PKU_ICL at SemEval-2017 Task 10: Keyphrase Extraction with Model Ensemble and External Knowledge

Proceedings of the 11th International Workshop on Semantic Evaluations (SemEval-2017), pages 934–937, Vancouver, Canada, August 3 - 4, 2017. c⃝2017 Association for Computational Linguistics PKU ICL at SemEval-2017 Task 10: Keyphrase Extraction with Model Ensemble and External Knowledge Liang Wang Key Laboratory of Computational Linguistics Peking University, China intfloat@pku.edu.cn Sujian Li Key Laboratory of Computational Linguistics Peking University, China lisujian@pku.edu.cn Abstract This paper presents a system that partic- ipated in SemEval 2017 Task 10 (subtask A and subtask B): Extracting Keyphrases and Relations from Scientific Publica- tions (Augenstein et al., 2017). Our pro- posed approach utilizes external knowl- edge to enrich feature representation of candidate keyphrase, including Wikipedia, IEEE taxonomy and pre-trained word em- beddings etc. Ensemble of unsupervised models, random forest and linear models are used for candidate keyphrase ranking and keyphrase type classification. Our sys- tem achieves the 3rd place in subtask A and 4th place in subtask B. 1 Introduction Keyphrases summarize the most important aspects of a document. They can be helpful in many areas such as information retrieval, topic modeling and text classification. However, manually labeling keyphrase would be far too time-consuming, and unrealistic especially when dealing with web-scale collection of documents. Therefore, automatic keyphrase extraction has drawn growing interests among NLP research communities for years. For state-of-the-art system on keyphrase extrac- tion, Hasan and Ng(2014) presents a comprehen- sive survey. Their experiments demonstrate that unsupervised approaches including graph-based ranking and topic modeling techniques perform best on News and Blogs dataset. In SemEval 2010 Task 5 (Kim et al., 2010) (Kim et al., 2013), which also aims to tackle the challenge of keyphrase extraction in scientific area, a majority of the participants adopt supervised approaches, and especially the top 2 systems are both super- vised. Thus, in our work, we argue that super- vised approaches can enable combination of rich features, with parameters learned efficiently and automatically, while their unsupervised counter- parts often involve simply designed features and manually fine-tuned hyperparameters. Based on the consideration above, for Se- mEval 2017 Task 10, our system is designed as a supervised one which also explore unsu- pervised techniques as auxiliaries. It involves three steps: candidate generation, keyphrase rank- ing and keyphrase type classification. For can- didate generation, we use chunking-based ap- proach to discover phrases that match a prede- fined part-of-speech pattern. Heuristic rules are manually designed by experience and applied to filter out those phrases which are unlikely to be keyphrases. For keyphrase ranking in subtask A, we use a straightforward regression-based point- wise ranking method. Here, two unsupervised al- gorithms TextRank (Mihalcea and Tarau, 2004) and SGRank (Danesh et al., 2015) are incorpo- rated into random forest by providing their out- put as complementary features. In our experi- ments, we find that stacking linear model upon random forest can provide extra performance gain. For keyphrase type classification in subtask B, we model it as a three-way classification problem, with the same feature set and classifiers used in subtask A. Feature engineering is a critical part …

Content truncated to 500 words