Identifying non-elliptical entity mentions in a coordinated NP with ellipses

Abstract

Named entities in the biomedical domain are often written using a Noun Phrase (NP) along with a coordinating conjunction such as 'and' and 'or'. In addition, repeated words among named entity mentions are frequently omitted. It is often difficult to identify named entities. Although various Named Entity Recognition (NER) methods have tried to solve this problem, these methods can only deal with relatively simple elliptical patterns in coordinated NPs. We propose a new NER method for identifying non-elliptical entity mentions with simple or complex ellipses using linguistic rules and an entity mention dictionary. The GENIA and CRAFT corpora were used to evaluate the performance of the proposed system. The GENIA corpus was used to evaluate the performance of the system according to the quality of the dictionary. The GENIA corpus comprises 3434 non-elliptical entity mentions in 1585 coordinated NPs with ellipses. The system achieves 92.11% precision, 95.20% recall, and 93.63% F-score in identification of non-elliptical entity mentions in coordinated NPs. The accuracy of the system in resolving simple and complex ellipses is 94.54% and 91.95%, respectively. The CRAFT corpus was used to evaluate the performance of the system under realistic conditions. The system achieved 78.47% precision, 67.10% recall, and 72.34% F-score in coordinated NPs. The performance evaluations of the system show that it efficiently solves the problem caused by ellipses, and improves NER performance. The algorithm is implemented in PHP and the code can be downloaded from https://code.google.com/p/medtextmining/.

tmChem: a high performance approach for chemical named entity recognition and normalization

“…Another issue common in other entity types is mentions containing coordination ellipsis [ 34 ], such as "oleic, linoleic, and palmitic acids." We observed both models having difficulties finding complete coordinations, instead, we often observed the models locating the non-elliptical part of the mention (such as "3-O-alkyl clarithromycin" from the mention "3-hydroxyl, 3-O-acyl and 3-O-alkyl clarithromycin").…”

Interactive Knowledge Discovery and Data Mining in Biomedical Informatics

“…Such data often have an unexpressed informative power, indeed the study of relations and correlations among them allows domain experts to understand the evolution of subtended behaviours or phenomena over time, as recently outlined by [115][116][117][118]. Such data often have an unexpressed informative power, indeed the study of relations and correlations among them allows domain experts to understand the evolution of subtended behaviours or phenomena over time, as recently outlined by [115][116][117][118].…”

“…In addition, such relations expressed over multiple sentences which makes it even more difficult to find them and typically requires co-reference resolution. Higher-level ambiguity such as processing of coordination structures and attachment of modifiers will require more linguistic methods and has been addressed in few systems [116,117] but remains a stumbling block. Some progress on ambiguity, especially term ambiguity, has been made; there exists a disambiguation module in the UMLS MetaMap concept recognition tool as well as other proposed approaches [112,113].…”

SimConcept: A Hybrid Approach for Simplifying Composite Named Entities in Biomedical Text

“…They evaluated their method using the GENIA [33] corpus, obtaining 86% accuracy. Due to the lower performance of this method on complex ellipsis (e.g., “recombinant human nm23- H1, -H2, mouse nm23-M1, and -M2” ), Chae et al, [34] developed a pattern based method, using lexicons to identify the region of each component (i.e., conjunction, conjuncts, and ellipsis antecedent) for each mention. However, these previous studies have focused on only one type of composite mention: mentions with coordination ellipsis.…”