Jump to: navigation, search

The ABCDE format for Semantic Conference Proceedings is a new way to publish easy-to-read semantically annotated conference proceedings.


Conference contributions (papers) should be easier to read and browse. Contributions are not written in isolation, but should directly be grafted onto the information space that they come from. This space contains of entities such as other scientific contributions, people, projects, software, etc. We want to publish contributions such that these relations are visible and usable, and the ideas the authors contribute are directly visible to readers and semantically enabled software.

Reading a contribution[edit]

Proceedings will be published on a webpage, that shows a paper and its surrounding context. It will look something like this:

 Title: The ABCDE Format for Semantic Conference Proceedings
  Anita de Waard - Elsevier (
  Eyal Oren - DERI (
  Heiko Haller - AIFB (
  Max Voelkel - AIFB (
 Conference: Semantic Wiki Workshop, SemWIKI at ECSW 2006
  Future conference proceedings should be easier to read and browse
  To establish this, the author should provide some basic rhetorical structure 
  We propose a Levels-of-Detail approach
  From a well-structured LaTeX source, we generate an abstract, from which the reader can drill down into the full text. 
 If this format spreads virally, it can radically enhance the accessibility of conference contributions.
 Since Entities are explicitly marked, minability and linking options are greatly enhanced. 
 Explicit markup of the Contributions allows faster access to the new aspects this work offers to man and machine. 

See figure: Abstract generated from and expandable to full content

Each of these sections can be expanded to read the complete text; a form of Stretch Text. Readers can quickly glance through the contribution and skip to the section they are interested in.

On the side, the context of the paper is shown (not in this example): references to and from this paper, related people (e.g. by co-authorship, reference, or social network), related projects, related software etc.

Readers can easily browse for related information. They can see other papers written in a project, other papers written by same people, similar papers in a research topic, etc.

Writing a contribution[edit]

Authoring should be as close to normal paper writing as possible. Almost all papers are written in the ABCDE structure. We will provide a LaTeX and Word template that needs only to be filled by the authors:

  • Annotation
  • Background
  • Contribution
  • Discussion
  • Entities (incl. references)

In LaTeX these could be denoted by \part{}. The only thing authors need to markup are the core sentences in each section - which saves them the work of finding synonyms of the article text to generate an abstract. See figure for an example of what the LaTeX could look like: Example LaTeX source

From these core sentences the abstract is generated that is shown when reading the paper.

Submissions are done in source format, we generate a PDF for offline reading and a stretch text for online reading. We also generate an RDF version for further processing.

Language processing[edit]

Paul Buitelaar of DFKI made a sample page at link his website: go to "Anita de Waard" with your cursor -- or to Heiko, Max (but not Eyal!) The VIEWS project at DFKI (see provides a system for automatic hyperlinking, which adds mouse-over menus to extracted/annotated NEs, such as person, location, etc. This could be done also for other entities in principle. Generating the menu can be configured to add links to web services (such as google, scholar, weather services, etc.) and/or information from (extracted) databases.

We are still engaged in discussions on how to progress - input is appreciated!

Collaborative Domain Ontology[edit]

A simple web service would be available for anyone to submit RDF data about entities in a domain: I can submit my foaf file, Max submits his homepage, and a doap description of the Nepomuk project.

When the annotation tool recognises a named entity, it queries the ontology for information about that resource. So it would point to my foaf data, to Max's homepage, etc.

We should not forget to allow people to submit this data to the system (and we need to figure out howto relate "entities" to a URI (to which the RDF data is bound), i.e. how to know that the entity "Eyal Oren" is the blank node with email ... mentioned in the foaf profile?

Workshop on Semantic Wikis in ESWC2006[edit]

We want to apply the idea to the Semantic Wiki workshop (currently being proposed for ESWC2006). All submissions would be in ABCDE Format, and we would generate both offline (PDF) and online (Stretch text) versions.

I have a draft paper for Semwiki2006 online at [1] - --Anita 14:15, 10 March 2006 (CET)


Don't forget to subscribe to the mailing list.