This is an entry into the Elsevier Article 2.0 Contest by 67 Bricks Ltd. It demonstrates how scientific article publishing can be improved by applying Web 1.0, Web 2.0 and Web 3.0/Semantic Web approaches to add value to article content. The application enhances content navigation, allows commenting on specific paragraphs and features of images, and allows facts to be asserted about the article and its contents.

Web 1.0 - Enriched content display and navigation

Inline reference display - this allows references to be shown more conveniently without having to go to the end of the document. Hovering over a reference to a footnote in the main body of the text will display the text of that reference directly (note that footnote references of the form "28,29" are supported, but the form "28-30" is not currently supported).

Unobtrusive paragraph-specific linking and bookmarking - this allows specific paragraphs within the document to be bookmarked, linked to, and sent to others. By hovering over a paragraph with the mouse, a hash character # is displayed at the end of the paragraph. Clicking on this marker will take the browser directly to that paragraph. The browser URL can then be bookmarked, copied into an email, put into a web link, or otherwise shared. These # markers do not appear except when the paragraph is hovered over, so do not interfere with the normal reading of the text.

Auto-generated table of contents - a table of contents is automatically created for the document to simplify navigation.

REST-based design - the system uses the REST architectural approach - pages are referred to by a simple URL naming scheme that leads to URLs such as "/elsevier/00012998/0036/0001/05000553/article.html#2", where the URL is structured by the ISSN, Volume, Issue and Article, and then optionally the specific paragraph within the article. This structure is simple to use with web analytics tools, and straightforward to use as a basis for writing additional services to extend the existing system.

Accessible design - the pages are designed for accessibility to make them usable by as wide a range of readers as possible; to do so is both ethical and commercially sensible. The system is using semantic markup where possible, and is using W3C standard HTML and CSS. It is resilient to text resizing and will reflow the text effectively for very small and very large text sizes. The content uses minimal non-essential HTML markup within the page, and uses the principles of Unobtrusive JavaScript to avoid cluttering page content with code that is not relevant to users.

Uniform navigation between journals, volumes, issues and articles - at each level, the Up link in the control panel at the top right of the page links to the table of contents for the level above.

Web 2.0 - Collaborative content creation

Paragraph level granular comments - comments can be added to specific paragraphs within the document, allowing specific claims by the author to be commented on or disputed. Comments are displayed in order of creation, so multiple users can comment on the same paragraph and discuss a particular point. Comments are added by clicking on the "Add comment" link to the right of each paragraph. These are attributed to the user who created them, if they are logged in to the application.

OpenID-based identity management allows users to log in using their existing identities, such as their Yahoo or AOL account, without having to create a username and password specifically for this website. A user enters the URL of an OpenID provider into the login box (for example, "http://yahoo.com/", or "http://openid.aol.com/[screenname]" or http://[username].wordpress.com), and then the third-party OpenID provider confirms the users identity to this website. OpenID is a grass-roots federated identity protocol supported by numerous websites and Open Source libraries, and corporate supporters include IBM, Google, Microsoft and Yahoo.

Authority-based comment display and filtering - the comments attached to the article are highlighted based on their source. Comments from anonymous users are all displayed in a uniform grey. Comments from authenticated users are displayed against a coloured background, and multiple comments from the same user will all be displayed using the same colour within a given article. Hovering over a comment from an authenticated user will display the real name of that user, if the user has chosen to make it available via their OpenID provider. Comments from an authenticated user who has been identified out-of-band as an expert on the topic are highlighted in green and displayed with a distinctive thick border - this allows specific commenters such as the author of the paper to be recognized by readers. The control panel at the top right of the page allows comments to be filtered based on their source - for example, so anonymous comments can be filtered out.

Region-specific image annotation - third parties or the original author are able to highlight particular features within images and add comments to those image areas. When hovering over an image, areas that have been annotated are highlighted by a square box. Hovering over this box shows the annotation that was made. By clicking on the image, an annotation can be added at the point that is clicked on. The same provenance information is recorded for these annotations as is recorded for the other comments.

Web 3.0 - adding factual assertions to content

Paragraph and reference level assertions - readers can add assertions to paragraphs to make factual claims about the content. This is using the Semantic Web RDF data model. This allows the author or other readers to state facts based on the content, such as that the article is for a study with a study size of 40. This enables researchers to more easily carry out meta-analysis of articles, and users to add rich content to articles. Assertions are added with the Add Assertion link displayed to the right of paragraphs, and then the type of assertion has to be selected.

Enhancing references with DOIs - Digital Object Identifiers can be added to references in the document, by the document author, or by third parties. This allows references to be made more useful for users. This is a specific example of the reference level assertions described above - DOIs are added by selecting the Add Assertion link next to an item of content, selecting a type of DOI, and entering the DOI.

Credits

As with any modern application, the original development within this project depends upon many third-party libraries and frameworks. The "nicetitle" JavaScript, the "annimg" JavaScript and the "generated_toc" JavaScript are scripts by Stuart Langridge released under an MIT license. Other scripts are using the Prototype JavaScript library, released under an MIT license. The framework in use is Apache Cocoon, released under the Apache license 2.0, running on Java. The OpenID authentication uses dyuproject, released under the Apache license 2.0. XSLT transforms are using Michael Kay's Saxon, released under the Mozilla Public License. The unobtrusive paragraph linking approach is inspired by the Purple Numbers mechanism invented and named by Doug and Christina Engelbart. The data is supplied by Elsevier under the terms of the Elsevier Article 2.0 Contest. Thanks to Eamonn Neylon for pointing out this contest and for early discussion of ideas, and to Dr. Michelle Reid for her review and suggestions. All other code is original and written by Inigo Surguy of 67 Bricks Ltd.

Current limitations and directions for future development

The current implementation only works in standards-compliant web browsers - it has been tested in Firefox, Safari and Chrome. It will not work in Internet Explorer. Nor is it optimized for small screens. However, it is not using any functionality that could not be supported in Internet Explorer given additional development time.

The current implementation does not support all possible Elsevier formatting - it is intended to be indicative of possibilities rather than fully complete. For example, mathematical formulae will not always be rendered accurately.

The Elsevier content does not contain unique identifiers at the paragraph level, so the application has to create its own paragraph identifiers. These will not be robust across revisions of the articles. It would be desirable for Elsevier to add paragraph-level identifiers within their XML content, so content can more dependably be linked to and annotated.

About the creators

67 Bricks Ltd. work with information specialists and content owners who are looking to generate more value from their information. We help them develop new ways to utilise and sell their information.