Evaluation Criteria

Match_m: For each instance X in the test set D_test, we select a set S_m^(x) of m ∊ (1. . .4) words with the top m probabilities according to the ground truth. Analogously, we select a prediction set set S^{^}_m^(x) for each m, based on the predicted probabilities.

We define the metric Match_m as follows:

Terms and Conditions

This page enumerated the terms and conditions of the competition.

Guidelines for SemEval System Papers

**Note: System papers are due May 15, 2020 by 23:59 anywhere on earth time (AOE). 

Participants of SemEval-2020 are encouraged to submit a system-description paper to provide details of their system, resources used, results, and analysis. These description papers will be part of the official SemEval-2020 proceedings, which will be published on http://aclanthology.info/.

 Dates

• COLING 2020, the host conference for SemEval 2020 has been postponed to December and have shifted their deadlines accordingly. SemEval will keep the schedule mostly the same, to avoid too much overlap with SemEval 2021. Deadlines for system description papers, task description papers, notification, and camera-ready submissions have all been shifted by two weeks. Updated timeline:

• 15 May 2020 - System description paper submissions due

• 22 May 2020 - Task description paper submissions due

• 24 June 2020 - Author notifications

• 8 July 2020 - Camera-ready submissions due

Guidelines and Formatting

• All papers for SemEval 2020 should follow the COLING camera-ready formatting guidelines. NOTE: this is a change from the previously-recommended formatting! Updated style files can be found here: http://alt.qcri.org/semeval2020/index.php?id=papers

• System description papers should be maximum 5 pages long, not including bibliography (no page limit for references). An extra page will be given for the camera-ready version to incorporate reviewer suggestions.

• You can find important details about system description paper submissions such as submission site, Best Task and Best Paper Awards here.

• Papers should follow the SemEval guidelines for writing system description papers.

• Note that SemEval submissions are not anonymous; author names should be included.

• Paper titles should be in this format: "<team name> at SemEval-2020 Task 10: [Some More Title Text]"

• It may also be helpful to look at some of the papers from past SemEval competitions, e.g., here.

• We strongly encourage authors to add results on dev and test sets when trained only on the training set. In this way, future works will be more easily compared with the SemEval's task/system papers.

• We'd very much like to see interesting interpretations of the results as well as discussion on the subjective quality of the used approach in the paper. 

• You do not have to repeat the details of the task and data. Just summarize the task and cite the task description paper (details below). Then you can get into details of your submissions, experiments, results, and analyses.

Citation

• Task description paper:

• @InProceedings{shirani2020semeval, author = {Shirani, Amirreza and Dernoncourt, Franck and Lipka, Nedim and Asente, Paul and Echevarria, Jose and Solorio, Thamar}, title = {SemEval-2020 Task 10: Emphasis selection for written text in visual media}, booktitle = {Proceedings of the 14th International Workshop on Semantic Evaluation}, year = {2020}}

• All system papers must cite the task paper. Additionally, they should also cite the ACL 2019 paper below, which introduces the task and dataset:

• Shirani, Amirreza, Franck Dernoncourt, Paul Asente, Nedim Lipka, Seokhwan Kim, Jose Echevarria, and Thamar Solorio. "Learning emphasis selection for written text in visual media from crowd-sourced label distributions." In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp. 1167-1172. 2019. [BIBTEX]

Awards

​Please consider submitting your system description paper regardless of your rank in the leaderboard, as ​any results, including negative results, are valuable scientific contributions to the community. 

In addition to the winners on the leaderboard, we will announce the following:

• The best result interpretation paper

• The best negative results paper