towards self-explanatory ontology visualization with contextual verbalization
TRANSCRIPT
Towards Self-explanatoryOntology Visualization
with Contextual VerbalizationRenārs Liepiņš, Uldis Bojārs, Normunds Gr zītisū̄ ,
Kārlis Čerāns, Edgars Celms
Institute of Mathematics and Computer ScienceUniversity of Latvia
12th International Baltic Conference on Databases and Information Systems (DB&IS), Riga, Latvia | 4-6 July 2016
Assistants are teachers. They are not professors. Professors are teachers who teach mandatory courses. Mandatory courses are taught only by professors. A course may be taught by more than one teacher.Courses are taken by students. Second-year students have passed at least two courses. Teachers may take courses as well but not the ones that they teach.Courses constitute academic programs. Each student is enrolled only in one academic program.
Domain experts vs. Knowledge engineers
ConceptualModelling
OntologyEngineering
<<owlClass>>Student
<<owlClass>>AcademicProgram
<<owlClass>>Course
<<owlClass>>Person
<<owlClass>>Teacher
<<owlClass>>MandatoryCourse
<<owlClass>>OptionalCourse
<<owlClass>>Professor
<<owlClass>>Assistent
<<disjointWith>>
<<objectProperty>>teaches
<<objectProperty>>takes
<<disjointWith>>
+teaches
+includes
+takes
+enrolls
+constitutes
Class: Assistant SubClassOf: Teacher DisjointWith: ProfessorClass: Professor SubClassOf: Teacher, teaches some MandatoryCourse DisjointWith: AssistantClass: MandatoryCourse SubClassOf: Course, inverse (teaches) only ProfessorObjectProperty: teaches Domain: Teacher Range: Course
Diagrams vs. Controlled natural language (CNL)
• Visual notations provide a better overview of the ontology structure and entity connections• However, one has to learn the notation to understand its semantics
• CNL is readable right away and often helps to understand complex ontology restrictions• However, the structure and interconnections become less apparent
Combining diagrams with CNL verbalization
• Every diagram element represents a set of ontology axioms – the axioms are generally presented locally in the diagram• A single ontology axiom can be related to several elements of the diagram
• Ontology symbols are lexically motivated and consistent• Preferably with lexical annotations
• We do not have to design a new verbalizer for each new graphical OWL notation, because they all are mapped to the same underlying axioms• By reusing ontology verbalizers, existing ontology visualization systems can
be easily extended with on-demand contextual explanations
• Contextual verbalization can be helpful in both exploring existing ontologies and authoring new ontologies• It would motivate to follow a lexical and consistent naming convention
Assumptions and consequences
OWLGrEd: a notation and editor for OWL
Developed at IMCS UL, used across the globe
• A compact UML-style graphical notation and editor for OWL 2• Uses the Manchester OWL Syntax for class expressions• Builds on 20-year experience in graphical modeling languages and tools• Integrated with Protégé
• The OWLGrEd/CNL extension:• Supports the lexicalization of ontology elements and the verbalization of
ontology axioms• Tested on the highly analytic English and the highly inflected Latvian
• Uses:• The OWL subset of Attempto Controlled English (ACE) as an interlingua• Grammatical Framework (GF) for the multilingual verbalization
OWLGrEd and its experimental CNL extension
interlingual ACE statements
OWLGrEd and its experimental CNL extension
OWLGrEd: editor
OWLGrEd: viewer
Multilingual domain
lexicon (GF)
Multilingual ontology grammar (GF)
compiled into
generates General, multilingual ACE
grammar (GF)
Ontology
Axioms Annotations
CNL Verbalizer
Attempto OWL
verbalizer
axioms
collected by
ACE statements using ontology
symbols
ACE-English / ACE-Latvian statements
1
2
yields adds
OWL axioms
CNL statements
3
Lexical terms in English and Latvian
Ontology symbol, generated form the English term
Domain lexicon: declaring a class
Properties are always declared / lexicalized in the context of their domain and range classes
In Latvian, the grammatical case of the subjectand object depends on the verb;The user can implicitly correct the case, if theautomatically suggested case is incorrect
Domain lexicon: declaring a property
Contextual verbalization in action (editor)
Contextual verbalization in action (viewer)
Contextual verbalization in action (viewer)
Contextual verbalization in action (viewer)
Contextual verbalization in action (viewer)
Contextual verbalization in action (viewer)
• CNL is more understandable to domain experts and end-users because it does not have to be learned, or the learning time is very short
• However, comparative studies have shown the domain experts still tend to prefer the graphical notations in the long run
• We have proposed an approach that combines the benefits of both a graphical notation and CNL verbalization• A graphical representation gives the user an overview of the ontology
while the contextual verbalization explains what the particular graphical element means
• The interactive CNL explanations help domain experts to learn and use the graphical notation without a special training (documentation)
• Target audience: domain experts, students, end-users• Also knowledge engineers, especially for validating the inferred axioms
Conclusion