STILL TO BE DONE AND ALREADY DISCUSSED IN TELECONFERENCES 1) clarify how to introduce terms that already have a different meaning in other ISO standards 3) Michael's proposal for restructuring scope - Michael 14) Is there any reasonable “implied default” for correspondences? (Yes, the "alignment API" equivalence) - Christoph 19)/20) possibly define some notion of “interoperability” that is tailored to this standard - Michael 21) possibly use ISO/IEC 2381-1 Information Technology Vocabulary - Michael 22) phrase "logical interoperability" more precisely, based on the previously introduced terms - also see: ability of two or more systems to exchange information and to mutually use the information that has been exchanged. Defined in: ISO/IEC TR 10000-3:1998, 3.2.7 - Michael 41) [Is this actually obsolete? We have a good mereology example now; maybe just put this into Ontohub, to which the "use cases" annex will link? CL] ask Michael Grüninger for his mereology example in CL - Michael - should be in Onothub 43) The following listings are actually not EBNF according to its definition – adapt them! - Christoph 48) say that default for ENTITY-MAP-ITEMS may be overridden by specific logics, such as CASL - Till SUNDAY 50.1) comment entity-to-term mappings in the abstract syntax; we are not going to introduce them now - Christoph 51) specify nonterminal SIMPLE-ID (for BOUND-VARIABLE), reuse something from RFC 3987/3986 (IRIs/URIs) - Christoph 52) need to figure out which of the ONTO-LANG-TRANS cases we actually want to keep - Till 55) mention that the default for ALIGN-ARITIES is twice “injective and total” - Oliver 56) add concrete syntax, plus explanation: applies current default correspondence to all entities with the same local names, using the “same local name” algorithm presented elsewhere - Oliver 57) How do we say that at least one of RELATION-REF and CONFIDENCE should be given? - Oliver 58) concrete syntax e.g. a = x, b my:similarTo y %(correspond-b-to-y)%, c my:similarTo 0.75 z - Oliver 59) say that, unless a different default is specified in a surrounding CORRESPONDENCE-BLOCK, the default is equivalent - Oliver 60) check if having CONFIDENCE in correspondences to be DOUBLE really makes sense for implementations, maybe we’d like to compare confidence values for equality - Oliver 65) Requirement for prefix maps needs to be rephrased to take into account DOL XML or any XML serialization of DOL RDF, which allow prefix maps everywhere. - Christoph 67) later on we also need to say something about the semantics of the syntax. - Till 68) Introduce exactness - Till 69a) The DIST-ONTO-NAME is an IRI that should (as a good practise, but not enforced) agree with the IRI of the document. Indeed, this applies to any usage of IRI in the standard. This should be stated in the standard - Christoph 70) specify semantics of module extraction - Till 71) specify semantics of projection - Till 72) specify semantics of implicit translations using default translations - Till 73) need to say something about the semantics of MODULE - Till. Oliver suggests to invite Luciano Serafini (who has worked on distributed description logics) in order to discuss this issue. 75) Extend translational semantics to all DOL constructs - Till 77) DOL Serializations: Essential points are: - need to be able to say: “the file at URL U is in OWL 2 Manchester syntax”, - maybe use packaging/wrapping format, - compare MIME types, HTTP content negotiation (but don’t go too deep into communication protocols) - Christoph PRAGMATIC SOLUTION OF 2012-09-26: Either the remote ontology contains pre-parse information about its format (e.g. by being linked data compliant, i.e. downloadable from its IRI, and having useful HTTP headers hinting at the syntax), or there is accessible information (outside of the DOL abstract syntax) about the file format (e.g. by the ontology IRI being annotated with some RDF, or maybe using ISO 11179). Or if none of these is the case, then it's up to application developers to implement some heuristics like "file". 78) MERGED INTO 77, 2012-09-26. Maybe we can implement something like the Linux command “file”? - Christoph 79) Properly integrate this text from our LaRC 2011 paper - Christoph DONE / DEFERED 10) annotation property: why a set of pairs and not just one pair? - Christoph DONE 11) adapt definition of module to new state of syntax/semantics, which supports Sigma-modules - Oliver DONE 16) declare "renaming" as a term? maybe also further structuring operations? - Till DONE 17) rename "logic" into some more specific? - Till - no, added institute and institution instead. DONE 26) should DOL-conformance of RIF be normative? - Till DEFER 27) term for "structural elements"? - Till DONE 28) [I consider this one as obsolete. TM] 34) meet the requirements of people who combine OWL reasoners with Prolog - Till DEFER (people can add this to the registry) 38) provide example of integrating two ontologies in a single-sorted logic by translating into many-sorted logic, where only many-sorted logic would guarantee consistency - Michael CANCELLED in the 2012-08-01 teleconference --- maybe introduce relativization of quantifiers also for OWL 39) Figure out what this feedback item from Michael Grüninger (?) means: "say that there should be a syntax for relationships btw. ontologies as well as a syntax for heterogeneous ontologies. (If you write down an ontology, it might involve constructs that only exist in OWL)" - Michael DELETED 45) Address conservativity aspect separately, e.g. in the terms section. - Till DONE 49) need to explain entity-to-term mappings with bound variables (“lambda” style) - Till (see 50.1) NOT NEEDED 50) add concrete syntax for map with bound variables: ENTITY(VAR,...,VAR) |-> TERM - Till (see 50.1) NOT NEEDED 54) introduce a case to specify one (default) relation type for all correspondences of the alignment, instead of requiring one relation type for each correspondence - Oliver DONE 66) somewhere we need to mention semantic annotations to embedded fragments in conforming ontology languages, e.g. %implied - Till NOT NEEDED 69) DIST-ONTO-NAME is not used in the semantics, and this is OK. - Till 81) There need to be certain criteria before one can do a translation at all. I.e. we need to say why the translation of OBO 1.4 is one that does actually establish conformance with DOL (via OWL’s conformance). - Oliver DONE 2012-09-26 104) reference for UTF-8 - Christoph DONE 110) we need to define “sublogic” as a term - how? I guess that would include the notion of an “OWL profile” - Till DONE 206) module sublanguage -> structuring sublanguage - Till DONE 211) define logic reductions and logic projections in the terminology section - Till DONE TO BE DONE AND TO BE DISCUSSED In TELECONFERENCES 83) Mind the possible difficulty of circular imports - Till 84) say something about “infrastructure theories”, i.e. axiomatizations of one logic in another logic. Providers of ontology language translations MAY also provide these (given that the translation is theoroidal). Note the possible trade-off between readability and theorem proving complexity (as the infrastructure axioms may be complex) - so maybe we should encourage multiple alternative translations to co-exist. - Till SUNDAY 85) maybe say something about ontology languages that support multiple logics (e.g. OWL and its profiles) - Till 88) Maybe we also need child elements from different namespaces? - Christoph 89) what if an ontology language does not support IRIs? e.g. OWL doesn’t specify IRIs for import declarations, so we can, e.g., not annotate them when using the RDF serialization of OWL. We could only do it via RDF reification, or by using an XML serialization. - Christoph 92) State relation between concrete and abstract syntax in a more formal way! – Is it comprehensible? Does it sufficiently state what we have in mind? - Till DEFER 100) Old question from Nov. 2011, not sure how to deal with this: allow for correspondences with confidence and other non-logical links in such maps - Oliver 101) LOGIC-OR-LANGUAGE? On the logic side maybe just the identity Prop-logic to Prop-logic, but CASL-language to CL-language. Or, instead of identity, Prop-logic -CNF- |-> Prop-logic - Till DEFER 102) would it make sense to merge ALIGN-TYPE with INTPR-TYPE? - Till 103) For has-instance and instance-of, the Alignment API does not quite have a symbolic notation, but simply “HasInstance” and “InstanceOf”, which, in our syntax, conflicts with abbreviated IRIs. I’d suggest either referring to these relations using normal DOL IRIs (abbreviated or not), or to come up with some symbolic notation. The one I gave here works for Unicode, but I don’t really know how to write it in ASCII. - Oliver 105) maybe we need to say something about encoding IRIs as URIs in the ASCII case - Christoph 106) figure out what forbidding key words as identifiers actually means. If we use OWL Manchester’s style of abbreviating IRIs, it probably means that in the worst case some IRIs can’t be abbreviated but must be given as complete global IRIs - Till 107) develop DOL XML schema in detail - Christoph xx1) complete DOL RDF vocabulary - Christoph 111) complete conformance of OWL 2 with DOL - Till xx2) complete conformance of Common Logic with DOL - Till xx3) complete conformance of RDFS with DOL - Till 112) = 26) 113) Provide linear syntax here (as in the paper) - ??? xx4) Provide translation OWL -> CL - Till xx5) Provide translation RDF -> RDFS - Till xx6) Provide translation RDF -> OWL - Till xx7) Provide translation RDFS -> OWL - Till xx8) say that remaining translations in the graph are sublogic inclusions - Till xx9) complete conformance of F-logic with DOL - Till 115) Conformance of UML Class Diagrams with DOL - Till / telco xx10) Conformance of OBO with DOL - Oliver 116) Provide linear syntax here (as in the paper) - ??? 117) The rest of the graph can be attached to that via informative annexes. We could even just informally refer to a REPOSITORY of such translations. (Christian Galinski knows related examples from other ISO standards.) That facilitates maintenance and makes it more sustainable. - Christoph xx11) complete institutional semantics - Till xx12) provide DOL examples for use cases - Till 119) module extraction combined with projection and RDF-based querying of annotation/metadata dimensions - Oliver 120) Maybe have an(other?) appendix that refers to the usage of DOL within ontology engineering methodologies, or at least to some good practices of using DOL - Till 122) maybe mention: How do we use the ISO 12620 DCR for our extension of the OMV? - Christoph 123) update reference [31]: UML as an ontology modelling language - Till 200) ackn that rules are important but we could not add them 201) The provisions (requirements, recommendations, instructions, information) can not be in the titles and must be in the normative text. For example, the subclause title: 5.2 DOL shall be a logic-agnostic metalanguage. is incorrectly presented. 202) we will consider simplifying the terminology section on annotations and remove the terms datatype and literal, because they are actually not necessary for specifying DOL's annotation support. Use RDF or RDFa for annotating DOL text syntax - Christoph 203) check wether we need the terms XML schema and standoff markup - Christoph 204) Relations present in Lola should become notes in the terms section of the WD. 205) add faithful extensions - Till 207) check use of term "basic ontology", should be used only for unstructured ontologies 208) include unsorted FOL as an ontology language 209) include many-sorted FOL as an ontology language 210) some segregated CL dialect as an ontology language? 212) entity -> symbol (note: symbol corresponds to OWL's entities) "of the signature the ontology" "symbols *in* an ontology" - more precise: non-logical symbols - better name for that? 213) signature = symbols that denote in models (does this exclude any languages?) do we want to restrict to languages with semantics? 214) write small statement about discussion about symbol vs. entity, signature etc. 215) note about philosophical onto defs. 216) semi-formal version of translations in the standard, and fully formal versi/on(s) (also with examples) on the web/registry (multiple formats possible) 217) in standard, maintainance agency (like DFKI or Uni Bremen) is mentioned that maintains formal text and is allowed to change them (cf. data category registry, but simpler) 218) check how the term "link" is used, and if it is only used in the "ontology" context, remove the "" restriction from the term - Christoph 219) look at quantifier restriction in OWL (relativised import), how to handle it in DOL in connection with the cancelled item 38 above 220) RDF to F-logic: http://stijnheymans.net/pubs/iswc2007.pdf 221) DOL language constructs for stating the equivalence between ontologies 222) an old but important issue that hadn't made it into this list: specify local prefix maps for basic ontologies, and prefix injection into in-place as well as basic ontologies (call it "[bulk] renaming" operation in terms of "scope of the standard"). As in the 2012-04-18 slides, but do the separator character c (slide 17) by a conformance criterion, not with a syntactic construct. Say how globally-scoped non-IRI identifiers are expanded to IRIs (and in the conformance specification of, say, CLIF, hard-code how to identify globally-scoped identifiers). ------------------------- US comments ------------------------- US te,ed The document needs to be converted to the ISO template, which will make the document easier to read. (--ff) we use an ISO template for LaTeX that makes it easier for us to collaboratively work on the document. We will convert to the ISO template at the stage where this is necessary later US te,ed The provisions (requirements, recommendations, instructions, information) can not be in the titles and must be in the normative text. For example, the subclause title: 5.2 DOL shall be a logic-agnostic metalanguage. is incorrectly presented. (--ff) we will correct this have asked Frank, later US te There are some significant misunderstandings about computation, such as the definition of datatypes. The ISO/IEC 11404 (General Purpose Datatypes - GPD) is the definitive source for understanding datatypes. (--ff) Use ISO/IEC 11404 for datatypes. (--ff) we will consider simplifying the terminology section on annotations and remove the terms datatype and literal, because they are actually not necessary for specifying DOL's annotation support Christoph DONE US te While some syntactic productions claim to use ISO/IEC 14977 EBNF (a Good Thing), they are missing semicolons that terminate the syntax rules. (--ff) Fix 14977 syntax problems. (--ff) we will fix this Christoph US 2 Normative references: Only items that are cited in a normative way in the body of the standard can be included in Clause 2. All other items should move to a bibliography at the end. Also, generally, items appearing here are ISO/IEC standards; if in doubt, check with Gottfried Herzog. (--sew) we will clean up the list. However, OntoIOp relies on several W3C standards, which will remain being referenced in this section Christoph US 2 ed The normative reference to 14977 is missing. (--ff) Add normative reference. (--ff) We will do this. Christoph DONE US 2 ed Many of the normative references can be removed because they, in fact, aren't used (normatively) in the normative text. (--ff) Check which normative references are actually used. (--ff) Will do; see above. Christoph US 3 te This clause (and indirectly the whole standard) suffers from ambiguous use of the term “ontology” and related terms. One needs to distinguish between (1) the kinds of entities and their relationships between them in a given domain (2) the conceptualization of (1) by people (3) the representation of (2) as set of expressions of a given serialization of a given ontology language with a particular language (4) the set of models of the ontology language that satisfy (3). (5) Assuming that (2) matches (1), one element of (4) is isomorphic to (1), that’s the intended model. Correspondingly, one needs to distinguish between (1) entities in a domain, (2) their conceptualizations, (3) terms in formal languages, (4) their interpretations in models, and (5) the intended denotation of a term -- which is an entity in the domain. Etc. (--fn) This standard is not deadling with the philosophical distinctions, but rather straightworwardly proceeds to the formal level. In the standard, we use ontology only in the sense of (3), and this is also stated ("ontology :a formal, explicit specification of a shared conceptualization"). We do not see the need to list all five possibilities. If we want to refer to (4), we just speak about the set of models of an ontology, and if we want to refer to (5), we just speak about the intended model of an ontology. (1) and (2) are not needed in the standard. Oliver US 3 te Giving the importance of models in the rest of the standard, the term should be included (--fn) Agreed. We will define model in the sense of model theory. Till DONE US 3 te Many of the definitions are faulty or constructed poorly (as per TC37's standards). (--ff) Definitions/terminology should conform to TC37 standards (704, 1087-1). (--ff) we will fix this Till DONE US 3 te Definitions can only contain "information" provisions and not requirements, recommendations, or instructions (e.g., linked data, literal, correspondence, logically interoperable, minimization, circumscription, etc.). (--ff) Fix faulty definitions. (--ff) we will fix this Till DONE US 3 te Definitions are circular (e.g., literal, subontology, extension, correspondence, import, XML schema, logic translation, etc.). (--ff) Fix faulty definitions. (--ff) Literal, subontology etc. are not really circular; there are just additional explanations that should be moved into a note. We will do this. XML schema and logic translation are not circular. Till DONE US 3 te Missing definitions (e.g., subterm, constituent, forgetful projection, etc.). (--ff) Fix faulty definitions. (--ff) We will circumscribe forgetful. We do not intend to define the concept subterms. ISO 24707 does not define subsentence either. It is in principle not possible to define all occurring terms, the question is only where to stop. Till DONE US 3 te Poorly worded and/or ambiguous definitions (e.g., ontology, entity, resource, element, linked data, datatype, literal, profile, logic translation). (--ff) Fix faulty definitions. (--ff) we intend to fix: entity: "atomic nonlogical syntactic constituent of an ontology" logic translation: "translation of a source logic into a target logic mapping signatures, sentences and models" linked data: definition will be shortened resource, element, profile: definitions should be kept. Note that these definitions need to be so general because they have to fit for any ontology language Till DONE US 3 te Lack of explanation, relationship, or rationale for core TC37 terms that are spelled the same, but mean something substantially different in this document (e.g., term, extension, correspondence, etc.). (--ff) Fix faulty definitions. (--ff) we will fix this Till MONDAY US 3.1.1 te Guarino and Giaretta (1995) argues against the definition of Gruber (1994, 1995), and instead arrives as the following definition: “ontology: (sense 1) a logical theory which gives an explicit, partial account of a conceptualization; (sense 2) synonym of conceptualization.” Where “conceptualization” is: “conceptualization: an intensional semantic structure which encodes the implicit rules constraining the structure of a piece of reality.” Guarino, N.; Giaretta, P. 1995. Ontologies and Knowledge Bases: Towards a Terminological Clarification. In: N. Mars, ed. Towards Very Large Knowledge Bases: Knowledge Building and Knowledge Sharing. IOS Press, Amsterdam: 25-32. http://www.loa-cnr.it/Papers/KBKS95.pdf. (--lo) Change to Guarino and Giaretta (1995) definition of “ontology”: “ontology: (sense 1) a logical theory which gives an explicit, partial account of a conceptualization; (sense 2) synonym of conceptualization.” Where “conceptualization” is: “conceptualization: an intensional semantic structure which encodes the implicit rules constraining the structure of a piece of reality.” (--lo) We will change definition according to this (sense 1) Oliver US 3.1.3 te Won’t an ontology signature have to include, minimally, the vocabulary: classes, properties/relations, instances, and the supported logical connectives and operators? (--lo) no. The signature is the same as the vocabulary of nonlogical symbols. Logical connectives are logical symbols, hence do not belong to the signature. Moreover, classes do belong to an OWL signature, but not to a Common Logic signature. Here, we need to be logic independent. DONE US 3.1.6 te This may be minor. I know there is some ambiguity in the notion of “theorem”, i.e., as you have it, something that has been proven. However, a theorem assumes a proof, and if the proof is not available, a theorem is more like a conjecture. However, we call it “Fermat’s Last Theorem”, though for many years, there was no proof. (--lo) We use "theorem" in the sense that some proof is available, e.g. in the Ontohub repository. According to our definition, FLT was a conjecture until 1995, and since then it has been a theorem. DONE US 3.2.3 te Implicit in your definition is apparently a graph data model for the language of Linked Data. Should we mention therefore: the RDF graph model + IRIs? (--lo) Note that the notion of linked data in general does not depend on RDF. Christoph DONE US 3.3.3 This is not a proper definition. (--sew) I was going to try to find one, but all the sources I looked at talk about literals without actually defining them. The group will need to rewrite the definition with a superordinate (broader) concept and proper differentiating information. The additional information should either be rewritten or placed in a note. (--sew) This definition will be removed. Christoph DONE US General ge ISO definitions use neither definite nor indefinite articles. (--sew) Remove articles. Check to make sure there are no periods at the end of definitions. (--sew) We will fix this. Till DONE US 3.4. ff. The statements that follow are not proper definitions. Revise all complete statements along the lines of the suggestion for monomorphic extension. (--sew) Example: monomorphic extension extension O2 of an ontology O1, where each model of O1 can be expanded to a model of O2 that is unique up to isomorphism (--sew) Many thanks! We will revise the definitions accordingly. US 3.4.6 te The definition does not work for CL, see mail exchange http://interop.cim3.net/forum//ontoiop-wg/2012-06/msg00003.html (--fn) We propose to use a notion of reduct for Common Logic such that the definition works. Till DONE US 3.4.11 There are several issues here – this looks like an inverted term, or is it a single term at all? Might this be: minimization specific procedure for implementing … Determine what the actual term is, and define accordingly. (--sew) We define two synonymous terms: minimisation and circumscrimption. We propose to change the definition into: "way of implementing the closed world assumption by restricting the interpretation to models that are minimal" Till DONE US 3.5.1 I think the definition is OK, but the encyclopaedic information that follows needs to be in a note. There is only a single sentence fragment in any given definition. (--sew) We will fix this. Till DONE US 3.5.2 te I think “mapping” would be preferable to “link”. A given link could simply be a reference within ontology A to a concept in Ontology B. Mapping is more general. (--lo) We tend to keep link. DONE US 3.5.5 Additional info in note. (--sew) We will fix this. Till DONE US 3, in particular 3.6.3 te How does one syntactically specify a sublanguage if an ontology language is an abstract syntax without a specific serialization or vocabulary? The relationships between logic, ontology language, sublanguage, serialization, vocabulary, signature should be made explicit. (--fn) A sublanguage can be very well specified by the abstract syntax. We will make the relations more explicit. Till DONE US 3.7.2 Check whether standoff markup has been defined already – we use standoff markup in corpus linguistics as well. (--sew) We will check. Christoph US 3.9 Good notes on interoperability. I found something similar in the CDB: (--sew) ability of two or more systems to exchange information and to mutually use the information that has been exchanged. Defined in: ISO/IEC TR 10000-3:1998, 3.2.7 Information technology -- Framework and taxonomy of International Standardized Profiles -- Part 3: Principles and Taxonomy for Open System Environment Profiles (--sew) Thanks. Oliver US 3.10.2 ed Add second sentence in a note (--sew) We will do so. Till DONE US 5.2 te I’m not sure how DOL can be a logic-agnostic meta-language, since such a language will have to presume mathematics and logic, right? (--lo) New formulation: "DOL shall be a logic-agnostic metalanguage, in the sense that its constructs can be used for many different logics." Of course, at the meta level, we also (but only informally) use logic. Till DONE US 5.2 ge I reference section 5.2, but the issue really transcends specific sections. It is this: it’s unclear to me what the nature of annotation is in this standard. E.g., in section Introduction, it is stated that “DOL’s modularity and annotation constructs can either be embedded into existing ontologies as non-disruptive annotations, or they can be provided as standoff markup, pointing to the ontologies they talk about; DOL specifies a syntax and semantics for both variants. DOL’s modularity constructs are semantically well-founded within a library of formal relationships between the logics underlying the different supported ontology languages.” Is annotation symbolic, i.e., interpretable itself in DOL, or does DOL add uninterpreted strings, i.e., documentation, to ontologies? (--lo) Annotations are uninterpreted. Possible remaining uses of the term "annotation" in a semantic context (e.g. conservative extensions) will be removed. We are no longer going to embed annotations into existing ontologies. Christoph US 5.4 3 te “DOL should be capable of expressing the application T(O) of an ontology language translation T to an ontology O. DOL need not be capable of expressing ontology language translations.” Even if DOL does not need to be capable to express it, the success of this standard relies on the existence of these translations. Thus, the standard needs either to specify itself a way, how to represent these language translations or refer to an already standardized language for the purpose. (--fn) We will write down translations in the semi-formal language of mathematics. Till SUNDAY US 7 te Conformity is problematic. The understanding "institute or an institutional logic" is ambiguous, these should be defined as terms in Clause 3. (--ff) This standard should use conformity wording (or similar paradigms of conformance) that other (ISO standardized) artificial languages use. (--ff) We will fix this. Till DONE US 7 te Conformity is problematic. There is discussion on the "establishment" of "conformance", which isn't the right way of thinking about this. (--ff) This standard should use conformity wording (or similar paradigms of conformance) that other (ISO standardized) artificial languages use. (--ff) We will fix this. ??? Christian Till MONDAY ------------------------- Maria's comments ------------------------- Maria-2-2 Oliver There are many more usages than "data and service interoperability", and if we include those, it'll make the standard look more relevant. To give an idea: a list I compiled in the IJMSO paper is as follows (I deleted the references that are examples of the purposes): \begin{enumerate} \item Querying data by means of an ontology (ontology-based data access) through linking databases to an ontology (i.e., a semantic layer on top of a database) ; \item Data(base) integration, most notably the strand of applications initiated by the Gene Ontology Consortium and a successor, the OBO Foundry; \item Structured controlled vocabulary to link data(base) records and navigate across databases on the Internet, also known as `linked data'; \item Using it as part of scientific discourse and advancing research at a faster pace (including hypothesis testing with ontologies), including experimental ontologies in a scientific discipline and usage in computing and engineering to build prototype software; \item Coordination among and integration of Web Services; \item Incorporating the ontology in an ontology-driven information system destined for run-time usage, such as in scientific workflows, multi-agent systems, ontology-mediated data clustering, and user interaction in e-learning; \item Ontologies for natural language processing, including development and use of ontologies in applications such as annotating and querying Digital Libraries and scientific literature, Question-Answering systems, and materials for e-learning; \item As full-fledged discipline ``Ontology (Science)'', where an ontology is a formal, logic-based, representation of a scientific theory; \item Tutorial ontologies to learn modelling in the ontology development environment (e.g., the wine and pizza ontologies). \end{enumerate} Keet, C.M. Dependencies between Ontology Design Parameters. International Journal of Metadata, Semantics and Ontologies. 2010, 5(4): 265-284. Reply: We will abstract your use cases to 1-2 sentences in this introduction clause (which should be limited to 1 page), and add them to a revised version of annex Q. Maria-6-2 Oliver "...or universals, or types, or classes, depending on the scientific community" Reply: OK, we will add these. Revised reply: No, instead, remove "concept" and "role". Maria-6-3 Oliver This example is limited to OWL. We will add one example using Common Logic. Maria-6-4 Oliver "...or relationships or associations, depending on the scientific/user community" Reply: OK, we will add these. Maria-7-1 Christoph DONE "machine readable" or "machine processable"? computers don't 'understand' things (yet). Reply: We will change this to "machine-processable" Maria-8-1 Christoph DONE or "a binary predicate"? Reply: We will remove this term, as we will fully delegate the responsibility for annotations to RDF (or RDFa in DOL XML). Maria-9-1 Till DONE with the definitions given as is, it is not really clear to me why we have to distinguish between "module" and "subontology". Reply: We will remove this definition. Maria-9-2 Oliver c.f. teh definition of an ontology. if we keep that definition, this doesn't match with it. Reply: We expect to resolve this by choosing a different definition of "ontology" Maria-9-3 Till DONE going by the term, this entails there is also a "submodule" (and "conservative" is a special type, c.f., e.g., 'non-conservative'). And then this module/subontology: woudl this be different from a "conservative subontology"? Reply: We will rename this term to "module". Maria-10-2 Till DONE My preference of wording: Consider an OWL DL ontology about wines, from which we would like to extract an OWL 2 EL module about white wines that will contain the declaration of the entity “white wine”, all declarations of entities related to “white wine”, and all and only those axioms about these entities that are compliant with the OWL 2 EL profile. if not updated, then at least "OWL EL" -> OWL 2 EL Reply: We will remove the aspect of projecting to EL but handle projection separately from module extraction. Maria-11-2 Till DONE can this be reworded? as I have read it several times over, and still wonder if I actually understood it or not. Reply: logical link referring to another ontology (and the rest will go into a note anyway) Maria-12-a-from-us Till DONE we will rename this to "structuring sublanguage" Maria-12-2 Christoph giving our own definition to "XML schema" that is different from everywhere else in the world is not a good idea, and asking for confusion and misunderstandings. Going by the definition, can't we call it something like "XML Grammar"? Reply: This is not our own invention but a wide consensus e.g. on Wikipedia. Independently from that we will reconsider whether we really need this tern; so far one conformance criterion depends on it. Maria-14 Christoph DONE (3.10.3) We will remove this; it's not necessary. Maria-15-1 Till DONE There's: OWL full OWL DL OWL Lite OWL 2 full OWL 2 DL OWL 2 EL OWL 2 QL OWL 2 RL Reply: we will adopt all of the "OWL 2 *", but we don't consider OWL 1 relevant. Maria-15-2 Christoph This section has to be structured properly. There are different types of requirements, a.o., features that the language itself has to have vs. availability/accessibility of the standard vs. ones motivated by implementations and serialization. And it is good practice to separate the shall from the should, with shall (/must have) first. On formatting: sometimes there are indents, sometimes itemized sometimes paragraphed; is there a specific reason for that? Reply: We will consider grouping the requirements section, by different types. Inside the sections we will order the requirements by importance. We will check the formatting; most likely it's an artifact of the LaTeX template we are using. Maria-16-1 Christoph why is this not a "shall"? Reply: because we can merely hope that DOL will really work with any ontology language. But, as agreed above, we will first mention the concrete languages that DOL shall support. Maria-16-2 (Note 27) Till DONE Reply: will rephrase whole item to "basic and structured ontologies (and facilities ...)" Maria-17-1 Till DONE where? and is it the link types that are defined, or the reuse? Reply: We will rephrase this. We mean to reuse correspondence types defined in external specifications or ontologies, e.g. in the Alignment API. Maria-17-2 Christoph Can we really get by with using just the other annotation vocabularies? E.g., I'm quite sure I'd want to annotate declared links. How would I be able to do that? Reply Annotating a declared link will be possible by giving the link an IRI. But then, any semantically meaningful annotation (other than just a source code comment) needs to have an RDF semantics, and therefore a predicate/property. As so many suitable properties exist already, we do not intend to reinvent any within DOL. Maria-17-3 (5.3) Christoph DONE shall?!? that is, if we want to facilitate wide tool support. Besides, further below it is stated that there what the serializations shall have to be capable of. Reply: OK, "shall" :) Maria-17-4 Till DONE uh? "Alternatively" indicates what comes afterward is another option, but I don't think that was meant here. or is it? Reply: Indeed, we rather meant "additionally". Maria-18-1 (5.4) Christoph DONE shall?! Reply: OK Maria-19-1 Till DONE uhm, no. it doesn't to me. Reply: We will remove the confusing example in the note, and instead give a clearer definition of "projection" and "reduction" in the terminology clause. Maria-19-2 (5.6) Christoph DONE shall?! Reply: OK Maria-19-3 Till DONE but on p17, paragraph 3, it says "defined externally to this standard" (which I read as a ''read up on source x'). So, what is it? I'm fine to have it within the standard (and have some ideas myself); or at least, we define several types and provide a 'template' for easy extensions for people who want to define more link types. Reply: we correct p.17, see above. We will have all the link types specified within the standard. Actually we need to revise the related terminology. We suggest the following wording: "This standard shall specify a fixed number of links types. Some of the link types will be restricted to logical links (e.g. interpretations), others will allow both logical and non-local links as instances (e.g. alignments)." Please go ahead with proposing link types! About a "template" we're not sure what it should be and whether it's necessary. Isn't it sufficient, e.g., that an author of a DOL ontology can add new non-logical correspondence types (to be used in alignments) by simply introducing a new subproperty of an existing externally defined property? We do not think it will make sense to have an extensible set of link types that are restricted to logical links, as it involves a lot of mathematics (as seen in this standard) to formalize their semantics. Instead we should aim at building all desirable such link types into DOL. Maria-20-1 Oliver the symbols are not clear. What does the arrow mean, what the diamond, and what the open circle, (beyond the labels of the lines)? And the coloured blobs in the upper DOL roundtangle? My guess is that the intention is that there's the DOL system with two imported ontologies, with one that is in OWL/XML syntax and is of some OWL species (the ASK-IT). Then there's another one of which neither the language not the serialization language is given (DOLCE is available written FOL, in KIF, and there are several OWL-DL versions serialised in RDF/XML); why that one is only linked to the CL-interpretation-circle is unclear to me. Reply: we thought we had made the edge and node types derby labeling all of them; e.g. an arrow with an open circle tip denotes an import. We will revise the bottom rectangles to clarify that we mean: There is an OWL formalization of ASK-IT, serialized in OWL XML, and a CL formalization of DOLCE, serialized in CLIF(for the sake of this example we are just assuming it has been made available in CLIF, by adapting the KIF source), both of which are stored in external files but nevertheless participate in the overall DOL distributed ontology. Maria-20-2 Till DONE this sounds really vague. Reply: We will rephrase: DOL should allow for expressing ontologies and links at different levels of detail. Maria-21-1 Christoph DONE to me, this "shall" sounds like a contradiction to the "should" in the section heading. Reply: Right, the heading shall also be "shall". Maria-23-1 Till DONE from prior terms&defs, it is not clear to me what is the difference between the two (and says todo in the semantics) Reply: We will explain this. Maria-23-2 Till DONE I presume you mean "of lower expressivity" Reply: OK, we will rephrase this. Maria-24-1 Till First it says there is no semantics to an alignment, but being able to distinguish between subsumtion, equality, and any other ones (see later comment), does amount to giving it a semantics at least implicitly, so I would assume we'd like to make that explicit as well. Reply: Agreed. We will introduce a semantics for alignments where correspondences have confidence value 1 and the correspondence type is such that the semantics is straightforward (e.g. equivalence, subsumes, is-subsumed, incompatible, has-instance, instance-of) Maria-25-1 Christoph DONE do you really mean *arity*, as in an alignment can involve multiple ontologies? From the line afterwards, I'd say you mean cardinality, not arity. Reply: Thanks, "cardinality" is the right term Maria-25-2 Till, Oliver it would be nice to specify more of them by default, like part, and a 'template' or something for user-defined ones. The latter may make the OBO people happy, as they are fond of so-called "cross-products", being relations across taxonomies (using, a.o., involved in, participates in, and so on). more info on that approach: Mungall, C. J., Bada, M., Berardini, T. Z., Deegan, J., Ireland, A., Harris, M. A., Hill, D. P., et al. (2010). Cross-Product Extensions of the Gene Ontology. Journal of biomedical informatics 44 (1), 80-86 Reply: we will consider adding this, even with a semantics. Maria-27-1 Till DONE manchester syntax is a Note, not a standard Reply: replace "standards" by "best practices" Maria-27-2 Till DONE why the difference? Reply: right, "changes" is enough. Maria-28-1 Till DONE This sounds odd. "can" seems more appropriate Reply: yes Maria-30-1 Till DONE "Signatures (\Sigma)..." one could say it is obvious, but it is neater to introduce all the symbols beforehand. Reply: yes Maria-30-2 Till DONE Def 1 has an additional "Sen" Reply: yes Maria-31-1 Till DONE Def 3 is an example, just like the others in this paragraph. It is out of place to make the example above a Definition, as that gives the impression it is part of all the institution stuff, which it is not. One easily can reformat it to put all of it in a clearly delineated example. Reply: Def. 3 provides an example of an institute. We will make this clearer. Maria-31-2 Till DONE an informal description, as done for others, would be appreciated (you're an expert on it, but very few others are familiar with these terms). Reply: yes Maria-34-1 Till SUNDAY is this the closing of the axioms for CWA? I'd hoped that would end up as a CLOSE-xxx or similar. Reply: We suggest to add close-world as a synonym for minimize Maria-40-1 Oliver OBO is a DAG, not random stuff like the UML standard is. Besides, there are several papers who claim to have 'The Real Mapping' from OBO to OWL. NO REPLY YET! Maria-41-1 Christoph how is that possible? you mean in the sense of not being one of those in the appendices? Reply: We will phrase this more precisely. It means that there could be ontology languages without a logic-based semantics (therefore not conforming with DOL) but having an XML syntax in which any element can be given an IRI (therefore bring accessible e.g. to non-logical DOL alignments) Maria-45-1 Christoph DONE difficult to choose. what about LOLA? (LOgic LAnguage) Reply: OK, let us use LOLA-REF Maria-61-1 Oliver see previous comment on OBO. e.g., based on http://www.cs.man.ac.uk/~horrocks/obo/ or the most recent trials and tribulations reported at the OWLEd'12 workshop last week: Vicky Dritsou, Elvira Mitraka, Pantelis Topalis, Christos Louis Getting the Best from Two Worlds: Converting Between OBO and OWL Formats http://ceur-ws.org/Vol-849/ or some other (whicheever one). Reply: thanks Maria-67-1 Christoph + a how that can be done Reply: yes, but probably this annex will be moved a repository of use cases Maria-67-2 Christoph well, a lot more than that as well. Reply: OK, we will improve this Maria-68-1 Christoph or DLRus, and we could make it more interesting by also adding TDL-Lite (a simplified DLRus) Reply: thanks for this comment