Home   >  Competitions   > 

 

Submission File

 

Each piece of description in the validation.csv summarizes a scientific paper in the candidate.csv. For each piece of description, the participants need to choose 3 best-matching papers. Please note that these predictions are ordered by their probability, from high to low.

 

Please refer to sample_submission.tsv for submission format. The submission files are in tsv format without any header:

 

description_id, paper_id (1), paper_id (2), paper_id (3)

 

with each line of the file corresponds to one piece of description (description_id), followed by 3 most relative papers (paper_id). These three predictions in one line shall be different from each other. Halfwidth comma shall be used between press_id and paper_id.

 

Please note that tsv file is separated by tab. If you use Python Pandas, you can use the parameter: sep='\t'  when saving files.

 

Evaluation Metrics

 

Submissions are evaluated according to the Mean Average Precision @3 (MAP@3)

 

                                                            

Where |U| is the number of press_id in the test set, P(k) is the precision at cutoff k, n is the number of predicted papers.

 

DigSci 2019

¥40,000 (~$5,714)

123 teams

start

Final Submissions

2019-10-02

2019-10-12