unibz-core / Scior

Identification of ontological categories for OWL ontologies.
https://purl.org/scior
Apache License 2.0
3 stars 1 forks source link

Better statistics calculation #25

Closed pedropaulofb closed 1 year ago

pedropaulofb commented 1 year ago

Currently, we calculate statistics regarding classes and classifications. For classes, the results are grouped into three categories: totally unknown classes, partially known classes, totally known classes.

Considering X the number of classifications available, and K the number of known classifications, the categories are:

As can be seen, classes with a single known classification and classes with a single unknown classification are both registered as PK classes.

Considering this, we could create a new statistics for classes indicating the % of information (number of known classifications/number of classifications available) that this every class have.

By implementing this, it would be possible to discover the average amount of information discovered for classes. E.g., the average known percentage starts in 10% and ends in 90%, then 80% were increased.

pedropaulofb commented 1 year ago

Problem solved with the creation of the knowledge_matrix in commit caeb012d2478c9c376dabc75a9b60ce8a10a91cb.