Using Rule-Based Natural Language, Article Critique Example

Kang, N., Singh, B., Afzal, Z., van Mulligen, E.M., et al. (2013). Using rule-based natural language processing to improve disease normalization in biomedical text. Journal of the American Medical Information Association 20, 876-881.

In this article, the authors discuss the need for a normalization system in regards to computers that extract information and data from unstructured text or that which does not possess a data model and usually contains numbers, dates, and facts in an ambiguous manner. In other words, the computer system does not recognize this type of unstructured text; thus, the authors argue that a normalization system is required in order to “link relevant concepts in a text to sources that contain further information” concerning the concept (Kang, Singh, Afzal, van Mulligen, et al., 2013, p. 876).

As to the methods utilized in this study, the authors made a side-by-side comparison of two popular biomedical concept normalization systems–MetaMap and Peregrine. Their decision to utilize these two systems was due to the fact that both are downloadable from their official websites and are equipped with default configurations and parameters. The overall results of this study showed that MetaMap and Peregrine were two excellent choices, due to scoring 61% and 63% accuracy respectively without the assistance of a NLP module (natural language processing) which generally helps to improve biomedical concept normalization techniques. However, when the NLP module was utilized, the scores increased by more than 10%, a finding which supports the author’s viewpoint that the module is advantageous in relation to biomedical normalization systems (Kang, Singh, Afzal, van Mulligen, et al., 2013, p. 876).

After closely reviewing this fascinating article, several key points stood out. First of all, the authors provide a very clear definition on concept normalization systems, in this case, MetaMap and Peregrine. Both of these systems are dictionary-based, meaning that information and data is alphabetized with a number of definitions and connotations. Interestingly, this system is currently used by the National Institutes of Health which focuses upon medical research in virtually every field of endeavor. Also, MetaMap is based upon what is known as the Unified Medical Language System (UMLS), a sort of complex thesaurus that contains key terms, coding standards, and a host of resources for biomedical information systems, especially electronic health records held by hospitals and clinics. As to the Peregrine system, this too is dictionary-based and acts as a data mining tool by being able to locate concepts via word/term search definitions.

Secondly, the authors clearly define the nature of the natural language processing module which is also dictionary-based and includes annotations or notes that researchers can use to find information and data more quickly. Overall, the NLP module is utilized by researches in the fields of linguistics and computer science and engineering. What is most interesting about this module is that in the near future, it will be able to communicate and interact with humans, such as when a user asks a question and the computer replies with the needed information or data. Thus, this module will be on the cutting edge of electronic cognition abilities.

This article was chosen for several reasons. First of all, it offers forty-three references which can be used for further research on concept normalization systems; secondly, it is a peer-reviewed article, meaning that outside researchers have read it and have found it appropriate for publication. Also, there are two references that seem to be related to this week’s topic–Ref. 31–Tsurucka, Y., McNaught, J., and Ananiadou, S. (2008). Normalizing biomedical terms by minimizing ambiguity and variability. BMC Bicinform 9, S2., and Ref. 4–Bodenreider, O. (2004). The unified medical language system (UMLS): Integrating biomedical terminology. Nucleic Acids Research (4)32, 267-270.

References

Kang, N., Singh, B., Afzal, Z., van Mulligen, E.M., et al. (2013). Using rule-based natural language processing to improve disease normalization in biomedical text. Journal of the American Medical Information Association 20, 876-881.