DocEng'19 Competition on Extractive Text Summarization

Call for competitors

This DocEng 2019 competition focuses at the challenges of automatic extractive and semi-extractive text summarization.

    • This competition quantitatively assesses the extractive summarization methodology/tool proposed using the 3,000 texts in the CNN-corpus in English for single-document extractive text summarization of news articles.
    • Some of the texts in CNN-corpus will be available for training at this competition webpage.
    • At the end of the competition, the testing dataset along with the required program to run the evaluation measures will be made publicly available.
    • The assessment methodology will be via the direct matching between the selected sentences and the gold-standard summary and also using ROUGE comparing each of the summaries with the highlights and the gold standards.
    • The number of broken anaphoric expressions in the summaries will also be computed.
    • The submitted methods will also be compared with the performance of other widely acknowledged publically available summarization tools.
    • A report on the competition will be published in the DocEng 2019 conference proceedings.
    • The authors of the two overall best summarization methods will be invited to be co-authors of the final report of this competition.

NOTE: You may participate in this contest even if you do not plan to attend the DocEng 2019 conference.

Important Dates

Mar. 15, 2019 Competition opens to the participants

May 25, 2019 Deadline for registration for the contest

June 25, 2019 Deadline for submission of the required executable code as well as a description of the participants’ summarization methodology.

July 16, 2019 Paper submission to the DocEng 2019 Program Committee describing the contest and the obtained results.

Sep. 23, 2019 Final contest results to be announced at the DocEng 2019 conference.


Rafael Dueire Lins
Rafael Ferreira
Steve Simske