minimum values (i.e., min_supA=5, min_supB=5, min_supAB=5, min_condBA=0, min_condAB=0, threshold=0).
The following lines show the results from experiments where only one of these parameters has a value above its respective minimum value.
The figures 1-3, shown in the paper, are based on the data in the file varied-localized-all-AllConf.csv.

The file qald-questions-combined.yml is in YAML format and contains the (question,query) pairs that we collected from the QALD challenge datasets.

The file qald-questions-details.yml is in YAML format and contains the (question,query) pairs that we collected from the QALD challenge datasets. In addition, it contains for each question the n-grams and for each query it contains the query parts (e.g., the list of terms that appear in subject position etc.).