Open Domain Web Keyphrase Extraction Beyond Language Modeling

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 5175–5184, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 5175 Open Domain Web Keyphrase Extraction Beyond Language Modeling Lee Xiong Chuan Hu Chenyan Xiong Daniel Campos Arnold Overwijk Microsoft AI and Research Redmond, WA 98052, USA {lexion, chuah, cxiong, dacamp, arnoldo}@microsoft.com Abstract This paper studies keyphrase extraction in real-world scenarios where documents are from diverse domains and have variant con- tent quality. We curate and release OpenKP, a large scale open domain keyphrase extraction dataset with near one hundred thousand web documents and expert keyphrase annotations. To handle the variations of domain and con- tent quality, we develop BLING-KPE, a neu- ral keyphrase extraction model that goes be- yond language understanding using visual pre- sentations of documents and weak supervision from search queries. Experimental results on OpenKP confirm the effectiveness of BLING- KPE and the contributions of its neural archi- tecture, visual features, and search log weak supervision. Zero-shot evaluations on DUC- 2001 demonstrate the improved generalization ability of learning from the open domain data compared to a specific domain. 1 Introduction Automatically extracting keyphrases that are salient to the document meanings is an essential step to semantic document understanding. An effective keyphrase extraction (KPE) system can benefit a wide range of natural language process- ing and information retrieval tasks (Turney, 2001; Hasan and Ng, 2014). Recent neural methods formulate the task as a document-to-keyphrase sequence-to-sequence task. These neural KPE models have shown promising results compared to previous systems (Chen et al., 2018; Meng et al., 2017; Ye and Wang, 2018). Noticeably, the recent progress in neural KPE is mostly observed in documents originating from the scientific domain (Meng et al., 2017; Augen- stein et al., 2017). Perhaps because the scientific domain has sufficient training data for these neu- ral methods: Authors are in the practice of as- signing keyphrases to their publications. In real- world scenarios, most potential applications of KPE deal with diverse documents originating from sparse sources that are rather different from scien- tific papers. They often include a much diverse document structure and reside in various domains whose contents target much wider audiences than scientists. It is unclear how well the neural meth- ods trained in the scientific domain generalize to other domains and in real-world scenarios. This paper focuses on the task of open domain web keyphrase extraction, which targets KPE for web documents without any restriction of the domain, quality, nor content of the documents. We curate and release a large scale open do- main KPE dataset, OpenKP, which includes about one hundred thousand web documents with ex- pert keyphrase annotations.1 The web documents are randomly sampled from the English fraction of a large web corpus and reflect the characteris- tics of typical web pages, with large variation in their domains and content qualities. To the best of our knowledge, this will be the first publicly avail- able …

Content truncated to 500 words