HURON: HUman Readability ONtologies

Welcome to the online version of HURON, a tool for the evaluation of human readable content of ontologies. Here, you can calculate a set of metrics related with the human readable content of ontologies over a corpus of ontologies. You can also include an analysis of the corpus based on the metrics obtained by the ontologies.

The corpus of ontologies

  • The ontologies must be publicly available in OBO or OWL format.
  • The corpus of ontologies has to be provided as a list of ID URL, where each entry should be in a different line.

The metrics

  • A set of readability metrics is available to be calculated over the ontology corpus. Metrics definitions are here.
  • The process discards ontologies that cannot be processed or that contain unsatisfiable classes.

The analysis of the ontology corpus

  • The analysis of the corpus will be performed if the user checks the corresponding checkbox.
  • The analysis computes, per each metric, its distribution along the ontology corpus. Additionally, a clustering analysis will be performed by obtanining the optimal number of groups per metric, given by evaluome. Finally, the correlation between the metrics will be also included in the analysis. More information about the analysis can be found here.

Output

  • The results of the request will be zipped and sent to the email address specified.
  • The zip file will contain a TSV file indicating the values obtained by each ontology for each metric
  • Additionally, if an analysis of the corpus was requested, the zip will contain a folder with several plots resulting of the analysis:
    • {metricName}_violin.pdf: For each metric, violin plots indicating its global distribution in the ontology corpus.
    • {metricName}_plot.pdf: For each metric, distribution of each cluster according to the optimal number of clusters found by Evaluome. These plots show how many groups of ontologies exist in the corpus according to their values for the corresponding metric, indicating the range for the metric for each group, and the quality of the resulting clustering.
    • metrics_correlation.pdf: Spearman correlation between the metrics, where correlations with a confidence level less than 0.95 are crossed out.
  • If there are no enough data to perform the analysis of the corpus, it will be skipped.

Execution






















--