'''Note:''' The work on Ontologies in Talia/Swickynotes/Philospace is ongoing, and this page may not always reflect the most current state. This page discusses some of the conceptual issues in using different ontologies. == Semantic Web Ontologies in Talia and Philospace == Talia and Philospace will share as much as possible the same ontologies. This should enable data compatibility and make data exchange among the two applications easier. Starting from the ontologies, that we created so far within the discovery project and that are used (more or less :-)) in Philospace, and from some the most promising ontologies around, we are trying to come up with a new and more comprehensive ontology. The ontologies currently used within Philospace are: * The Scholarship Ontology. Link: [http://semedia.deit.univpm.it/ontologies/philo_0.2.owl]. A simple ontology that defines very basic classes and properties. * Domain Ontologies. Extensions of the Scholarship Ontology for specific domains. They define new classes, properties and instances. An example is the one used for the Nietszche domain: [http://dbin.org/brainlets/discovery/philo_nietzschesource_0.3.owl] The ontologies currently used in Talia are: * The Talia structural ontology, see source:trunk/discovery_app/ontologies - it includes and maps to * The Talia ontology re-uses/maps elements from existing ontologies. See the config file source:trunk/discovery_app/config/talia_core.yml.app_example (at the bottom) for the configured namespaces. The re-used ontologies are * Dublin Core elements * Dublin Core Terms * Dublin Core Types * FOAF Existing ontologies: * Bibliographic Ontology. * Link: [http://bibliontology.com/] * Functional Requirements for Bibliographic Records (FRBR). * Conceptual schema: [http://www.ifla.org/VII/s13/frbr/frbr.htm] * Expression of Core FRBR Concepts in RDF: [http://vocab.org/frbr/core] * Dublin Core [http://dublincore.org/] * DC in RDF [http://dublincore.org/documents/dc-rdf/] '''Considerations and toughts''' (work in progress...) The FRBR is a conceptual schema that aims at describing bibliographical objects that can be books, videos, maps, 3d objects. Thus The domain is similar to the one we are facing in Discovery. Objects in FRBR can be described at different levels: '''works, expressions, manifestations and items'''. The question is: '''do we need to consider such a distinction? and Why?''' A distinction among the phisical (or particular) representation, e.g. of a book, and the idea of book itself can be useful: If someone says that a given version of "I promessi sposi" (a particular edition of the book) talks about marriage, as soon as a new edition or version of the same book is added one would like to automatically infer that also this new version talks about marriage ("annotation reuse"). However, if the annotation says "this edition sucks" such a behavior is no more desirable. N.B.: annotation reuse is complex in the case the annotation refers to a fragment of a book (e.g. a sentence): is a new edition of the book is loaded it might be very difficult of impractical to automatically find the right sentence in such a new edition (this can be in a different language, or translated in a different way, etc.). At a first glance all Talia content will be "Items". However, it might be useful to display to a user all the properties that are associated to the corresponding manifestation, expression or work (e.g. the author is associated to the work, the editor to the expression, etc.). FRBR defines a property "has as subject" that matches exactly with the "has subject" that we are using in the current version of the scholarship ontology. We also have different specifications (sub-properties) of such a property (e.g. quotes, contradicts, defines, etc.). Possible ranges for such a property are Persons, Concepts, Objects, Events and Places. FRBR also defines a sets of relations that might occur among Endevours (i.e. Works, Expressions, Manifestations and Items), these are a superset of those we encision to use for our purposes (isPartOf, translationOf, revisionOf, etc.). FRBR defines a set of attributes that can be associated to Endevours. The RDF representation developed by Talis (http://vocab.org/frbr/core) uses OWL and captures all these things except for the attributes. The feeling is that all the attributes (and relations) defined in the FRBR final report can be easily mapped to Dublin Core elements/terms. This is what, more or less, some people around has experienced. For example [http://w3.uniroma1.it/ssab/er/relazioni/jorgensen_eng.pdf] describes a project (Visual-Cat, [http://visualcat.udel.edu/]) that uses a RDF representation of FRBR as a normalization layer to/from different metadata formats (DC, MARC). While [http://dc2006.ucol.mx/papers/jueves/14.00/happyoufinal.ppt] argues that each DC element can be mapped to FRBR model. '''Work types''' Among the FRBR attributes there is one that might be of particular importance to us: "form". It represents the "class to which the work belongs (e.g. novel, play, poem, essay, biography, etc.)". However no controlled vocabulary is indicated to be used in conjunction with the form attribute. This is somehow related (if not equivalent) to the DC term "type", that allows as values the classes listed in the dc:types module. This list is very general and contains few classes (see http://dublincore.org/documents/dcmi-type-vocabulary/). The Bibliographic Ontology provides a rich hierarchy modeling resource types which include all the basic classes contained in dc:types. '''Form vs. Intention''' While all the previous mentioned vocabularies define categories of objects under the point of view of their "form", some of the Domain Ontologies developed in Discovery, for example the Nietzsche Domain Ontology (one of the latest version here [http://dbin.org/brainlets/discovery/philo_nietzschesource_0.3.owl]) defines categories of objects based on their "intention". For example there is a Interpreting class which includes different kind of interpretation of something (e.g. a Lecture, an Interview, a Review, etc.). Each one of these types of objects can be represented in different forms. For example an interview can be a video, a transcription, or even an jpg image! This said, I would model the two meaning of "type" (Form and Intention) in separate hierarchies. For the form I would use the Bibliographic Ontology (perhaps integrated with some additional kinf of forms in needed), for the Intention I would use something like the Nietzsche Domain Ontology.