The NLM Indexing Initiative.

The objective of NLM's Indexing Initiative (IND) is to investigate methods whereby automated indexing methods partially or completely substitute for current indexing practices. The project will be considered a success if methods can be designed and implemented that result in retrieval performance that is equal to or better than the retrieval performance of systems based principally on humanly assigned index terms. We describe the current state of the project and discuss our plans for the future.

Argument identification for arterial branching predications asserted in cardiac catheterization reports.

The language describing coronary vasculature provides a suitable paradigm for research in semantic interpretation of anatomical text. As a pilot project we investigate the possibility of highly accurate retrieval of arterial branching relationships asserted in cardiac catheterization reports. Our methodology relies on the cooperation of underspecified linguistic analysis and structured domain knowledge. The satisfactory results of formal evaluation on both a training and testing set support the promise of this approach.

EDGAR: extraction of drugs, genes and relations from the biomedical literature.

EDGAR (Extraction of Drugs, Genes and Relations) is a natural language processing system that extracts information about drugs and genes relevant to cancer from the biomedical literature. This automatically extracted information has remarkable potential to facilitate computational analysis in the molecular biology of cancer, and the technology is straightforwardly generalizable to many areas of biomedicine. This paper reports on the mechanisms for automatically generating such assertions and on a simple application, conceptual clustering of documents. The system uses a stochastic part of speech tagger, generates an underspecified syntactic parse and then uses semantic and pragmatic information to construct its assertions. The system builds on two important existing resources: the MEDLINE database of biomedical citations and abstracts and the Unified Medical Language System, which provides syntactic and semantic information about the terms found in biomedical abstracts.

Mining molecular binding terminology from biomedical text.

Automatic access to information regarding macromolecular binding relationships would provide a valuable resource to the biomedical community. We report on a pilot project to mine such information from the molecular biology literature. The program being developed takes advantage of natural language processing techniques and is supported by two repositories of biomolecular knowledge. A formative evaluation has been conducted on a subset of MEDLINE abstracts.

Identification of anatomical terminology in medical text.

We report on an experiment to use the natural language processing tools being developed in the SPECIALIST system to accurately identify terminology associated with the coronary arteries as expressed in coronary catheterization reports. The ultimate goal is to map from any anatomically-oriented medical text to online images, using the UMLS as an intermediate knowledge source. We describe some of the problems encountered when processing coronary artery terminology and report on the results of a formative evaluation of a tool for addressing these problems.

Evaluating lexical variant generation to improve information retrieval.

Techniques for managing lexical variation constitute an integral part of information retrieval systems. We report on a series of experiments aimed at evaluating LVG, a lexical variant management tool which addresses the particular problems involved in matching health related vocabularies to concepts in the Unified Medical Language System (UMLS) Metathesaurus. Experiments conducted on data from the Large Scale Vocabulary Test indicate the effectiveness of this approach to managing biomedical information.

Automatic semantic interpretation of anatomic spatial relationships in clinical text.

A set of semantic interpretation rules to link the syntax and semantics of locative relationships among anatomic entities was developed and implemented in a natural language processing system. Two experiments assessed the ability of the system to identify and characterize physico-spatial relationships in coronary angiography reports. Branching relationships were by far the most common observed (75%), followed by PATH (20%) and PART/WHOLE relationships. Recall and precision scores were 0.78 and 0.67 overall, suggesting the viability of this approach in semantic processing of clinical text.

Query expansion using the UMLS Metathesaurus.

Recent work has demonstrated the importance of query expansion for improving retrieval effectiveness when applying statistically-based systems to MEDLINE citations. The research has suggested the use of retrieval feedback for enhancing the original text of users' queries. As an alternative method of query expansion, we propose the use of the MetaMap program for associating UMLS Metathesaurus concepts with the original query. Our experiments show that query expansion based on MetaMap compares favorably with retrieval feedback. We conclude that the optimal strategy would be to combine the two techniques.

Finding the findings: identification of findings in medical literature using restricted natural language processing.

The ability to search the biomedical literature based on findings would provide enhanced access to information. We describe a computer program called FINDX which relies on the UMLS Metathesuaurus and restricted natural language processing to identify findings in free text. Such identification can serve as a filtering mechanism while selecting relevant papers. After discussing the salient characteristics of findings on which FINDX depends, we report on the results of an experiment in which we tested the program on a set of MEDLINE abstracts pertaining to the diagnosis of Parkinson Disease.

Ambiguity resolution while mapping free text to the UMLS Metathesaurus.

We propose a method for resolving ambiguities encountered when mapping free text to the UMLS Metathesaurus. Much of the research in medical informatics involves the manipulation of free text. The Metathesaurus contains extensive information which supports solutions to problems encountered while processing such text. After discussing the process of mapping free text to the Metathesaurus and describing the ambiguities which are often the result of such mapping, we provide examples of rules designed to eliminate mapping ambiguities. These rules refer to the context in which the ambiguity occurs and crucially depend on semantic types obtained from the Metathesaurus. We have conducted a preliminary test of the methodology and the results obtained indicate that the rules successfully resolve ambiguity around 80% of the time.

UMLS knowledge for biomedical language processing.

This paper describes efforts to provide access to the free text in biomedical databases. The focus of the effort is the development of SPECIALIST, an experimental natural language processing system for the biomedical domain. The system includes a broad coverage parser supported by a large lexicon, modules that provide access to the extensive Unified Medical Language System (UMLS) Knowledge Sources, and a retrieval module that permits experiments in information retrieval. The UMLS Metathesaurus and Semantic Network provide a rich source of biomedical concepts and their interrelationships. Investigations have been conducted to determine the type of information required to effect a map between the language of queries and the language of relevant documents. Mappings are never straightforward and often involve multiple inferences.

Semantic processing in information retrieval.

Intuition suggests that one way to enhance the information retrieval process would be the use of phrases to characterize the contents of text. A number of researchers, however, have noted that phrases alone do not improve retrieval effectiveness. In this paper we briefly review the use of phrases in information retrieval and then suggest extensions to this paradigm using semantic information. We claim that semantic processing, which can be viewed as expressing relations between the concepts represented by phrases, will in fact enhance retrieval effectiveness. The availability of the UMLS domain model, which we exploit extensively, significantly contributes to the feasibility of this processing.