Minimizing conservativity violations in ontology alignments: algorithms and evaluation
read more
Citations
Ontology Based Data Access in Statoil
From Polynomial Procedures to Efficient Reasoning with EL Ontologies
Results of the Ontology Alignment Evaluation Initiative 2019
LogMap family participation in the OAEI 2017
Large-Scale Ontology Matching: State-of-the-Art Analysis
References
Depth-First Search and Linear Graph Algorithms
The Unified Medical Language System (UMLS): integrating biomedical terminology
A theory of diagnosis from first principles
Ontology Matching
Similarity flooding: a versatile graph matching algorithm and its application to schema matching
Related Papers (5)
Frequently Asked Questions (16)
Q2. What future works have the authors mentioned in the paper "Minimizing conservativity violations in ontology alignments: algorithms and evaluation" ?
In order to mitigate incompleteness, the authors plan to study extensions of their techniques to more expressive logical fragments, while keeping the current scalability properties. Nevertheless the authors plan to explore alternative methods to address the conservativity violations. For example, domain experts could be involved in the assessment of the additional disjointness [ 20, 35 ], and to suggest extensions to the input ontologies [ 31 ] for violations recognised as false positives. The authors consider, however, that the proposed methods have also potential in scenarios others than Optique.
Q3. What are the common ontology alignment repair systems?
State-of-the-art ontology alignment repair systems, such as ALCOMO [54], AML [66], ASMOV [32], Lily [85], LogMap [33], and YAM++ [60], typically consider the input ontologies as immutable and their repair techniques focus on the mappings.
Q4. What is the impact of alignment repair?
The impact of alignment repair is computed as the percentual of gain (resp. loss for negative values) for each measure computed for a repaired alignment, compared to the same measure computed for the original alignment.
Q5. What is the principle of conservativity in ontology alignment?
The conservativity principle in ontology alignment aims at capturing the differences in the ontology classification between the input ontologies and the aligned ontology [36] (i.e., new subsumptions and/or new equivalences among concepts).
Q6. What is the simplest way to encode the structural index of ontologies?
Given that queries over the structural relationships of ontologies are heavily employed in their approach, the authors rely on the optimized structural index of LogMap [33, 39], based on the interval labelling schema techniques presented in [1]
Q7. What is the definition of equivalence violations?
In addition, the authors also define violations between concepts that may have been already involved in a subsumption relationship (i.e., resulting in an equivalence between them), denoted as equivalence conservativity principle violations, or simply equivalence violations.
Q8. What are the three principles proposed to minimize the number of potentially unintended consequences?
In [36] three principles were proposed to minimize the number of potentially unintended consequences, namely: (i) consistency principle, the mappings should not lead to unsatisfiable concepts in the integrated ontology, (ii) conservativityReceived xxx Revised xxx Accepted xxxprinciple, the mappings should not introduce new semantic relationships between concepts from one of the input ontologies, (iii) locality principle, the mappings should link entities that have similar neighbourhoods.
Q9. How many entities were considered not equivalent?
In a later release of such ontology, 15 entities were merged, while 18 were judged as not equivalent by domain experts (NCI ontology curators).
Q10. How is the graph representation of the ontology created?
The graph representationG of the aligned ontology w.r.t.O1,O2 andM, is built by means of createDigraph function (line 1 of Algorithm 1).
Q11. What is the sum of the detection and repair time of EqRepair?
The experimental results considering EqRepair algorithm, can be summarized as follows:(i) The sum of the detection and repair time of EqRepair is very low due to the linear cost of the detection technique and the efficient parallelization of the diagnosis computation.(ii)
Q12. What is the corrective strategy for the ontology?
The correction strategy aims at adding to the input ontologies a minimal set of axioms, so that the input ontologies (in isolation) can entail the novel axiom (solving, in this way, the violation).
Q13. What is the definition of a diagnosis for a graph representation of an ontology?
In Definition 4.4 the authors formalize a diagnosis as the set of arcs of the graph representation of an aligned ontology that, once removed, breaks all the unsafe cycles.
Q14. What is the average size of the repairs?
The computed repairs are typically of limited size (less than 10%), but can reach a significant portion of the the original alignment.
Q15. What is the simplest way to characterize a restricted version of the conservativity principle?
Starting from the results of Proposition 4.2, the authors can characterize a restricted version of the conservativity principle using graph-theoretical concepts only, applied on the graph representation, without the need to refer to the aligned ontology.
Q16. What is the mapping repair algorithm for the extended horn propositional formulas?
The step 8 of Algorithm 2 uses the mapping (incoherence) repair algorithm of LogMap, for the extended Horn propositional formulas Pd1 and Pd2 , and the input mappings M. The mapping repair process exploits the Dowling-Gallier (D&G) algorithm [14, 23] for propositional Horn satisfiability (refer to [73], Section 6.3, for more details) and checks, for every propositionA of a given formula P , the satisfiability of the propositional formula PA = P ∪ {> →