Scientific Information Extraction with Semi-supervised Neural Tagging

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2641–2651 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Scientific Information Extraction with Semi-supervised Neural Tagging Yi Luan Mari Ostendorf Hannaneh Hajishirzi Department of Electrical Engineering, University of Washington {luanyi, ostendor, hannaneh}@uw.edu Abstract This paper addresses the problem of ex- tracting keyphrases from scientific articles and categorizing them as corresponding to a task, process, or material. We cast the problem as sequence tagging and intro- duce semi-supervised methods to a neu- ral tagging model, which builds on re- cent advances in named entity recognition. Since annotated training data is scarce in this domain, we introduce a graph-based semi-supervised algorithm together with a data selection scheme to leverage unanno- tated articles. Both inductive and trans- ductive semi-supervised learning strate- gies outperform state-of-the-art informa- tion extraction performance on the 2017 SemEval Task 10 ScienceIE task. 1 Introduction As a research community grows, more and more papers are published each year. As a result there is increasing demand for improved methods for finding relevant papers and automatically under- standing the key ideas in those papers. However, due to the large variety of domains and extremely limited annotated resources, there has been rel- atively little work on scientific information ex- traction. Previous research has focused on unsu- pervised approaches such as bootstrapping (Gupta and Manning, 2011; Tsai et al., 2013), where hand-designed templates are used to extract sci- entific keyphrases, and more templates are added through bootstrapping. Very recently a new challenge on Scientific Information Extraction (ScienceIE) (Augenstein et al., 2017)1 provides a dataset consisting of 500 1SemEval (Task 10)https://scienceie.github. io/index.html Computer Science: This paper addresses the task of [named en- tity recognition]Task, using [conditional random fields]Process. Our method is evlauated on the [ConLL NER Corpus]Material. Physics: [Local field effects] Process on spontaneous emission rates within [nanostructure photonics material]Material for example are familiar, and have been well used. Material Science: The [Kelvin probe force microscopy technique] Process allows [detection of local EWF]Task be- tween an [atomic force micorscopy]Material and [metal surface]Material. Figure 1: Annotated ScienceIE examples. scientific paragraphs with keyphrase annotations for three categories: TASK, PROCESS, MATERIAL across three scientific domains, Computer Science (CS), Material Science (MS), and Physics (Phy), as in Figure 1. This dataset enables the use of more advanced approaches such as neural network (NN) models. To that end, we cast the keyphrase extraction task as a sequence tagging problem, and build on recent progress in another informa- tion extraction task: Named Entity Recognition (NER) (Lample et al., 2016; Peng and Dredze, 2015). Like named entities, keyphrases can be identified by their linguistic context, e.g. re- searchers ”use” methods. In addition, keyphrases can be associated with different categories in dif- ferent contexts. For example, ‘semantic parsing’ can be labeled as a TASK in one article and as a PROCESS in another. Scientific keyphrases dif- fer in that they can include both noun phrases and verb phrases and in that non-standard “words” (equations, chemical compounds, references) can …

Content truncated to 500 words