W3C

Ontology of Rhetorical Blocks (ORB)

W3C Interest Group Note 20 October 2011

This version:
http://www.w3.org/TR/2011/NOTE-hcls-orb-20111020/
Latest version:
http://www.w3.org/TR/hcls-orb/
Editors:
Paolo Ciccarese, Massachusetts General Hospital / Harvard Medical School <paolo.ciccarese@gmail.com>
Tudor Groza, School of ITEE, The University of Queensland / Digital Enterprise Research Institute (DERI), Galway <tudor.groza@uq.edu.au>
Contributors:
Tim Clark, Massachusetts General Hospital / Harvard Medical School
Anita De Waard, Elsevier Laboratories

Abstract

The Ontology of Rhetorical Blocks is a formalization capturing the coarse-grained rhetorical structure of scientific publications. This note is designed to provide a general overview of the motivation and use-cases supporting ORB, in addition to the actual conceptual elements, as well as, practical examples of how to use it in conjunction with different representation languages.

Status of this document

This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the W3C technical reports index at http://www.w3.org/TR/.

The document was produced by the the Scientific Discourse Task Force, part of the Semantic Web in Health Care and Life Sciences Interest Group (HCLS), part of the W3C Semantic Web Activity (see charter). Comments may be sent to the publicly archived public-semweb-lifesci@w3.org mailing list.

Publication as an Interest Group Note does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress.

The IPR status of information provided in this document is in accordance with Section 6 of the W3C Patent Policy. The disclosure obligations of the Participants of this group are described in the charter.

Table of Contents

1. Introduction

In the last the decade, the consistently growing number of scientific publishing spheres (journals, conferences, workshops, etc) led to a significant increase in the amount of publications, with the biomedical and pharmaceutical domains being some of the most heavily affected. For example, MedLine now hosts over 18 million articles, and has a growth rate of 0.5 million / year (around 1300 articles/day) [Tsuji2009]. This makes the process of finding and associating relevant work in a particular field a cumbersome task.

One of the key issues behind this information overload problem is the inefficiency proved by the current indexing mechanisms based solely on syntactic resources. Scientific publications represent collections of artefacts contributed by authors, at different granularity levels. For example, they state claims, positions and arguments in relation to their own achievements, or the results achieved by other researchers (corresponding to a fine-grained level), or synthesise background information or experimental results (corresponding to a coarse-grained level). These epistemic items represent the key to decoding the rhetoric captured within the publications' content, and thus identifying and acquiring them could lead to novel ways of dealing with the information overload. Additionally, they may also contain important domain knowledge that can then be directly coupled with the rhetoric and argumentation semantics of the content.

Our goal is to research and develop a formal structure able to represent all these discourse knowledge items. However, in the context of this note, we focus only on modeling larger spans of text that carry a rhetorical role, and thus forming the coarse-grained rhetorical structure of the publication, via the Ontology of Rhetorical Blocks. At a later stage, we will present a formalization also for the fine-grained rhetorical structure built by means of parts of sentences, sentences or phrases.

1.1. Document Scope and Target Audience

This document describes the Ontology of Rhetorical Blocks (ORB). ORB can be used to add semantics to the structure of newly-written scientific articles as well as annotating already existing ones.

1.2. Document Conventions

The following namespace prefix bindings are assumed unless otherwise stated:

Prefix URI Description
rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns# The RDF Vocabulary
rdfs: http://www.w3.org/2000/01/rdf-schema# The RDF Schema vocabulary
xsd: http://www.w3.org/2001/XMLSchema# XML Schema
dc: http://purl.org/dc/elements/1.1/ Dublin Core Metadata Element Set, Version 1.1
dcterms: http://purl.org/dc/terms/ Dublin Core Metadata Terms
orb: http://purl.org/orb/ Latest version of the ORB ontology

2. Use Cases

The range of activities that researchers usually perform is quite broad, and usually context dependent. To simplify the description, one can classify these contexts into: actual research, project administration / management, dissemination and supervision. While this classification is quite naive and far from being unique, it is good enough to serve our purposes. From the four above-mentioned contexts, three rely on "using" scientific literature to achieve specific goals. For example, consulting the state of the art for comparative purposes (actual research), writing about own research achievements (dissemination) or teaching students how to perform research and disseminate its results (supervision). Consequently, in one way or another, publications represent the center-piece of our daily research activities, and their rhetorical and argumentation structures represent key elements in decoding the message that the authors tried to convey. In the following we present a series of use cases showing how the externalization of these structures can improve the current research workflow.

2.1. Use Case #1: Finding appropriate information within publications

Composing a literature survey requires a thorough analysis of the targeted scientific area, and hence the understanding of the contribution brought by the existing publications in that particular domain. At the same time, understanding such contributions requires one to read each and every interesting paper, which could represent a quite overwhelming task. Alternatively, one could use a search engine to find publications that may contain some specific keywords, however this will result in a set of publications that might have tangential topics to the one of interest and not to their actual contribution.

Having the rhetorical block structure externalized and attached to the digital publications would enable a richer and more expressive searching and browsing experience. One would be able to quickly spot the METHODS blocks within the publication and possibly resume the reading activity only to those, thus reducing the time usually spent on reading the entire publication. On the other hand, being able to formulate queries for content specific only to such blocks could already improve the quality (and possibly the quantity) of the set of relevant publications (e.g. methods: "autosomal-dominant mutations in APP").

In practice, the externalisation process requires human or machine annotation of the original publication, in order to create the corresponding ORB metadata. This metadata could then reside within the publication (subject to the publishing format - see below for an example of using RDFa and XHTML) or outside the publication in a metadata repository. In both cases, the publication and the metadata should be exposed on the Web following the Linked Data principles, in particular to enable the identification of the rhetorical blocks and their retrievability. Hence, to get to the METHODS block of this document one should be able to simply use the following URI: http://www.w3.org/TR/2011/NOTE-hcls-orb-20111020/#orb-methods. The actual content negotiation process is deferred to the server that publishes the document and/or metadata, while the interpretation of the hash-based identification is the responsibility of the environment that renders the publication and metadata. This last aspect is especially relevant when one tries to address a non-existing rhetorical block, in which case, the addressing could be interpreted as a declaration.

2.2. Use Case #2: Publication authoring

In general, the authoring process of a publication usually starts by creating the linear (section) structure of the publication and continues with laying the argumentative thread within this structure. This process, together with its outcome (i.e., the document), suffers from two major problems. Firstly, in the last decade, we observe a shift from printed publications to electronic publications. While for the former, the presence of the typical linear (print-driven) structure is compulsory, the latter could make use of novel structuring approaches, that would enable the externalization of both the semantics of the content and its argumentative support. Secondly, considering the current information overload, the existence of an explicit semantic structure within the publication (created already at authoring time) would reduce the overhead of post-processing the publications in order to externalize it.

The process of making the semantics explicit can be easily improved by using a fixed rhetorical structure. One could start authoring the publication from a given template of a rhetorical structure, following a core model, which could, at a later stage, be customized according to the domain of the publication. For example, if the publication is in the Experimental Biology domain, the core model only would be sufficient, while if the publication is in Mathematics, several specific modules capturing particular rhetorical blocks could be added. This scenario has two advantages: firstly, the author would have from the start a clear structure which to fill in, and secondly, each block of the structure would have a clear rhetorical role (unlike today's publication sections), making it easy to mine, retrieve and browse.

3. The Ontology

Subject to the scientific field, the coarse-grained rhetorical structure of scientific publications can take several shapes. While in the Computer Science domain we would commonly find blocks carrying rhetorical roles such as Scenario, Related Work or Evaluation, in the biomedical field we would encounter others, like Background, Experimental Results or Discussion. In principle, most of these blocks can be found in the vast majority of the domains, however, with slightly different names. As such, independently of the domain, the coarse rhetoric emerging from the publications' content will have a commonly shared semantics, but will be materialized in different domains using different terminologies.

The Ontology of Rhetorical Blocks (ORB) aims at providing a minimalistic set of common rhetorical blocks that can be leveraged from scientific publications. At the same time, it acts as an entry point to the fine-grained semantics not only of the actual publication content, but also of the publication as a document, thus capturing its provenance and external references information. Consequently, the ontology models a publication by means of three artefacts: the Header, the Body and the Tail. As described in the following sections, out of these artefacts only the Header will be materialised into an ontological entity, the rest being defined by their actual rhetoric components.

3.1. Header

Class Definition
orb:Header The part of the publication that models, or captures, meta-information about the publication, including fields such as title, authors, affiliations, publishing venue or abstract.

Every scientific publication, independently of its subject area, is accompanied by a Header. The Header of the publication contains meta-information about it, describing the provenance and the rhetorical summary. Hence, it will capture the publication's title, authors, affiliations, publishing venue, but also the abstract, as a rhetorical summary.

3.2. Body

As mentioned above, subject to the scientific domain, the rhetorical blocks may have different names. In order to be abstract enough, yet inclined towards HealthCare and Life Sciences, ORB adopts the IMRAD (Introduction, Methods, Results And Discussion) [IMRAD] structure and terminology to define the rhetorical blocks of the Body. This structure is preferred by several scientific journals and explicitly recommended in the "Uniform Requirements for Manuscripts Submitted to Biomedical Journals: Writing and Editing for Biomedical Publication" issued by the International Committee of Medical Journal Editors (ICMJE).

IMRAD [IMRAD-XML], and consequently ORB, describes the standard main structure of a scientific publication, which typically includes the following four sections in the following order:

Class Definition
orb:Introduction The section describing why was the study in the publication undertaken, what was the tested hypothesis or what was the purpose of the research. It lays down the rationale behind the existence of the publication.
orb:Methods The section describing when, where, and how was the study done. It includes the materials used as part of the study and who was included in the study groups (patients, etc.).
orb:Results The part synthesizing the results of the study presented in the paper.
orb:Discussion The section analyzing whether the tested hypothesis was confirmed. It also interprets the results to understand their consequences and importance. And finally, it shows how the approach and results fit to what other researchers in the field have discovered, including possible perspectives of future research.

3.3. Tail

Similar to the Header, the Tail provides additional meta-information about the paper, however, this time in the context of external references included in the paper. Here we distinguish between two categories that define the Tail artefact:

Class Definition
orb:Acknowledgments List pointing to funding bodies or individuals that contributed in a way or the other to enabling or supporting the work presented in the publication.
orb:References External references to other works (e.g., scientific publications, websites, software) that are relevant for the content of the current publication.

It should be pointed out that in the context of ORB, the References are not limited to the usual list of citations provided by the publication. Instead, References denote any external resources that complement, in a way or another, the work described in the paper under scrutiny, including elements such as, data, websites, software, scientific publications, etc. There is, however, a distinction being made between references to resources and to people (or organizations), via the Acknowledgments concept.

4. ORB Extensions

As already described, the Ontology of Rhetorical Blocks provides a high-level coarse-grained rhetorical structure for scientific publications. Subject to the actual rhetorical block, this can be further specified or decomposed into finer-grained elements carrying a rhetorical role. For example, the Methods block could comprise elements such as Purpose, Objects of study and Tools and procedures. As a note, this decomposition process, or the creation of a fine-grained rhetorical structure, is currently in progress in the Scientific Discourse Task of the HCLS Working Group.

One block that particular requires more attention is the Header, as it contains clear and well-established components. In accordance with the spirit of the Semantic Web, in the context of the ORB Header, we did not re-define (or introduce recycled) terms or entities to model such information. Instead, the ORB Header should be interpreted as a container of the textual representation of the publication metadata. Hence, in the following, we propose a series of guidelines (see table below) with respect to the information that should be considered when analyzing the Header, and defer the definition of the actual elements to widely used semantic models. Also, it should be noted that the resulting metadata fields should be attached to the overall publication, and not only to the Header, in order to provide a clean semantics and to be compliant with most of the current ontologies that deal with this subject. Two examples of such ontologies are the Bibliographic Ontology (BIBO) [BIBO] and the FRBR aligned Bibliographic Ontology (FaBiO) [FABIO]. Another valuable resource in represented by the Publishing Requirements for Industry Standard Metadata (PRISM) [PRISM] that makes already use of a subset of the Dublin Core namespaces [PRISM-DC]. More precisely, PRISM version 2.1 - published on 15 May 2009 - is in fact making use of the Dublin Core Metadata Element Set, Version 1.1 [DCMI] and not of the latest DCMI Metadata Terms [DCMI-Terms].

Field Description Formal definition
Title The title of the publication. dc:title
Author / Contributor Each individual that contributed to the publication. Individuals can be modeled as a foaf:Person in conjunction with the properties specified in the formal definition. dc:creator + prism:role + foaf:Person
Affiliation The affiliation of each author. Can be modeled as foaf:Organization in conjunction with the properties specified in the formal definition. swrc:affiliation
Publishing venue The publishing venue of the publication. Modeled in PRISM as publicationName: Title of the magazine, or other publication, in which a resource was/will be published prism:publicationName
Date The publishing date of the publication. dc:date
Abstract The abstract of the publication dcterms:abstract

Similar to the Header, the Tail's components can also be further decomposed into several finer-grained elements. For example, the References block will always contain an element about bibliographic citations. It may also contain elements pointing to the data used in the experiments, or even pointers to additional sets of experiments performed. However, it is important to note that, as in the case of the Header, such decompositions are not within the scope of ORB. Their definition is deferred either to a middle/fine-grained representation, or to specific models that target particular instances of scientific publications (or publishing venues).

5. Using ORB

Some examples of ORB usage are listed in the following sections. Every section is dealing with a different data format.

5.1. With XML

ORB can be easily integrated in any XML format. However, as PRISM has been defined primarily for being implemented in XML, the following XML format is based on the PRISM Aggregator Message (PAM) [PRISM-PAM] currently used by several publishers.

The PAM format allows XHTML elements to carry the prism:class= attribute. Best practice is to use such attributes on the <div> tag when special structures are complex such as several paragraphs or a paragraph and a table [PRISM-PAM-GUIDE]. However, as the semantic of such attribute has a structural connotation, the attribute new orb:class has been introduced. We believe this design choice simplifies the rhetorical blocks declaration and introduces little risk of confusion. The ORB classes are the allowed values for the attribute orb:class and no further namespaces declarations are needed. We introduced also orbext:class to enable ORB extensions.

<pam:article>     
    <head>       
    	<dc:identifier>doi:10.1523/JNEUROSCI.3107-09.2009</dc:identifier>       
        <dc:title>Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.</dc:title>       
        <dc:creator prism:role=”writer”>S.A. Austin</dc:creator>       
        <dc:creator prism:role=”writer”>M.A. Sens</dc:creator> 
        <dc:creator prism:role=”writer”>C.K. Combslt;/dc:creator>   
        <dc:publisher>The Society For Neuroscience</dc:publisher>
        <prism:publicationDate>2009-11-18</prism:publicationDate>  
        <prism:publicationName>The Journal of Neuroscience</prism:publicationName>       
        <prism:issn>0270-6474</prism:issn>       
        <prism:eIssn>1529-2401</prism:eIssn>   
        <prism:doi>10.1523/JNEUROSCI.3107-09.2009</prism:doi>                   
        <prism:volume>29</prism:volume>       
        <prism:number>46</prism:number>      
        <prism:startingPage>14451</prism:startingPage>       
        <prism:url>http://dx.doi.org/10.1523/JNEUROSCI.3107-09.2009</prism:url>       
        <prism:copyright>© 2010 The Society For Neuroscience</prism:copyright>     
    </head>  
    <body>
       <div orb:class="Header">
          ...
          <span class="citation-abbreviation">J Neurosci. </span>
          <span class="citation-publication-date" orbext:class="dc:date">2009 November 18; </span>
          <span class="citation-volume">29</span>
          <span class="citation-issue" >(<span>46</span>)</span>
          <span class="citation-flpages">: 14451.</span>
          <span class="fm-vol-iss-date">doi: <span>10.1523/JNEUROSCI.3107-09.2009</span></span>
          ...   
            
          <div class="fm-title" orbext:class="dc:title">
              Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.
          </div>
          <div class="contrib-group fm-author">
              <span orbext:class="dc:creator">S.A. Austin</span>,
              <span orbext:class="dc:creator"> M.A. Sens</span>, and 
              <span orbext:class="dc:creator">C.K. Combs</span>
          </div>
    
          <div orbext:class="dcterms:abstract">
             Amyloid precursor protein (APP) ...
          </div>
       </div>
       ...
       <div class="sec" id="S1" orb:class="Introduction">
          <div style="text-transform: none;" id="S1titletitle" class="head1 section-title">
             <div>Introduction</div>
          </div>
          <div id="S1content" class="section-content">
             <div id="P2" class="p p-first">
                Endothelial cells ...
             </div>
             ...
          </div>
       </div>
       ...
    </body> 
 </pam:article>   	
   

5.2. With RDF

As you might expect, ORB can be used to define content in RDF format [RDF]. The following snippet is showing how to use ORB in conjunction with the Bibliographic Ontology (BIBO). A similar snippet could be defined using ORB in conjunction with the FRBR aligned Bibliographic Ontology (FaBiO). In both cases, the ORB blocks are defined as parts of the specific document, which in the following example is a journal article and therefore a bibo:Article.

This is the Turtle RDF for the open access article Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells. It combines ORB, Dublin Core and BIBO:

@prefix bibo: <http://purl.org/ontology/bibo/> .
@prefix dc: <http://purl.org/dc/elements/1.1/> .
@prefix dcterms: <http://purl.org/dc/terms/> .
@prefix orb: <http://purl.org/orb/> .

[ a bibo:Article ;
     dc:creator "S.A. Austin" ;
     dc:creator "M.A. Sens" ;
     dc:creator "C.K. Combs" ;
     dc:title "Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells." ;
     dcterms:abstract "Amyloid precursor protein (APP) ..."
     dcterms:hasPart
         [ a orb:Header ;
             dc:description "Header content"
         ],
         [ a orb:Introduction ;
             dc:description "Introduction content"
         ]
         ...
].
 

5.3. With HTML and RDF

If the document is published in HTML format, it is possible to leverage HTML IDs and add RDF triples about them:

Given the HTML of the article Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells:

 
   ...
   <div class="sec" id="S1">
      <div style="text-transform: none;" id="S1titletitle" class="head1 section-title">
         <div>Introduction</div>
      </div>
      <div id="S1content" class="section-content">
         <div id="P2" class="p p-first">
            Endothelial cells ...
         </div>
         ...
      </div>
   </div>
   ...

We can create XML/RDF as follow:

<rdf:Description rdf:about="http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2820274/#S1">                 
    <rdf:type rdf:resource="http://purl.org/orb/Introduction"/> 
</rdf:Description>

In the process of authoring a new document, it is preferrable to use IDs identifying ORB sections (See Use Case #1). These IDs are obtained by adding the prefix 'orb-' to all the ORB terms. For example, the 'methods' section will be identified by the ID 'orb-methods'. In doing this, it will be possible to resolve the section by appending '#orb-methods' to the document URI

While authoring a document it is preferrable to use IDs identifying ORB sections:

   ...
   <div class="sec" id="orb-introduction">
      <div style="text-transform: none;" id="S1titletitle" class="head1 section-title">
         <div>Introduction</div>
      </div>
      <div id="S1content" class="section-content">
         <div id="P2" class="p p-first">
            Endothelial cells ...
         </div>
         ...
      </div>
   </div>
   ...
<rdf:Description rdf:about="http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2820274/#orb-introduction">                 
    <rdf:type rdf:resource="http://purl.org/orb/Introduction"/> 
</rdf:Description>

5.4. With HTML and RDFa

Another possible way of using ORB is through RDFa [RDFa Primer]. Using a few simple XHTML attributes, authors could mark up human-readable data with machine-readable indicators for browsers and other programs to interpret. In fact, RDFa benefits from the extensive power of RDF.

This is the RDFa 1.0 injected into the open access article Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells. The example includes some extensions such as title and abstract. It combines ORB, Dublin Core and FaBiO:

<div typeof="fabio:Article" about="_:a"  xmlns:fabio="http://purl.org/spar/fabio/" xmlns:orb="http://purl.org/orb/" 
  xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:dcterms="http://purl.org/dc/terms/" rel="dcterms:hasPart">
   <div typeof="orb:Header">
      ...
      <span class="citation-abbreviation">J Neurosci. </span>
      <span about="_:a"  class="citation-publication-date" property="dc:date">2009 November 18; </span>
      <span class="citation-volume" property="prism:volume">29</span>
      <span class="citation-issue" >(<span property="prism:issueIdentifier">46</span>)</span>
      <span class="citation-flpages">: 14451.</span>
      <span class="fm-vol-iss-date">doi: <span property="prism:doi">10.1523/JNEUROSCI.3107-09.2009</span></span>
      ...   
        
      <div about="_:a" class="fm-title" property="dc:title">
          Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.
      </div>
      <div about="_:a"  class="contrib-group fm-author">
          <span property="dc:creator">S.A. Austin</span>,
          <span property="dc:creator"> M.A. Sens</span>, and 
          <span property="dc:creator">C.K. Combs</span>
      </div>

      <div about="_:a" property="dcterms:abstract" >
         Amyloid precursor protein (APP) ...
      </div>
   </div>
   ...
   <div class="sec" id="orb-introduction" typeof="orb:Introduction" property="dc:description">
      <div style="text-transform: none;" id="S1titletitle" class="head1 section-title">
         <div>Introduction</div>
      </div>
      <div id="S1content" class="section-content">
         <div id="P2" class="p p-first">
            Endothelial cells ...
         </div>
         ...
      </div>
   </div>
   ...

</div>

You might notice the use of the blank node _:a through the Turtle idiom _:XXX. Alternatively it is possible to use an explicit URI identifying the article.

Triples can be extracted from the above snippets. For example, by copying and pasting the above snippet into the W3C RDFa 1.0 Distiller and Parser [RDFa-Distiller], the correspondent RDF structure will be obtained:

<rdf:RDF
  xmlns:fabio="http://purl.org/spar/fabio/"
  xmlns:dc="http://purl.org/dc/elements/1.1/"
  xmlns:dcterms="http://purl.org/dc/terms/"
  xmlns:orb="http://purl.org/orb/"
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
>
  <fabio:Article rdf:about="_:a">
    <dc:date>2009 November 18; </dc:date>
    <dc:creator>C.K. Combs</dc:creator>
    <dc:creator> M.A. Sens</dc:creator>
    <dc:creator>S.A. Austin</dc:creator>
    <dc:title>
          Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.
    </dc:title>
    <dcterms:abstract>
         Amyloid precursor protein (APP) ...
    </dcterms:abstract>
    <dcterms:hasPart>
      <orb:Header/>
    </dcterms:hasPart>
    <dcterms:hasPart>
      <orb:Introduction>
        <dc:description rdf:parseType="Literal">
        ...
        </dc:description>
      </orb:Introduction>
    </dcterms:hasPart>
    ...
  </fabio:Article>
</rdf:RDF>
 

As you might have noticed, the Header element is empty. This behavior is related to some limitations of RDFa 1.0. With the next version or RDFa 1.1 - expected before the end of 2011 - the code can include the property dc:description for the Header element without disrupting the nested declarations. Similarly it will be possible to include RDFa declarations in the other ORB sections, for example Introduction.

This is the RDFa 1.1 injected into the open access article Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells. It combines ORB, Dublin Core and FaBiO:

<div typeof="fabio:Article" about="_:a"  xmlns:fabio="http://purl.org/spar/fabio/" xmlns:orb="http://purl.org/orb/" 
  xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:dcterms="http://purl.org/dc/terms/" rel="dcterms:hasPart">
   <div typeof="orb:Header" property="dc:description">
      ...
      <span class="citation-abbreviation">J Neurosci. </span>
      <span about="_:a"  class="citation-publication-date" property="dc:date">2009 November 18; </span>
      <span class="citation-volume">29</span>
      <span class="citation-issue" >(<span>46</span>)</span>
      <span class="citation-flpages">: 14451.</span>
      <span class="fm-vol-iss-date">doi: <span>10.1523/JNEUROSCI.3107-09.2009</span></span>
      ...   
        
      <div about="_:a" class="fm-title" property="dc:title">
          Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.
      </div>
      <div about="_:a"  class="contrib-group fm-author">
          <span property="dc:creator">S.A. Austin</span>,
          <span property="dc:creator"> M.A. Sens</span>, and 
          <span property="dc:creator">C.K. Combs</span>
      </div>

      <div about="_:a" property="dcterms:abstract" >
         Amyloid precursor protein (APP) ...
      </div>
   </div>
   ...
   <div class="sec" id="orb-introduction" typeof="orb:Introduction" property="dc:description">
      <div style="text-transform: none;" id="S1titletitle" class="head1 section-title">
         <div>Introduction</div>
      </div>
      <div id="S1content" class="section-content">
         <div id="P2" class="p p-first">
            Endothelial cells ...
         </div>
         ...
      </div>
   </div>
   ...

</div>

You might still notice the use of the blank node _:a through the Turtle idiom _:XXX. Alternatively it is possible to use an explicit URI identifying the article.

Triples can be extracted from the above snippets. For example, by copying and pasting the above snippet into the W3C RDFa 1.1 Distiller and Parser [RDFa-1.1-Distiller], the correspondent RDF structure will be obtained:

<rdf:RDF
  xmlns:dc="http://purl.org/dc/elements/1.1/"
  xmlns:dcterms="http://purl.org/dc/terms/"
  xmlns:fabio="http://purl.org/spar/fabio/"
  xmlns:orb="http://purl.org/orb/"
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
>
  <fabio:Article>
    <dc:creator> M.A. Sens</dc:creator>
    <dc:creator>S.A. Austin</dc:creator>
    <dc:creator>C.K. Combs</dc:creator>
    <dc:title>
          Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.
    </dc:title>
    <dc:date>2009 November 18; </dc:date>
    <dcterms:abstract>
         Amyloid precursor protein (APP) ...
    </dcterms:abstract>
    
    <dcterms:hasPart>
      <orb:Header>
        <dc:description>
	      J Neurosci. 
	      2009 November 18; 
	      29
	      (46)
	      : 14451.
	      doi: 10.1523/JNEUROSCI.3107-09.2009
	      Amyloid Precursor Protein Mediates a Tyrosine-kinase Dependent Activation Response in Endothelial Cells.
	      S.A. Austin, M.A. Sens, and  C.K. Combs
	      Amyloid precursor protein (APP)   
        ...
        </dc:description>
      </orb:Header>
    </dcterms:hasPart>

    <dcterms:hasPart>
      <orb:Introduction>
        <dc:description>
          Introduction
          Endothelial cells ...
          ...
   	  </dc:description>
      </orb:Introduction>
    </dcterms:hasPart>

    ...
  </fabio:Article>
</rdf:RDF>

This last snippet is looking more similar to the pure RDF snippet in section 4.2.

Acknowledgments

We would like to thank Ivan Herman (W3C Semantic Web Activity Lead) for his help and support for RDF and RDFa technologies, Ron Daniel, Jr. (Elsevier Laboratories) for the help with PRISM-related issues, and Eric Prud'hommeaux (W3C) for the technical assistance with the creation of this document.

Also, we would like to thank Jodi Schneider (Digital Enterprise Research Institute), David R Newman (University of Southampton) and Scott Marshall (Leiden University Medical Center) for their valuable comments and feedback during the writing and editing of this note.

References

[Tsuji2009]
Tsujii, J. (2009). Refine and PathText, which combines Text Mining with Pathways. Keynote at Semantic Enrichment of the Scientific Literature Workshop 2009 (SESL 2009), Cambridge, UK
[DCMI]
Dublin Core Metadata Element Set, Version 1.1
http://dublincore.org/documents/dces/
[BIBO]
Bibliographic Ontology
http://bibliontology.com/
[FaBiO]
FRBR aligned Bibliographic Ontology
http://purl.org/spar/fabio
[PRISM]
PRISM Namespaces, Version 2.1
http://www.prismstandard.org/specifications/2.1/PRISM_prism_namespace_2.1.pdf
[PRISM-DC]
PRISM Subset of the Dublin Core Namespaces, Version 2.1
http://www.prismstandard.org/specifications/2.1/PRISM_dublin_core_namespaces_2.1.pdf
[PRISM-RCV]
PRISM Controlled VocabulariesNamespaces, Version 2.1
http://www.prismstandard.org/specifications/2.1/PRISM_controlled_vocabulary_namespace_2.1.pdf
[PRISM-PAM]
PRISM Aggregator Message Namespace, Version 2.1
http://www.prismstandard.org/specifications/2.1/PRISM_aggregator_message_namespace_2.1.pdf
[PRISM-PAM-GUIDE]
PRISM Aggregator Message Namespace Guide, Version 2.1
http://www.prismstandard.org/pam_2.1/PAM_guide_2.1.pdf
[DCMI-Terms]
DCMI Metadata Terms
http://dublincore.org/documents/dcmi-terms/
[IMRAD]
http://en.wikipedia.org/wiki/IMRAD
[IMRAD-XML]
http://www.uio.no/studier/emner/hf/imk/MEVIT4725/h04/resources/imrad.xml
[RDF]
RDF Primer - W3C Recommendation 10 February 2004
http://www.w3.org/TR/2004/REC-rdf-primer-20040210/
[RDFa-Primer]
RDFa Primer 1.0 - Bridging the Human and Data Webs
http://www.w3.org/TR/xhtml-rdfa-primer/
[RDFa-Distiller]
W3C RDFa Distiller and Parser
http://www.w3.org/2007/08/pyRdfa/
[RDFa-1.1-Distiller]
W3C RDFa 1.1 Distiller and Parser
http://www.w3.org/2007/08/pyRdfa/Shadow.html