Several ontologies have been created in the last years for the semantic annotation of scholarly publications and scienti c documents. This rich variety of ontologies makes it di cult for those willing to annotate their documents to know which ones they should select for such activity. This paper presents a classi cation and description of these state-of-the-art ontologies, together with the rationale behind the di erent approaches. Finally, we provide an example of how some of these ontologies can be used for the annotation of a scienti c document.
Semantic publishing [
Therefore, the result of this work is a classi cation of the most important
ontologies for describing scholarly documents. The proposed classi cation divides
ontologies into three main groups: ontologies for describing the document
structure (sections, paragraphs, etc.), ontologies for describing the rhetorical elements
(introduction, results, etc.) and ontologies for describing bibliographies and
citations. In what follows, we expand on these ontologies, and show how they
can be employed to describe a scienti c paper. We also illustrate how some of
them could be applied to a published article from the journal Future Generation
Computer Systems, which is already encapsulated as a Research Object [
The rest of the paper is organized as follows: Section 2 explains the main ontologies that describe documents; Section 3 focuses on describing ontologies for the scienti c discourse; Section 4 presents the works that attempt to annotate the references of a document; Section 5 introduces other vocabularies that allow annotating documents. Finally, Section 6 concludes the paper and depicts some recommendations to annotate a scienti c document. 2
In this section we describe ontologies that allow describing the structure of a scholarly article or, more generally, of a document. Each ontology is presented with its main characteristics (classes and properties) and an example of use.
One of the earliest works in this direction was the no-longer maintained Document ontology3, implemented in the SHOE language. This ontology focuses only on the document type. Some of the documents types de ned in this ontology are: Abstract, Letter, Form, Lecture, etc.
The Ontology of Rhetorical Blocks (ORB)4 captures a coarse-grained
rhetorical structure of scienti c publications, independently of their domain. The
ontology models a publication by means of three artefacts: the header, the body and
the tail. The header is the part of the publication that models meta-information
about the publication, including title, authors, a liations, publishing venue and
abstract. The body is composed by four rhetorical blocks: introduction, methods,
results and discussion, according to the IMRAD [
A recent work that attempts to annotate the entire characteristics of a
document is the Semantic Publishing and Referencing Ontologies5 a set of ontologies
that allow describing books and journal articles, citations, bibliographic records,
the component parts of documents, and various aspects of the scholarly
publication process. This set of ontologies is composed by:
2 http://rohub.linkeddata.es/motifs_bundle_page-FGCS/
3 http://www.cs.umd.edu/projects/plus/SHOE/onts/docmnt1.0.html
4 http://www.w3.org/2001/sw/hcls/notes/orb/#ontology
5 SPAR, namespace http://purl.org/spar
{ FaBiO6, the FRBR-aligned Bibliographic Ontology, which allows recording
and publishing bibliographic records of scholarly documents.
{ CiTO7[
DoCO, Documents Components Ontology The DoCO ontology provides a broad number of classes and relationships that allow describing a document based on its structure and content. DoCO imports two ontologies: Deo and the Document Structural Patterns Ontology. Deo is an OWL2 ontology that describes the major rhetorical elements of a document. It also provides a structured vocabulary for rhetorical elements within documents and it uses all the rhetorical block elements from the SALT Rhetorical Ontology [9]. The pattern ontology de nes formally patterns for segmenting a document into atomic components, in order to be manipulated independently and re- owed in di erent contexts.
DoCO describes the vast majority of document components such as chapter, preface, glossary, etc. Table 1 shows some of the classes from this ontology. 6 Namespace http://purl.org/spar/fabio/ 7 Namespace http://purl.org/spar/cito 8 Namespace http://purl.org/spar/biro 9 Namespace http://purl.org/spar/c4o 10 Namespace http://purl.org/spar/doco 11 Namespace, http://purl.org/spar/deo 12 Namespace, http://www.essepuntato.it/2008/12/pattern 13 Namespace http://purl.org/spar/pro 14 Namespace http://purl.org/spar/pso 15 Namespace http://purl.org/spar/pwo
As far as Deo is concerned, Deo supports the main rhetorical elements in a document, e.g., Introduction, Methods, Results and Conclusions. These elements give a de ned rhetorical structure to the paper, which assists readers to identify the important aspects of the paper. Notice that the rhetorical organization of a paper does not necessarily correspond neatly to its structural components (sections, paragraphs, etc.). In this sense, Deo and DoCO complement one another. Table 2 shows some of the most relevant classes from Deo. acknowledgements introduction related work motivation background conclusion future work methods
results discussion problem statement biography 3
The scienti c discourse has particular characteristics that are not covered by the aforementioned ontologies. Particularly, a scienti c discourse has goals, claims, experiments, evaluations and so on. Indeed, the reasoning of the assertion of the scienti c document is crucial for scholarly and scienti c publishing, in proposing hypotheses and advancing evidence in their support. Several works have been proposed to model the discourse argumentation normally present in scienti c articles.
One of the rst works to address the modelling of the scholarly discourse was ScholOnto[10]. The ScholOnto ontology provided a small set of conceptual and relational types. The main class of the ontology is the Claim. All claims are owned by an agent, and have some form of justi cation. Claims assert new relationships with other claims, or between concepts.
The work proposed by [11,12] identi es the main components of scienti c
investigations and construct the Core Information about Scienti c Papers (CISP)
metadata about the content of papers. The main classes proposed in CISP are:
Goal of investigation, Motivation, Object of investigation, Research method,
Experiment, Observation, Result and Conclusion. CISP metadata makes use of the
ontology of experiments EXPO16[
As aformentioned, CISP makes use of EXPO, a very complete ontology about scienti c experiments. The aim of this ontology is to provide a controlled vocabulary of scienti c experiments. For this purpose EXPO de nes over 200 concepts to allow providing a formal description of experiments for e cient analysis, annotation and sharing of results. EXPO is able to describe computational and physical experiments, experiments with explicit and implicit hypothesis. EXPO de nes general classes including Scienti cExperiment, ExperimentGoal, ExperimentTechnology, ExperimentResult, etc. (see Table 4).
Inspired by EXPO and CISP metadata, the work described in [13] proposes the Core Scienti c Concepts, CoreSCs. The CoreSCs is a scheme built upon eleven categories at the sentence level that allows the automatic recognition of each one of the categories in scienti c articles. The CoreSC includes: hypothesis, motivation, goal, object, background, method, experiment, model, observation, result and conclusion. The authors argued that these categories describe the main components of a scienti c investigation. The rst application of this scheme has been used to automatically annotate papers in Biochemistry and Chemistry.
Other works, such as The Argument Model Ontology17, use the `Toulmin Model of Argument' [14]. Toulmin proposed a layout containing six interrelated components for analysing arguments: claim, evidence, warrant, backing, quali er and rebuttal. The Argument Model Ontology models this components through a set of 8 classes and 21 properties (see Table 5). The following snippet shows an example of use of this ontology: 17 http://www.essepuntato.it/2011/02/argumentmodel : sentence1 dcterms : description " We propose a catalog of domain independent conceptual abstractions for workflow steps that we call scientific workflow motifs " . : sentence2 dcterms : description " We present an empirical analysis performed over 260 scientific workflow descriptions . " . : argument1 a amo : Argument ; amo : hasClaim : sentence1 ; amo : hasEvidence : sentence2 .
A very recent work is the one proposed in [15] MicroPublications18. In this work the authors employ the Toulmin's model updated by Bart Verheij [16] and then they propose a semantic model of scienti c argument and evidence designed for representing the key arguments and evidence in scienti c articles. MicroPublications proposes a model to construct an argumentation network linking textual statements and data as evidence for claims.
Beyond the works that employ a linguistic model there are other works that are focused on describing the scienti c discourse itself and the relations among the claims and hypotheses made by the author of the document. That is the case of the last two works that we present here.
The SWAN19 ontology [5] models the scienti c discourse. The SWAN project is part of the Annotation Ontology [17] and it has evolved into the Domeo annotation toolkit20 (a web application enabling users to create and share ontologybased annotations on HTML and XML documents). The core of the SWAN ontology models the discourse elements providing a model of assertions, questions and hypotheses. The SWAN discourse elements are: { Research statements: a claim or an hypothesis. { Research questions: topics under investigation. { Structured comments: the structured representation of a comment published in a digital resource.
On the other hand the SWAN ontology also provides discourse relationships, which are a set of relationships that can be used to build scienti c discourse. Some of the discourse relationships are in Table 6. 18 http://purl.org/mp 19 Semantic Web Applications in Neuromedicine http://www.w3.org/TR/hcls-swan/ 20 http://swan.mindinformatics.org/ refers to inconsistent with alternative to relevant to arises from motivates 4
The references of a document play an important role in the paper. One of the most widely used ontologies for describing bibliographic entities is BIBO, The Bibliographic Ontology Speci cation [18]. BIBO de nes a set of classes to identify the type of document based on its origin (journal, book, webpage, etc.), where bibo:Document is the key class of this model. BIBO includes Dublin Core terms to cover common needs, uses FOAF (Friend of a Friend)21 to describe authors, and adds other classes and properties, as shown in Table 7.
Although the BIBO ontology identi es in a unique way each paper, BIBO is unable to express the history of a paper. In this sense, the work proposed by [19] extends the BIBO ontology in order to include the internal work ow of journals, conferences, and so forth. The result of this work allows tracking the history of a scienti c paper.
As mentioned in Section 2, there are other ontologies focused on describing
bibliographic entities, such as FaBiO22 [
Finally, the Citation Typing Ontology23 enables the characterization of the nature or type of citations, both factually and rhetorically. CiTO contains 41 object properties that add more information to the cite (e.g., agrees with, corrects, likes, uses method in). CiTO allows characterising citations in three ways: explicit citations (e.g. the reference list of a journal article), indirect citations (e.g. a citation to a more recent paper by the same research group on the same topic), or implicit citations (e.g. as in artistic quotations or parodies, or in cases of plagiarism). Some of the CiTO properties are enumerated in Table 8.
There are also other general ontologies that have been used for the annotation of any type of documents, not just scienti c ones.
The most obvious and extended one is Dublin Core Metadata Terms (DCT), which contains fteen properties to specify the characteristics of electronic documents (creator, date, contributor, description, format, etc). The terms in DCT are intended to be used in combination with terms from other vocabularies, as we saw above in the ORB vocabulary.
Friend of a Friend (FOAF) is a stable ontology that contains some classes such as Agent, Person, Organization, Group, Project, Document, Image, etc., and some properties to describe the instances of these classes. This vocabulary 23 CiTO, namespace http://purl.org/spar/cito allows describing the authors of documents, their a liations and other relevant information about them.
Finally, the Semantic Web Conference Ontology24 is an ontology for describing academic conferences. This ontology establishes how to use classes and properties from other ontologies (FOAF, Dublin Core, SIOC25 and iCal/RDF Calendar26) and provides some classes for things relative to conferences. Some of these classes are: AcademicEvent, ConferenceVenuePlace, Proceedings, etc. 6
24 http://data.semanticweb.org/ns/swc/ontology 25 http://sioc-project.org/ 26 http://www.w3.org/TR/rdfcal/ the cited paper, because the author of the citing paper says that this document extends their previous work.
Finally, we point out how to use the FOAF ontology to annotate each one of the authors of the document. The use of this ontology allows describing people and their relations.
In brief, we propose to use the DoCO ontology for describing the document structure, the Deo ontology gives way to describe the vast majority of rhetorical elements, but we believe that it is necessary to extend this ontology in order to cover the concrete elements of the application domain. The BIBO ontology describes references according the kind of document and all the characteristics involved in the publication process. We propose to use the CiTO ontology for describing the rhetoric of the citations (in this way we can establish a network with other works). In order to describe the scienti c discourse it is necessary to describe the most important elements of the domain (e.g. the biological domain has probably di erent discourse elements than the computer graphics domain). An important part in the scienti c discourse are the claims done by the author of the paper and how they are contextualized. For this purpose we propose to use the discourse model proposed by [11]. This model can be annotated with most of the CiTO properties. In this sense, our nearly future work is to provide an ontology for describing the scienti c discourse according to the aforementioned model by doing an extension of the proposed ontologies.
In this paper we have described the most relevant ontologies used for describing scienti c documents. We have classi ed these ontologies into three main groups, those that describe the discourse of a document (either scienti c or generic), those that allow describing the document structure and those dedicated to describe references and citations. At last, we have presented other well-known vocabularies that provide generic information about any type of document. Moreover, we have also sketched out a model for the semantic enhancement of documents based on some of these ontologies, mainly those that describe the document structure, their rhetorical elements and their references.
This work is supported by the FP7 European project Dr Inventor FP7-611383. doco:title
Common motifs in scientific workflows: An empirical analysis doco:list of authors
Daniel Garijo, Pinar Alper, Khalid Belhajjame, Oscar Corcho , Yolanda Gil ,Carole Goble foaf:person Ontology Engineering Group, Universidad Politécnica de Madrid, Spain
School of Computer Science, University of Manchester, United Kingdom
Information Sciences Institute, Department of Computer Science, University of Southern California, United States Abstract. Workflow technology continues to play an important role as a means for specifying and enacting computational experiments in modern science……… scientific:hypothesis deo:background Acknowledgments This research was supported by the…
doco:bibliography
Fig. 1. The proposed model for describing a scienti c document taking together some of the described ontologies. 5. Ciccarese, P., Wu, E., Wong, G., Ocana, M., Kinoshita, J., Ruttenberg, A., Clark, T.: The SWAN biomedical discourse ontology. Journal of Biomedical Informatics 41(5) (October 2008) 739{751 6. Bechhofer, S., Buchan, I., De Roure, D., Missier, P., Ainsworth, J., Bhagat, J., Couch, P., Cruickshank, D., Delder eld, M., Dunlop, I., Gamble, M., Michaelides, D., Owen, S., Newman, D., Su , S., Goble, C.: Why linked data is not enough for scientists. Future Generation Computer Systems 29(2) (February 2013) 599{611 7. Sollaci, L.B., Pereira, M.G.: The introduction, methods, results, and discussion (IMRAD) structure: a fty-year survey. Journal of the Medical Library Association 92(3) (July 2004) 364{371 8. Peroni, S., Shotton, D.: FaBiO and CiTO: ontologies for describing bibliographic resources and citations. Web Semantics: Science, Services and Agents on the World Wide Web 17 (December 2012) 33{43 9. Groza, T., Handschuh, S., Moller, K., Decker, S.: SALT - semantically annotated LaTeX for scienti c publications. In Franconi, E., Kifer, M., May, W., eds.: The Semantic Web: Research and Applications. Number 4519 in Lecture Notes in Computer Science. Springer Berlin Heidelberg (January 2007) 518{532 10. Shum, S.B., Motta, E., Domingue, J.: ScholOnto: an ontology-based digital library server for research documents and discourse. International Journal on Digital Libraries 3(3) (October 2000) 237{248 11. Liakata, M., Teufel, S., Siddharthan, A., Batchelor, C.R.: Corpora for the conceptualisation and zoning of scienti c papers. In: Proceedings of the International Conference on Language Resources and Evaluation. (2010) 12. Soldatova, L., Liakata, M.: An ontology methodology and CISP - the proposed core information about scienti c papers. Programme/Project deposit (December 2007) 13. Liakata, M., Saha, S., Dobnik, S., Batchelor, C., Rebholz-Schuhmann, D.: Automatic recognition of conceptualization zones in scienti c articles and two life science applications. Bioinformatics 28(7) (January 2012) 991{1000 14. Toulmin, S.E.: The Uses of Argument. Cambridge University Press (July 2003) 15. Clark, T., Ciccarese, P.N., Goble, C.A.: Micropublications: a semantic model for claims, evidence, arguments and annotations in biomedical communications. Submitted to the Journal of Biomedical Semantics (May 2013) 16. Verheij, B.: The toulmin argument model in arti cial intelligence. In Simari, G., Rahwan, I., eds.: Argumentation in Arti cial Intelligence. Springer US (January 2009) 219{238 17. Ciccarese, P., Ocana, M., Garcia Castro, L., Das, S., Clark, T.: An open annotation ontology for science on web 3.0. Journal of Biomedical Semantics 2(2) (July 2011) 1{24 18. D'Arcus, B., Giasson, F.: Bibliographic ontology speci cation (November 2009) 19. Hu, Y., Janowicz, K., McKenzie, G., Sengupta, K., Hitzler, P.: A linked-datadriven and semantically-enabled journal portal for scientometrics. In Alani, H., Kagal, L., Fokoue, A., Groth, P., Biemann, C., Parreira, J.X., Aroyo, L., Noy, N., Welty, C., Janowicz, K., eds.: The Semantic Web. Number 8219 in Lecture Notes in Computer Science. Springer Berlin Heidelberg (January 2013) 114{129 20. Tillett, B.: What is FRBR? a conceptual model for the bibliographic universe.
The Australian Library Journal 54(1) (2005) 24{30 21. Liakata, M., Soldatova, L.: Guidelines for the annotation of general scienti c concepts. Aberystwyth University, JISC Project Report (November 2008)