Measure

Contents

1. Overview
2. Types of metrics
3. Output formats

Overview

ROBOT can compute a number of metrics about your ontology, such as entity and axiom counts, qualitative information such as OWL 2 profiles and more complex metrics aimed at informing ontology developers such as logical expressivity and axiom shape. A number of essential metrics can be generated very simply as follows:

robot measure --input uberon_fragment.owl --output results/metrics.tsv

This will generate a table with metrics such as:

• Entity metrics (number of classes, object properties etc.)
• Datatypes used
• Number of axioms (logical and otherwise)
• TBox, RBox, ABox size (size as number of axioms)
• OWL2 profile information

Types of metrics

There are three general modes for computing metrics:

1. Essential metrics (--metrics essential, default)
2. Extended metrics (--metrics extended)
3. All metrics (--metrics all)
4. Reasoner metrics (--metrics all-reasoner, --metrics essential-reasoner, --metrics extended-reasoner)

Essential metrics include basic ontology entity counts (number of classes, individuals and properties), axiom counts and a few qualitiative metrics such as OWL profiles, see the initial example above.

robot measure --input uberon_fragment.owl --format json --metrics essential --output results/metrics_essential.json

Extended metrics contain all the essential metrics, and additional details on axiom types, logical expressivity and datatypes. Use: --metrics extended.

All metrics include all the essential and extended metrics, as well as a range of more complex metrics targeted at OWL and reasoning specialists, such as information about GCIs, shape of the class hierarchy and potential cyclicity. Use: --metrics all.

Reasoner metrics (currently only satisfiability status and number of unsatisfiable classes) can be collected by appending -reasoner to the --metrics parameter, for example: --metrics essential-reasoner.

robot measure --input uberon_fragment.owl --format json --metrics essential-reasoner --output results/metrics_essential_reasoner.json

A full breakdown of all metrics can be found in the table below. In the actual output, you will also see the same metrics ending with _incl. This indicates that the metric is the ontology including its imports. Otherwise, the metric is just representative of the ontology without imports.

Some abbreviations useful to understand the following:

• abox: The part of the ontology that is about individuals, see Wikipedia
• rbox: The part of the ontology that is about object properties
• tbox: The part of the ontology that is about classes, see Wikipedia
• avg: average
• max: maximum
• rhs: ride-hand-side. If used in conjunction with axioms, usually the “superclass” part of a SubClassOf axiom.
• dt: datatype
• gci: General concept inclusion. Formally, all SubClassOf axioms are GCIs, but here we mean those that have a complex left-hand-side (subclass part).
• OWL 2 Profiles:
  • OWL 2 DL: Roughly the subset of OWL that conforms to description logics, plus annotations.
  • OWL 2 EL: Subset of OWL 2 DL, see W3C.
  • OWL 2 QL: Subset of OWL 2 DL, see W3C.
  • OWL 2 RL: Subset of OWL 2 DL, see W3C.
• Logical expressivity and constructs. A breakdown of the meaning of the letters can be found here.

metric	metric_value	metric_type	explanation
abox_axiom_count	0	single_value	Number of axioms in the ABox
annotation_property_count	24	single_value	Number of annotation properties
assert_n_subclass_avg	1.0	single_value	Average number of (asserted) subclasses per class
assert_n_superclass_avg	1.0416666666666667	single_value	Average number of (asserted) superclasses per class
ax_complexrhs_count	15	single_value	Number of axioms with a complex right hand side
axiom_count	703	single_value	Number of axioms
axiom_length_max	4	single_value	Longest axiom in terms of number of entities used (including duplicate uses).
axiom_type_count	SubClassOf 37	map_value	Number of axioms of a specific type, such as SubClassOf
axiom_types	AnnotationAssertion	list_value	Indicates the presence of an axiom type in the ontology.
certain_cycle	false	single_value	If true, there is a definite cycle in the ontology. If false, cyclicity is unknown (experimental).
class_count	24	single_value	Number of classes in the ontology
class_sgl_subcl_count	10	single_value	Number of super-classes which have more than one subclass.
constructs	AL	list_value	Logical constructs used
dataproperty_count	0	single_value	Number of distinct data properties.
datatypes_builtin	XSD_BOOLEAN	list_value	Datatypes used from the built-in datatype map (like XSD schema datatypes).
datatypes_count	2	single_value	Total number of distinct datatypes.
dt_builtin_count	2	single_value	Total number of distinct built-in datatypes.
dt_notbuiltin_count	0	single_value	Total number of distinct custom (not built-in) datatypes.
expressivity	ALEH+	single_value	Logical expressivity.
gci_count	0	single_value	Total count of GCI axioms.
gci_hidden_count	2	single_value	Total count of hidden GCIs.
individual_count	0	single_value	Number of individuals.
instance_per_class_avg	0.0	single_value	Average number of individuals per class.
namespace_axiom_count	http://purl.obolibrary.org/obo/BFO_ 17	map_value	Number of axioms using at least one term in a namespace.
namespace_entity_count	http://purl.obolibrary.org/obo/BFO_ 17	map_value	Number of distinct entities used from a namespace.
logical_axiom_count	41	single_value	Number of logical axioms in ontology
most_freq_concept	UBERON:0000475	single_value	Most frequently used class.
multiple_inheritance_count	2	single_value	Number of classes with multiple inheritance.
named_supercl_count_max	1	single_value	Maximum number of named superclasses.
obj_property_count	2	single_value	Number of object properties.
ontology_anno_count	0	single_value	Number of ontology annotations.
ontology_id	https://github.com/ontodev/robot/examples/edit.owl	single_value	The ontology id.
owl2_dl	false	single_value	Does the ontology correspond to the OWL2 DL profile?
owl2_el	false	single_value	Does the ontology correspond to the OWL2 EL profile?
owl2_ql	false	single_value	Does the ontology correspond to the OWL2 QL profile?
owl2_rl	false	single_value	Does the ontology correspond to the OWL2 RL profile?
owl2	true	single_value	Does the ontology correspond to the OWL2 Full profile?
owl2dl_profile_violation	UseOfUndeclaredAnnotationProperty 531	map_value	Counts of OWL2 DL profile violations.
rbox_axiom_count	2	single_value	How many axioms in the RBOX?
rdfs	false	single_value	Does the ontology correspond to the RDFS profile?
rule_count	0	single_value	Number of SWRL rules used.
signature_entity_count	52	single_value	Total number of entities in signature, including classes and individuals.
syntax	RDF/XML Syntax	single_value	What serialisation is used for the ontology?
tautology_count	0	single_value	Number of tautological axioms.
tbox_axiom_count	39	single_value	Number of TBOX axioms (without RBOX).
tbox_nominals	false	single_value	Number of TBOX axioms with nominals.
tboxrbox_axiom_count	41	single_value	Number of TBOX axioms (with RBOX).
undecl_entity_count	17	single_value	Number of undeclared entities.

Output formats

Currently, the following serialisation formats are supported:

• tsv
• csv
• html
• json
• yaml

To serialise the metrics as json, for example, we run:

robot measure --input uberon_fragment.owl --format json --metrics all --output results/metrics_all.json

Note that if no --format/-f is explicitly provided, ROBOT will try to infer the inteded format from the file extension instead. Failing that, it will generate a TSV file.

Error Messages

Metrics format error

Only the following --format/-f options are currently supported: tsv, csv, json, yaml, html.

Metrics type error

Only the following --metrics/-m options are currently supported: essential, essential-reasoner, extended, extended-reasoner, all, all-reasoner.