Banner: An Executable Survey of Advances in Biomedical Named Entity Recognition

Abstract

There has been an increasing amount of research on biomedical named entity recognition, the most basic text extraction problem, resulting in significant progress by different research teams around the world. This has created a need for a freely-available, open source system implementing the advances described in the literature. In this paper we present BANNER, an open-source, executable survey of advances in biomedical named entity recognition, intended to serve as a benchmark for the field. BANNER is implemented in Java as a machine-learning system based on conditional random fields and includes a wide survey of the best techniques recently described in the literature. It is designed to maximize domain independence by not employing brittle semantic features or rule-based processing steps, and achieves significantly better performance than existing baseline systems. It is therefore useful to developers as an extensible NER implementation, to researchers as a standard for comparing innovative techniques, and to biologists requiring the ability to find novel entities in large amounts of text.BANNER is available for download at http://banner.sourceforge.net.

Disease Named Entity Recognition by Machine Learning Using Semantic Type of Metathesaurus

“…The IO encoding scheme gives the slightly better F-score than BIO and BIOEW schemes. This is in agreement with the finding in [17] that uses the BioCreative II corpus for gene/protein NER task. In this paper, the IO setting is retained for our experiments.…”

“…Three NER systems for disease recognition using the Biotext corpus and 5 × 2 cross-validation was reported in [17]. Comparing with their results, our semantic type feature based method gives the highest F-score of 56.67 (BANNER: 54.84, ABNER: 53.44, and LingPipe: 51.15).…”

“…The corpus was first pre-processed by tokenization and lemmatization before feature extraction. Following [17], we used feature set consisting of POS, lemma, orthographical and morphological features (patterns for word capitalization, letter and digit combinations, prefixes and suffixes). Numbers were normalized by converting digits to single digit "0".…”

Towards Confidence Estimation for Typed Protein-Protein Relation Extraction

“…It has also spurred a number of (shared) tasks and system competitions of which the best known are the BioNLP Shared Task 1 and the BioCreative challenge 2 . Relevant subtasks include named entity recognition (NER, Leaman and Gonzalez, 2008), entity linking and normalization to unique database identifiers (Zheng et al, 2014), event (EE, Björne and Salakoski, 2015) and relation extraction (RE, Tymoshenko et al, 2012;Airola et al, 2008;Choi, 2016). The overall goal is to identify biomedical entity mentions, disambiguate them w.r.t.…”

“…The main contributions of this paper are: (1) We build a distantly supervised RE pipeline based on BANNER (Leaman and Gonzalez, 2008) for NER, TEES (Björne and Salakoski, 2015) for EE and RE, and GNAT and Gnorm (Hackenberg et al, 2011;Wei et al, 2015) to link protein mentions to the STRING protein interaction database (von Mering et al, 2005), to distantly determine the truth and falsity of the discovered typed protein-protein relations. (2) We define confidence measures for each component of our pipeline and analyze their impact on relation prediction.…”