Unsupervised inference of implicit biomedical events using context triggers

Abstract

Background: Event extraction from the biomedical literature is one of the most actively researched areas in biomedical text mining and natural language processing. However, most approaches have focused on events within single sentence boundaries, and have thus paid much less attention to events spanning multiple sentences. The Bacteria-Biotope event (BB-event) subtask presented in BioNLP Shared Task 2016 is one such example; a significant amount of relations between bacteria and biotope span more than one sentence, but existing systems have treated them as false negatives because labeled data is not sufficiently large enough to model a complex reasoning process using supervised learning frameworks.

Results: We present an unsupervised method for inferring cross-sentence events by propagating intra-sentence information to adjacent sentences using context trigger expressions that strongly signal the implicit presence of entities of interest. Such expressions can be collected from a large amount of unlabeled plain text based on simple syntactic constraints, helping to overcome the limitation of relying only on a small number of training examples available. The experimental results demonstrate that our unsupervised system extracts cross-sentence events quite well and outperforms all the state-of-the-art supervised systems when combined with existing methods for intra-sentence event extraction. Moreover, our system is also found effective at detecting long-distance intra-sentence events, compared favorably with existing high-dimensional models such as deep neural networks, without any supervised learning techniques.

Conclusions: Our linguistically motivated inference model is shown to be effective at detecting implicit events that have not been covered by previous work, without relying on training data or curated knowledge bases. Moreover, it also helps to boost the performance of existing systems by allowing them to detect additional cross-sentence events. We believe that the proposed model offers an effective way to infer implicit information beyond sentence boundaries, especially when human-annotated data is not sufficient enough to train a robust supervised system.

Keywords: Bacteria; Biomedical event extraction; Biotope; Cross-sentence relations; Natural language processing; Text mining; Unsupervised inference.

Fig. 1

Screenshot of the PubMed original abstract text for the BB-event document (PMID: 10738994). The BB-event datasets do not contain headers such as OBJECTIVES, DESIGN, and METHODS

Fig. 2

Overview of the proposed model

Fig. 3

Using a trigger pattern to collect a context trigger from training data

Fig. 4

Example of extracting long-range intra-sentence events

Fig. 5

Example of propagating event labels to other pairs of bacteria and locations

Fig. 6

Example of the mapping between context triggers and bacteria mentions, and of linking each trigger to location mentions. Blue-shaded words such as cultures and prevalence are context triggers. Dashed curved arrows are the mappings between context triggers and bacteria mentions within the context window of each bacteria mention. Solid curved arrows connecting context triggers to location mentions are intra-sentence relations between them. Solid vertical arrows on the right indicate the sliding context window (i.e., sentence range) of each of the three bacteria mentions (B1, B2, and B3)