Query Expansion Using Medical Information Extraction for Improving Information Retrieval in French Medical Domain

Query Expansion Using Medical Information Extraction for Improving Information Retrieval in French Medical Domain

Aicha Ghoulam (Department of Computer Science, University of Oran 1, Ahmed Ben Bella, Oran, Algeria), Fatiha Barigou (Department of Computer Science, University of Oran 1, Ahmed Ben Bella, Oran, Algeria), Ghalem Belalem (Department of Computer Science, University of Oran 1, Ahmed Ben Bella, Oran, Algeria) and Farid Meziane (University of Salford, UK)
Copyright: © 2018 |Pages: 17
DOI: 10.4018/IJIIT.2018070101
OnDemand PDF Download:
No Current Special Offers


This article describes how many users' queries contain references to named entities, and this is particularly true in the medical field. Doctors express their information needs using medical entities as they are element rich with information that helps better target relevant documents. At the same time, many resources have been recognized as a large container of medical entities and relationships between them such as clinical reports; which are medical texts written by doctors. In this article, the authors present a query expansion method that uses medical entities and their semantic relations in the query context based on an external resource in OWL. The goal of this method is to evaluate the effectiveness of an information retrieval system to support doctors in accessing easily relevant information. Experiments on a collection of real clinical reports show that their approach reveals interesting improvements in precision, recall and MAP in medical information retrieval.
Article Preview

1. Introduction And Motivation

With the growing amount of available information in the medical field, accessing useful and relevant biomedical information in real time is becoming of a paramount importance for practitioners and researchers. Indeed, information retrieval systems support users in their daily activities to satisfy their needs. Usually, the user formulates his information need into a query; in return, an information retrieval system (IRS) provides the most relevant documents that satisfy the user query. However, there are many difficulties in developing an effective IRS. One of these difficulties is the word mismatching problem (vocabulary mismatch). The users can express their needs using different words with similar meanings (synonyms) and same words with different meanings (polysemy). According to (Bhatnagar & Pareek, 2014), concepts may be described in different words in user’s queries and/or documents. Many techniques were proposed to solve this problem; one of them is query expansion techniques.

For a long time, query expansion (QE) has been the main motivation for improving the retrieval effectiveness of an IRS. The QE can be performed in different ways such as manual (the user chooses expansion terms), interactive (the user chooses expansion terms from suggestions provided by the system) and automatic (all the process is invisible to the user).

Researchers developed efficient techniques for automatic query expansion; a survey of these techniques is given in (Carpineto & Romano, 2012). Sources for selecting the query expansion terms can be grouped into: document corpus (global, local, relevance feedback-based query expansion), linguistic resources (dictionaries, thesaurus, WordNet ontology, for semantic query expansion) and world knowledge-based resources (Wikipedia). Recently, systems based on query expansion make use of external resources such as ontologies and lexical hierarchies and they have significantly improved their results. In the medical field, most of the medical ontologies such as Medical Subject Heading (MeSH) thesaurus were used to improve medical information retrieval. In (Díaz-Galiano et al., 2009) terms associated with MeSH descriptors are considered as synonyms and used to expand queries, the experiments have shown improvements in the performance of the information retrieval in the medical domain.

In medicine, most of the ontologies have been translated to French to cover the general concepts in the domain. For example, the Health Terminology/Ontology Portal (HeTOP1) contains different medical ontologies translated to French such as Medical Subject Headings (MeSH), Systematized Nomenclature of MEDicine (SNOMED int), National Cancer Institute Thesaurus (NCIt) and so on. However, these ontologies are too large to be processed in a specific system. Thus, a domain-specific ontology is needed to solve this problem. This is what led us to construct our own specific resource in OWL and then use it to expand user’s query.

In this paper, we used an external resource in OWL for query expansion process in the French medical domain. It contains medical entities and relations between them that were extracted from real clinical reports and improved from the work developed in (Ghoulam et al., 2015b). We used medical entities, their synonyms and the semantic relations between them to expand the user’s query. Then, we transformed the expanded query to Boolean query using Boolean operators.

This work is motivated by the fact that, the clinical reports can have a positive impact on the quality of care, patient safety and efficiencies in medical procedures. The doctors need a quality search, to consult and search through these informative reports so that they can make a decision in the shortest time to improve healing. These kinds of medical retrieval systems have become very necessary tools; they will enable researchers to access accurate data and the required information and reducing the time spent by doctors on making decisions about patient’s diseases.

The remainder of this paper is organized as follows; section 2 presents related works on information retrieval systems and query expansion methods in the medical field. In section 3, we describe the proposed system and the external resources we used. Our contribution to retrieve medical reports using query expansion will be discussed in section 4. Section 5 presents the experiments and their results. The paper ends with a conclusion and some suggestions on future developments.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 18: 4 Issues (2022): Forthcoming, Available for Pre-Order
Volume 17: 4 Issues (2021)
Volume 16: 4 Issues (2020)
Volume 15: 4 Issues (2019)
Volume 14: 4 Issues (2018)
Volume 13: 4 Issues (2017)
Volume 12: 4 Issues (2016)
Volume 11: 4 Issues (2015)
Volume 10: 4 Issues (2014)
Volume 9: 4 Issues (2013)
Volume 8: 4 Issues (2012)
Volume 7: 4 Issues (2011)
Volume 6: 4 Issues (2010)
Volume 5: 4 Issues (2009)
Volume 4: 4 Issues (2008)
Volume 3: 4 Issues (2007)
Volume 2: 4 Issues (2006)
Volume 1: 4 Issues (2005)
View Complete Journal Contents Listing