CSIBS: The Citation-Sensitive In-Browser Summarizer

Guide to this Demo

Click to see an example of an article (HTML format) annotated by CSIBS

CSIBS Functionality Overview

Scientific researchers need to stay up-to-date with the latest advances in their fields. However, in a world where everyone is increasingly time-poor, near exponential growth in the available literature makes this a challenging task. Based on a user requirements analysis, we present a new research tool, called the Citation-Sensitive In-Browser Summariser (CSIBS), which supports researchers in browsing through scientific literature. As one reads an online article and encounters a citation that looks important, CSIBS creates a preview summary of the cited document. The key innovation is the contextual tailoring of the automatically generated summaries using the citation and its surrounding text. As this context changes, so too does the citation-specific summary portion of the preview, which contains contextually-relevant sentences extracted from the cited document. The CSIBS preview presents relevant information required to appraise the citation, containing meta-data about the reference, the abstract and the citation-specific summary. Thus, CSIBS, alleviates information overload by enabling the reader to determine whether or not to invest time in exploring the cited article further.

Related Links:

• Videos of the 4 finalists in the Elsevier Grand Challenge: To hear more about CSIBS, view the 4th presentation entitled: "Browsing through Scientific Literature with Citation-Sensitive Previews"
• Our paper at the 2009 Joint Conference on Digital Libraries (JCDL 2009) entitled "Whetting the Appetite of Scientists: Producing  Summaries Tailored to the Citation Context"  by Stephen Wan, Cécile Paris, and Robert Dale.
  • Please cite this paper if you wish to refer to this prototype
  • Short summary: This paper presents an overview of the CSIBS tool
• Our paper at the ACL 2009 Workshop on text and citation analysis for scholarly digital libraries (NLPIR4DL) entitled "Designing a Citation-Sensitive Research Tool: An Initial Study of Browsing-Specific Information Needs" by
  Stephen Wan, Cécile Paris, Michael Muthukrishna and Robert Dale.
  • Short summary: This paper presents the initial user requirements analysis that we conducted.  In it, we describe  information needs related to the citation-based browsing scenario in the domain of scientific articles.

General Notes:

• In this demo, we provide enhanced versions of articles available via the Elsevier ScienceDirect portal (a cached sample is provided above).

• Although this demo works on all browsers, it has been optimised for viewing with Firefox and Chrome. Internet Explorer (IE) users might notice a series of click when pages are loaded. This is an IE interface issue and this sound does not occur with Firefox or Chrome.

What you should see:

• Web version:

  • Viewing the enhanced web article: All citations have the default ScienceDirect pop-up containing the full reference and citation counts for the cited document. We have augmented this pop-up with meta-data, which includes the abstract. Starred citations indicate that the cited document can be found in the Elsevier dataset that we are using, and thus a Citation-Sensitive preview can be generated. (Note that the pop-up will display below the mouse cursor, if possible. If not, when it displays, it may block reading of the article, in which case, readers should first scroll down. This is the default interface behaviour of the ScienceDirect website.)

  • What to look out for in the preview: Automatically extracted sentences that are relevant to the citation context are now grouped together under the immediate section heading, since the initial user requirements indicated that information about article structure was desirable. Words matching the context are highlighted. Note that the preview adapts to the citation context.

    
    Click to see a larger version

  • Caveats: Some special characters may display as a question mark.