SZTERGAK : Feature Engineering for Keyphrase Extraction

Proceedings of the 5th International Workshop on Semantic Evaluation, ACL 2010, pages 186–189, Uppsala, Sweden, 15-16 July 2010. c⃝2010 Association for Computational Linguistics SZTERGAK : Feature Engineering for Keyphrase Extraction G´abor Berend Department of Informatics University of Szeged 2. ´Arp´ad t´er Szeged, H-6720, Hungary berendg@inf.u-szeged.hu Rich´ard Farkas Hungarian Academy of Sciences 103. Tisza Lajos k¨or´ut Szeged, H-6720, Hungary rfarkas@inf.u-szeged.hu Abstract Automatically assigning keyphrases to documents has a great variety of applica- tions. Here we focus on the keyphrase extraction of scientific publications and present a novel set of features for the su- pervised learning of keyphraseness. Al- though these features are intended for ex- tracting keyphrases from scientific papers, because of their generality and robust- ness, they should have uses in other do- mains as well. With the help of these fea- tures SZTERGAK achieved top results on the SemEval-2 shared task on Automatic Keyphrase Extraction from Scientific Arti- cles and exceeded its baseline by 10%. 1 Introduction Keyphrases summarize the content of documents with the most important phrases. They can be valuable in many application areas, ranging from information retrieval to topic detection. However, since manually assigned keyphrases are rarely pro- vided and creating them by hand would be costly and time-consuming, their automatic generation is of great interest nowadays. Recent state-of- the-art systems treat this kind of task as a super- vised learning task, in which phrases of a docu- ment should be classified with respect to their key phrase characteristics based on manually labeled corpora and various feature values. This paper focuses on the task of keyphrase ex- traction from scientific papers and we shall intro- duce new features that can significantly improve the overall performance. Although the experimen- tal results presented here are solely based on sci- entific articles, due to the robustness and univer- sality of the features, our approach is expected to achieve good results when applied on other do- mains as well. 2 Related work In keyphrase extraction tasks, phrases are ex- tracted from one document that are the most char- acteristic of its content (Liu et al., 2009; Wit- ten et al., 1999). In these approaches keyphrase extraction is treated as a classification task, in which certain n-grams of a specific document act as keyphrase candidates, and the task is to classify them as proper keyphrases or not. While Frank et al. (1999) exploited domain spe- cific knowledge to improve the quality of auto- matic tagging, others like Liu et al. (2009) analyze term co-occurence graphs. It was Nguyen and Kan (2007) who dealt with the special characteristics of scientific papers and introduced the state-of-the- art feature set to keyphrase extraction tasks. Here we will follow a similar approach and make sig- nificant improvements by the introduction of novel features. 3 The SZTERGAK system The SZTERGAK framework treats the reproduc- tion of reader-assigned keyphrases as a supervised learning task. In our setting a restricted set of to- ken sequences extracted from the documents was used as classification instances. These instances were ranked …

Content truncated to 500 words