IMoJIE: Iterative Memory-Based Joint Open Information Extraction

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5871–5886 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 5871 IMOJIE: Iterative Memory-Based Joint Open Information Extraction Keshav Kolluru1, Samarth Aggarwal1, Vipul Rathore1, Mausam1, and Soumen Chakrabarti2 1 Indian Institute of Technology Delhi keshav.kolluru@gmail.com, samarth.aggarwal.2510@gmail.com rathorevipul28@gmail.com, mausam@cse.iitd.ac.in 2 Indian Institute of Technology Bombay soumen@cse.iitb.ac.in Abstract While traditional systems for Open Informa- tion Extraction were statistical and rule-based, recently neural models have been introduced for the task. Our work builds upon CopyAt- tention, a sequence generation OpenIE model (Cui et al., 2018). Our analysis reveals that CopyAttention produces a constant number of extractions per sentence, and its extracted tu- ples often express redundant information. We present IMOJIE, an extension to Copy- Attention, which produces the next extraction conditioned on all previously extracted tuples. This approach overcomes both shortcomings of CopyAttention, resulting in a variable num- ber of diverse extractions per sentence. We train IMOJIE on training data bootstrapped from extractions of several non-neural sys- tems, which have been automatically filtered to reduce redundancy and noise. IMOJIE out- performs CopyAttention by about 18 F1 pts, and a BERT-based strong baseline by 2 F1 pts, establishing a new state of the art for the task. 1 Introduction Extracting structured information from unstruc- tured text has been a key research area within NLP. The paradigm of Open Information Extrac- tion (OpenIE) (Banko et al., 2007) uses an open vo- cabulary to convert natural text to semi-structured representations, by extracting a set of (subject, rela- tion, object) tuples. OpenIE has found wide use in many downstream NLP tasks (Mausam, 2016) like multi-document question answering and summa- rization (Fan et al., 2019), event schema induction (Balasubramanian et al., 2013) and word embed- ding generation (Stanovsky et al., 2015). Traditional OpenIE systems are statistical or rule-based. They are largely unsupervised in nature, or bootstrapped from extractions made by earlier systems. They often consist of several components like POS tagging, and syntactic parsing. To bypass error accumulation in such pipelines, end-to-end neural systems have been proposed recently. Recent neural OpenIE methods belong to two categories: sequence labeling, e.g., RnnOIE (Stanovsky et al., 2018) and sequence generation, e.g., CopyAttention (Cui et al., 2018). In princi- ple, generation is more powerful because it can introduce auxiliary words or change word order. However, our analysis of CopyAttention reveals that it suffers from two drawbacks. First, it does not naturally adapt the number of extractions to the length or complexity of the input sentence. Second, it is susceptible to stuttering: extraction of multiple triples bearing redundant information. These limitations arise because its decoder has no explicit mechanism to remember what parts of the sentence have already been ‘consumed’ or what triples have already been generated. Its decoder uses a fixed-size beam for inference. However, beam search can only ensure that the extractions are not exact duplicates. In response, we design the first neural OpenIE system that uses sequential decoding of tuples con- ditioned on previous tuples. We …

Content truncated to 500 words