Complex event extraction at PubMed scale

Abstract

Motivation: There has recently been a notable shift in biomedical information extraction (IE) from relation models toward the more expressive event model, facilitated by the maturation of basic tools for biomedical text analysis and the availability of manually annotated resources. The event model allows detailed representation of complex natural language statements and can support a number of advanced text mining applications ranging from semantic search to pathway extraction. A recent collaborative evaluation demonstrated the potential of event extraction systems, yet there have so far been no studies of the generalization ability of the systems nor the feasibility of large-scale extraction.

Results: This study considers event-based IE at PubMed scale. We introduce a system combining publicly available, state-of-the-art methods for domain parsing, named entity recognition and event extraction, and test the system on a representative 1% sample of all PubMed citations. We present the first evaluation of the generalization performance of event extraction systems to this scale and show that despite its computational complexity, event extraction from the entire PubMed is feasible. We further illustrate the value of the extraction approach through a number of analyses of the extracted information.

Availability: The event detection system and extracted data are open source licensed and available at http://bionlp.utu.fi/.

Fig. 1.

Event extraction. A multiphased system is used to generate an event graph, a formal representation for the semantic content of the sentence. Before event detection, sentences are parsed (A) to generate a suitable syntactic graph to be used in detecting semantic relationships. Event detection starts with identification of named entities (B) with BANNER (parses are not used at this step). Once named entities have been identified, the trigger detector (C) uses them and the parse for predicting triggers, words which define potential events. The edge detector (D) predicts relationship edges (event arguments) between triggers and named entities. Finally, the resulting semantic graph is divided into individual events by (E) duplicating trigger nodes and regrouping argument edges.

Fig. 2.

Total number of citations and citations with tagged gene/protein mentions and events in the sample by year.

Fig. 3.

Number of citations with tagged mentions of insulin, IgG and TNF-alpha (normalized for capitalization and hyphenization), as well as extracted events of these proteins. The counts are cumulative for every five years to smooth the curves.

Fig. 4.

Extracted event network around interleukin-4. This graph shows a subset of the predicted event network, including only named entities with at least 50 extracted instances. The round event nodes are (P)ositive regulation, (N)egative regulation, (R)egulation, gene (E)xpression, (B)inding, p(H)osphorylation and (L)ocalization. For clarity, single-argument events (E, B, H and L) are displayed only when they also act as arguments of regulation events.