No abstract

No abstract

Proper understanding of textual data requires the exploitation and integration of unstructured and heterogeneous clinical sources, healthcare records or scientific literature, which are fundamental aspects in clinical and translational research. The determination of semantic similarity between word pairs is an important component of text understanding that enables the processing, classification and structuring of textual resources. In the past, several approaches for assessing word similarity by exploiting different knowledge sources (ontologies, thesauri, domain corpora, etc.) have been proposed. Some of these measures have been adapted to the biomedical field by incorporating domain information extracted from clinical data or from medical ontologies (such as MeSH or SNOMED CT). In this paper, these approaches are introduced and analyzed in order to determine their advantages and limitations with respect to the considered knowledge bases. After that, a new measure based on the exploitation of the taxonomical structure of a biomedical ontology is proposed. Using SNOMED CT as the input ontology, the accuracy of our proposal is evaluated and compared against other approaches according to a standard benchmark of manually ranked medical terms. The correlation between the results of the evaluated measures and the human experts' ratings shows that our proposal outperforms most of the previous measures avoiding, at the same time, some of their limitations.

Semantic similarity estimation is an important component of analysing natural language resources like clinical records. Proper understanding of concept semantics allows for improved use and integration of heterogeneous clinical sources as well as higher information retrieval accuracy. Semantic similarity has been the focus of much research, which has led to the definition of heterogeneous measures using different theoretical principles and knowledge resources in a variety of contexts and application domains. In this paper, we study several of these measures, in addition to other similarity coefficients (not necessarily framed in a semantic context) that may be useful in determining the similarity of sets of terms. In order to make them easier to interpret and improve their applicability and accuracy, we propose a framework grounded in information theory that allows the measures studied to be uniformly redefined. Our framework is based on approximating concept semantics in terms of Information Content (IC). We also propose computing IC in a scalable and efficient manner from the taxonomical knowledge modelled in biomedical ontologies. As a result, new semantic similarity measures expressed in terms of concept Information Content are presented. These measures are evaluated and compared to related works using a benchmark of medical terms and a standard biomedical ontology. We found that an information-theoretical redefinition of well-known semantic measures and similarity coefficients, and an intrinsic estimation of concept IC result in noticeable improvements in their accuracy.