@inproceedings{eger-etal-2017-eelection,
    title = "{EELECTION} at {S}em{E}val-2017 Task 10: Ensemble of n{E}ural Learners for k{E}yphrase {C}lassifica{TION}",
    author = "Eger, Steffen  and
      Do Dinh, Erik-L{\^a}n  and
      Kuznetsov, Ilia  and
      Kiaeeha, Masoud  and
      Gurevych, Iryna",
    editor = "Bethard, Steven  and
      Carpuat, Marine  and
      Apidianaki, Marianna  and
      Mohammad, Saif M.  and
      Cer, Daniel  and
      Jurgens, David",
    booktitle = "Proceedings of the 11th International Workshop on Semantic Evaluation ({S}em{E}val-2017)",
    month = aug,
    year = "2017",
    address = "Vancouver, Canada",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/jlcl-multiple-ingestion/S17-2163/",
    doi = "10.18653/v1/S17-2163",
    pages = "942--946",
    abstract = "This paper describes our approach to the SemEval 2017 Task 10: Extracting Keyphrases and Relations from Scientific Publications, specifically to Subtask (B): Classification of identified keyphrases. We explored three different deep learning approaches: a character-level convolutional neural network (CNN), a stacked learner with an MLP meta-classifier, and an attention based Bi-LSTM. From these approaches, we created an ensemble of differently hyper-parameterized systems, achieving a micro-$F_1$-score of 0.63 on the test data. Our approach ranks 2nd (score of 1st placed system: 0.64) out of four according to this official score. However, we erroneously trained 2 out of 3 neural nets (the stacker and the CNN) on only roughly 15{\%} of the full data, namely, the original development set. When trained on the full data (training+development), our ensemble has a micro-$F_{1}$-score of 0.69. Our code is available from \url{https://github.com/UKPLab/semeval2017-scienceie}."
}