Ontology authoring involves making choices about what subject domain knowledge to include. This may concern sorting out ontological differences as well as making choices of conflicting axioms due to limitations in the logic. Examples are different foundational ontologies in ontology matching and OWL 2 DL's transitive object property versus qualified cardinality constraints. Such conflicts have to be resolved. However, there is currently only isolated and fragmented guidance for doing so, which therefore results in ad hoc decision-making. This work aims to address this by working towards a framework dealing with the various types of modeling conflicts through meaning negotiation and conflict resolution in a systematic way. The approach was evaluated with an actual case of domain knowledge usage in the context of epizootic disease outbreak.
An increase in the use of ontologies brings with it the task of reusing existing ones, which
is already an aspect of the OBO Foundry approach [
How to proceed? While one could discard a relevant ontology and start afresh, we
assume that a user may wish to attempt to resolve any issues that may arise. A few tools
are available to assist with detecting and inspecting issues; e.g., the explanation features
in Prote´ge´ [
This work aims to contribute to addressing these obstacles by devising a novel approach for meaning negotiation in ontology development and (re)use, and for conflict resolution. We examine the possible principal sources of conflicts for both individual ontologies and multiple ontologies. For each case, there is a fixed set of feasible solution strategies, so that then explanatory implications may be automatically generated. Some of the components can be computed automatically, whereas others require the human-inthe-loop to make the final, but now well-informed, decision. The approach is illustrated and evaluated with a case of domain knowledge usage to manage an epizootic disease outbreak in Switzerland (avian influenza), which involved negotiation and resolving conflicts regarding the ‘appropriate’ mereotopological theory, its trade-off with the OWL species, and exceeding OWL 2 DL when combining ontologies.
The remainder of the paper is structured as follows: An approach to meaning negotiation and conflict resolution is introduced in Sections 2 and 3, respectively. The use case is presented in Section 4. A conclusion is drawn in Section 5.
Negotiating the meaning of the knowledge to be represented in an ontology involves reaching an agreement on: 1) the exact elements required, 2) the domain theory that will provide these elements, and 3) the required ontology language to represent the former. This may involve meaning negotiation and conflict resolution, which will be disam
biguated and illustrated first. Potential sources of conflict are identified and the resolution processes elaborated on in more detail.
We define informally the notions of meaning negotiation and conflict resolution, illustrate them, and then outline how they arise.
Meaning negotiation concerns deliberations to figure out the precise semantics one wants to represent in the ontology. They are all positive choices in the sense of ‘which of the options is applicable? then we take that one’.
Conflict resolution concerns choosing one option among a set of two or more options, where that choice is deemed the ‘lesser among evils’ for that scenario, necessarily involves a compromise, and making it work requires reengineering something in at least one of the ontologies or as a whole. Subtypes: Language conflict resolution A conflict arises within the same family of languages or with a more distant one. This is a zero-sum game (i.e., with a winner and a loser) or there may be a joint outside option.
Ontological conflict resolution The ontologies adhere to different theories. They may be foundational philosophical decisions that affect the overall structure of the ontology or subject domain arguments with competing theories. This is likely a zero-sum game (no joint outside option).
when modelers are not sure whether they need a mereological theory with or without atom, offering them a dialogue “if you add Atom to ground mereology, then you obtain the following novel deductions [listing]; do you want that?” or frame negotiation of alternative commitments as an imperative, e.g., “take either parthood or proper parthood as primitive for your mereological theory, but not both.”.
nality constraints, but not both, with the same object property, as illustrated in the
tion’ (a term from game theory) where neither wins, but there is an alternative option.
side either setting and into the DOL framework [
An example of an ontological conflict is a clash in the top-level organisation of the ontology, such as between BFO and GFO, and related philosophical differences, such as whether concepts are allowed in the ontology. At the subject domain level, this may be, e.g., whether a virus is a living thing or not, and, more generally, competing scientific theories. In principle, they do not have a joint outside option, other than reverting it back to the domain experts to resolve this by, e.g., conducting experiments in the lab.
Where do such conflicts come from? The sources of issues arising can be manifold.
Six principal cases were discerned, which are non-exclusive and possibly not exhaustive:
1. Ontological differences between established theories; e.g., extensional mereology
(EM) vs. minimal mereology (MM) and DOLCE vs. BFO as top-level ontology.
2. Ontological differences at the axiom-level; e.g., whether part-of is antisymmetric.
3. Different modeling styles; e.g., foundational ontology-inspired or conceptual
model-influenced; e.g., reification or not [
unsatisfiable class due to signed but disjoint ancestors. While this may also have
as source an ontological difference at the axiom-level, it manifests either after
adding the axioms, during test-driven development (TDD) [
The first three are, in principle, a priori negotiations by an ontologist, but may manifest only upon ontology matching. Cases 4 and 5 emerge during ontology authoring. The last one may or may not be a priori. 2.2. The Conflict Set Conflict detection offers opportunities for automation and, even though there is no single way of how conflicts can be detected, some tasks can be carried out with the aid of state-of-the-art ontology development environments (ODEs) (see Section 3).
The data structure in which the detected conflicts are stored, and upon which the resolution of conflicts operates, is called conflict set. A conflict set is generated in all cases where a conflict is detected. We will illustrate this in Section 4. Without loss of generality, it is assumed that, when matching more than two ontologies, a conflict set is generated for every pair. Conflict sets can be described in a context-free grammar in Backus-Naur Form as follows (the productions for most terminals are omitted): <conflict-set> ::= <ontology> <ontology> [<diff>] <ontology> ::= <IRI> [<species>] <axiom> {<axiom>} [<inference>] <species> ::= "OWL DL" | "OWL Lite" | "OWL Full" | "OWL 2 EL" | "OWL 2 QL" | "OWL 2 RL" | "OWL 2 DL" | "OWL 2 Full" | "FOL" | "HOL" <axiom> ::= [<number>] <formula> [<description>] {<theory>} {<dl-expressivity>} <theory> ::= <IRI> | <name> | <IRI> <name> <diff> ::= difference between the inferred axioms sets of the two ontologies Accordingly, there are two ontologies (or two fragments of the same ontology), each identified by an IRI (or another identifier) and composed of a (possibly singleton) set of axioms. An axiom may adhere to an ontologically well-founded theory, such as BFO or ground mereology.
In practice, conflict resolution often starts with some issue raised by the ODE, specifically, when an axiom is added, or an ontology is merged or integrated into the active ontology. Examples of such issues are undecidability, language profile violation, and incoherence. They can be seen as cues indicating that something is wrong with the active ontology. The author then has to find out what raised the issue. Thereby, they may be supported by the ODE. Proceeding that way is not as straightforward as one might expect, because there is no one-to-one correspondence between conflict and issue. Examples of such ‘causal investigations’ will be given in Section 3.2–3.5. For the rest of this section, the following principal choices are presupposed: (i) The author sticks to Occam’s razor when authoring an ontology for the case at hand: the least expressive language in which the required axioms can be represented fully is preferred over all more expressive ones. (ii) The author wants to capture as much of the semantics of the domain theory as possible. (iii) The author prefers a decidable language over FOL or HOL for representing a domain theory and a coherent ontology over an incoherent one.
The first point is a general principle in almost every situation in life. The second point assumes that the author prefers representing a full axiomatization over a partial axiomatization and, by extension, a partial one is better than mere primitives without any axioms. While the third point may not hold in all situations, we deemed it realistic to include, since most software infrastructure caters for decidable ontology languages and coherent ontologies, and Semantic Web and Knowledge Graph languages in particular. 4 status of an element competing (scientific) theories No. Conflict 1 foundational 2
mereological 3 domain theory 5 ontological 6 7 within-language family modeling style
Description Examples
Conflicting theories at the top-level ontologies adhere to conflicting theories BFO, DOLCE, GFO, SUMO, UFO,
YAMATO conflicting mereological theories with vs. without Atom, whether parthood is antisymmetric or not, weak vs.
strong supplementation
Conflicting theories at the subject domain level competing theories monotheism vs. polytheism, marxism vs. leninism whether virus is a living thing or not
Axiom-level conflicts conflicting theories acting out on the see rows 1–3 axiom-level undecidable violation of a language pro- some of the non-admissible axiom comfile binations in Example 2 decidable violation of a language profile functional and transitive properties in
OWL 2 QL
Other conflicts applied vs. foundational class vs. object property there is / there is no data property axiom Infection vs. infected-by
If an ontologically well-founded theory underlying some axioms to add or an ontology to integrate is in conflict with the ontology representing the desired theory (nos. 1–3 in Table 1), the respective IRIs must be added to the conflict set. This presupposes that the conflict is known and the pair of IRIs is listed somewhere, for instance, in a library of common conflicts. To give an example, if one considers adding the part-whole relations taxonomy that happens to be aligned with DOLCE to a BFO-aligned infectious disease ontology (IDO), then the theory conflict (BFO vs. DOLCE) will be detected by looking up the library of common conflicts. Conflict resolution, in this case, aims at preserving a consistent theory. Since for conflicting theories there is no joint outside option, the ontology author has to decide in favor of one theory and discard the other. Their decision may be informed by the deliberations of what should be represented in the ontology made during meaning negotiating.
State-of-the-art ODEs support the import of ontologies. After import, their IRIs can be read from the metadata of the active ontology and looked up manually for common conflicts in a library. Accordingly, this conflict detection approach is straightforward to implement. Uncommon conflicts are harder to detect, hence to resolve. The use of ontology design patterns for a theory would be helpful in automating detection, as would annotations. Further, the library of common conflicts may grow upon finding more conflicts, so that it can prevent the same or a similar conflict from emerging later on in the project.
See nos. 5–6 in Table 1. Here, conflict resolution aims at preserving a decidable ontology
language or raising awareness of undecidability when opting for a joint outside option.
In the first case, this most often is a zero-sum game: the ontology author has to chose
which ones of some conflicting axioms in the conflict set to keep. For mereotopological
theories, these types of conflicts are well-investigated [
Least number of axioms affected; preferred axiom type identified by assigning weights; least number of inferences lost.
The least number of axioms affected can be read from the conflict set. Assigning weights to axiom types implies that certain types are a priori considered more valuable than others. For instance, one may weigh existentials more than universals and unqualified cardinality more than qualified cardinality. The least number of inferences lost requires an additional step at which the inferences of the ontologies are computed and recorded in the conflict set. If undecidability is caused by an ontological conflict at the axiom-level that was not resolved along with conflicting theories (e.g., weak vs. strong supplementation in Example 2), then also the decisions taken when negotiating meaning upfront may serve as a criterion. The authors’ decision and the criteria upon which it is based should be recorded, in case the same or a similar conflict emerges later on in the project.
The second case (opting for a joint outside option) requires that principal choice (iii)
(preferring a decidable language) is relaxed. Theories that are represented in different
logics can be dealt with by the DOL framework [
State-of-the-art ODEs provide some support for detecting and resolving these kinds
of conflicts. To give an example, the OWL API [
of minimum cardinality, maximum cardinality, exact cardinality, functionality, inverse functionality, reflexivity, irreflexivity, asymmetry, role disjointness. Mereology: weak supplementation (pp(x; y) ! 9z(p(z; y) ^ :o(z; x))) in MM vs. strong supplementation (:p(y; x) ! 9z(p(z; y) ^ :o(z; x))) in EM. Temporal logics: dense time (8t; t0 2 T; t < t0; 9t00:t < t00 < t0) vs. discrete time (there is a first and last time point t, and no time point between t and t + 1).
See nos. 5–6 in Table 1. Here, the case where conflict resolution aims at preserving the language to the extent that it is decidable again has to be distinguished from the case where the original language profile should be preserved. Presupposing the principal choices above, conflict resolution in the first case aims at capturing as much of the semantics of the domain theory as possible. Since the violated language profile is not the most expressive one, there may be room for a (decidable) joint outside option. To give an example, the axiom O(x; y) =de f 9z(P(z; x) ^ P(z; y)) (i.e., ‘overlaps’) cannot be expressed in any decidable OWL species. While preserving decidability, the author may still want to state that P (x; z) ^ P(z; y) is a sufficient condition for O(x; y). Or they may want to state that O(x; y) is a reflexive and symmetric property. Doing so may violate the original language profile. Whether the ontology language still is undecidable with the modified axioms and conditions can be figured out in the same way as described in Section 3.2. Weakening the theory step by step this way will end up in a decidable language, since the representation of properties as mere primitives is always possible in OWL and other ontology languages.
In the second case, conflict resolution aims at preserving the original language
profile at the expense of relaxing principal choice (ii), i.e., accepting that ‘as much
semantics as possible’ is less than anticipated. This applies to conflicts emerging from what
is called design for scalability in Section 2.1. Here, conflict resolution is likely to be a
zero-sum game similar to that described in Section 3.2 (undecidability). The OWL 2 QL
profile, for instance, is aimed at applications that use very large volumes of instance data,
such as conventional relational database systems, and where query answering is the most
important reasoning task. Violating this profile means accepting that query answering
may no longer be implementable by rewriting queries into a standard relational query
language [
In the first case, tool support is the same as that described in Section 3.2
(undecidability). In the second case, the OWL Species Classifier supports authors of OWL
ontologies by listing which axioms violate which OWL species (see footnote 2). The OWL
Species Classifier was also used to search through the 417 axioms of the CIDO ontology
for COVID-19 [
exceed the OWL 2 EL profile, considering scalability and compatibility with typical OBO
has a class expression with a universal quantifier on the right-hand side; more specifically, `Yale New Haven Hospital SARS-CoV-2 assay' v 8'FDA EUA-authorized organization' is one of the multiple axioms that violate the OWL 2 EL expressiveness restrictions in the cido.owl of 14 June 2020, and is also present in the cido-base.owl of 18 June 2020.
Conflict resolution, in this case, aims at preserving a coherent ontology. Examples include ontological misspecifications at the axiom-level, such as disjoint ancestors, resulting in unsatisfiable classes (no. 5 in Table 1). Such conflicts manifest only when making deductions by a reasoner. In the simplest case, they are resolved by keeping some of the conflicting axioms and removing others in a way similar to that described in Section 3.2 (undecidability). In the example, either the disjointness axiom on the ancestors or the subclass axioms on the class may be kept, but not both.
State-of-the-art ODEs allow for making deductions. After running the OWL
reasonser in Prote´ge´ 5.2, for instance, unsatisfiable classes and properties are displayed in
red color. In order to find out what made them unsatisfiable, justifications can be
computed using the respective plug-in. A justification is a set of axioms from an ontology that
is sufficient for an entailment to hold [
These conflicts arise from source 3 in Section 2.1. Resolving them aims at restructuring (parts of) an ontology such that correspondences with entities of a different ontology can be established. For instance, if the same notion is modeled in one ontology as a class and in another as an object property (see Figure 1), or even in both ways in the same ontology, and the ontology language does not permit heterogenous alignments, then either the object property has to be reified or the class has to be recast as an object property (see no. 7 in Table 1). Typical examples are object properties such as o1:married-to and o1:hasmember and corresponding reifications as o2:Marriage and o2:Member, respectively. A concrete difference is illustrated in Example 4.
Example 4 Consider again the CIDO ontology and now also the CODO ontology3 for COVID-19: codo:`laboratory test nding' fpositive, pending, negativeg, i.e., the outcomes are instances, whereas in CIDO, there is a cido:`COVID-19 diagnosis' class with three subclasses [negative/positive/presumptive positive] COVID-19 diagnosis. This is an example of class vs. instance modeling of the same idea.
What is recorded in the conflict set depends on the case at hand; for the class vs.
property example, these would be the respective axioms to match and the axioms they
are used in, which may be found by using an NLP-based algorithm with POS tagging
and stemming. Generally speaking, there are two different options of dealing with
conflicting modeling styles. The first is to convert one modeling pattern into the other, and
the second option is to match patterns by a set of axioms, rather than by a single bridging
axiom, which is referred to as a heterogeneous TBox mapping [
The way how state-of-the-art ODEs deal with conflicting styles depends on
the kind of conflict and the ODE. For instance, Prote´ge´ 5.2 restrict alignments to
owl:equivalentClass statements. Simply put, it does not provide the necessary means
to bridge a class and an object property. A joint outside option may be the DOL
framework [
3https://bioportal.bioontology.org/ontologies/CODO
The approach proposed here is tested against a case of epizootic disease outbreak in the
Lemanic Arc (France, Switzerland) in 2006 [
Assume the administrative division of Switzerland is represented in
administrative ontology O1 and the finds of infected birds as well as protection and surveillance
zones are represented in epidemiology ontology O2 (the ontologies can be downloaded
from https://www.envidat.ch/dataset/icbo2020). In order to construct a query
against a geodatabase to figure out which municipalities to include in which zones, the
two ontologies need to be merged. Both are OWL 2 DL ontologies with an
expressivity of ALCRIF and SRIF , respectively. They have been implemented using Prote´ge´
5.2 [
In order to represent the administrative division properly, every region occupied by a municipality is assigned to exactly one region occupied by a district. Accordingly, the object property partOf is functional in ontology O1. For the finds of infected birds in ontology O2, on the other hand, the same object property needs to be transitive: The (small) regions occupied by the finds are contained in the regions occupied by the protection zones. These are contained in the regions occupied by the surveillance zones. Merging the two ontologies, thus, results in a conflict which is reported by the following conflict set: Ontology: O1 IRI: appl:administrative No.: 1.17 Axiom: has 2D u has 2D inv u located in u partOf v ? Description: disjointness of roles Theory: n/a DL: (:); R No.: 1.22 Axiom: > v ( 1 partOf) Description: functionality Theory: n/a DL: F ; Q Inference: (O1 + O2 2:32) j= O10
Ontology: O2 IRI: appl:epidemiology No.: 2.32 Axiom: Tr(partOf) Description: transitivity of roles Theory: M DL: S; R Inference: (O1 + O2 1:17 Diff: (O10 - O20) = 0/ 1:22) j= O20
The conflict is resolved by trading transitivity of appl:epidemiology#partOf for functionality of appl:administrative#partOf and disjointness of roles in the administrative ontology. Doing so affects less, but equally preferred, axioms than the other way round, namely, one axiom vs. seven axioms (please note: ‘axiom’ 1.17 is a shorthand notation for six individual axioms omitted due to space limitations). It loses exactly the same inferences (diff is empty) as trading in the opposite direction.
First steps towards a framework dealing with the various types of modeling conflicts through meaning negotiation and conflict resolution in a systematic way have been proposed. We introduced and specified the notions of meaning negotiation and conflict resolution, made clear what their components are, and took a first step towards conceiving a library of conflicts. The notion of conflict set was introduced as a minimal data structure in which the detected conflicts can be stored and upon which a software-mediated conflict resolution can operate. This approach was evaluated with an actual case of domain knowledge usage in the context of epizootic disease outbreak.
As has been shown, there is no single way of detecting and resolving conflicts. A number of common cases were described by distinguishing (1) ontology from language, (2) theory-level from axiom-level, and (3) upfront negotiation from resolution upon manifestation. While there are some tools and plugins that can assist with meaning negotiation and conflict resolution, no integrated support is currently provided. Future work includes refining the framework such as to specify software design requirements for automating conflict detection, based on which algorithms for conflict resolution can be devloped or fine-tuned, as well as establishing the conflict library in more concrete terms.