Provide abstracts of computer science-related papers on arXiv. Participants need to use this data to predict the classification (Background, Objectives, Methods, Results, Conclusions, Others) of each sentence in the paper. link
- Number of training set : 7000 papers
- Number of testing set : 20000 papers
- init embedding+BiGRU(baseline)
- glove+BiGRU(final)
Through the expriment, we found the GRU better than the LSTM in this task
All participating teams have 469 https://tbrain.trendmicro.com.tw/Competitions/Details/8