Query Expansion Using Medical Subject Headings Terms in the Biomedical Documents

Thesprasith, Ornuma; Jaruskulchai, Chuleerat

doi:10.1007/978-3-319-05476-6_10

Cited by 12 publications

(8 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our best performance in the official challenge results is achieved using query expansion limited to five MeSH terms per query token with a relatively light weighting, and yields an overall improvement in infNDCG of 11% over the baseline system. This is approximately near the middle of the range of improvement seen in other studies using vocabulary-based query expansion techniques ( 16 , 18 , 24 ).…”

Section: Discussionsupporting

confidence: 49%

Query expansion using MeSH terms for dataset retrieval: OHSU at the bioCADDIE 2016 dataset retrieval challenge

2017

View full text Add to dashboard Cite

Scientific data are being generated at an ever-increasing rate. The Biomedical and Healthcare Data Discovery Index Ecosystem (bioCADDIE) is an NIH-funded Data Discovery Index that aims to provide a platform for researchers to locate, retrieve, and share research datasets. The bioCADDIE 2016 Dataset Retrieval Challenge was held to identify the most effective dataset retrieval methods. We aimed to assess the value of Medical Subject Heading (MeSH) term-based query expansion to improve retrieval. Our system, based on the open-source search engine, Elasticsearch, expands queries by identifying synonyms from the MeSH vocabulary and adding these to the original query. The number and relative weighting of MeSH terms is variable. The top 1000 search results for the 15 challenge queries were submitted for evaluation. After the challenge, we performed additional runs to determine the optimal number of MeSH terms and weighting. Our best overall score used five MeSH terms with a 1:5 terms:words weighting ratio, achieving an inferred normalized distributed cumulative gain (infNDCG) of 0.445, which was the third highest score among the 10 research groups who participated in the challenge. Further testing revealed our initial combination of MeSH terms and weighting yielded the best overall performance. Scores varied considerably between queries as well as with different variations of MeSH terms and weights. Query expansion using MeSH terms can enhance search relevance of biomedical datasets. High variability between queries and system variables suggest room for improvement and directions for further research. Database URL: https://biocaddie.org/benchmark-data

show abstract

Section: Discussionsupporting

confidence: 49%

Query expansion using MeSH terms for dataset retrieval: OHSU at the bioCADDIE 2016 dataset retrieval challenge

2017

View full text Add to dashboard Cite

show abstract

“…Queries in the dataset are, on average, 14 terms long, which is much shorter than the queries considered in this article (80 terms). After its introduction, the OHSUMED collection has been extensively used to evaluate classification (e.g., Genkin, Lewis, & Madigan, ; Han & Karypis, ; Xu & Li, ), learning to rank (e.g., Cao et al, ; Duh & Kirchhoff, ; Liu, Xu, Qin, Xiong, & Li, ), and query reformulation (Abdou & Savoy, ; Dong, Srimani, & Wang, ; Haveliwala, ; Hersh, Price, & Donohoe, ; Jalali & Borujerdi, ; Liu & Chu, ; Srinivasan, ; Thesprasith & Jaruskulchai, ). Works in the latter group are the most similar to our systems; they can be further partitioned based on the approach used: ontology‐based reformulation, Pseudo Relevance Feedback (PRF), and a combination of the two.…”

Section: Related Workmentioning

confidence: 99%

“…At query time, terms in the query are mapped to UMLS concepts; highly ranked concepts related to query concepts are then used for query expansion. Finally, Thesprasith and Jaruskulchai () introduced RABAM‐PRF, a variant of PRF that ranks MeSH terms found in the top documents and uses them for query expansion.…”

Section: Related Workmentioning

confidence: 99%

Learning to reformulate long queries for clinical decision support

Soldaini

Yates

Goharian

2017

Asso for Info Science & Tech

View full text Add to dashboard Cite

The large volume of biomedical literature poses a serious problem for medical professionals, who are often struggling to keep current with it. At the same time, many health providers consider knowledge of the latest literature in their field a key component for successful clinical practice. In this work, we introduce two systems designed to help retrieving medical literature. Both receive a long, discursive clinical note as input query, and return highly relevant literature that could be used in support of clinical practice. The first system is an improved version of a method previously proposed by the authors; it combines pseudo relevance feedback and a domain-specific term filter to reformulate the query. The second is an approach that uses a deep neural network to reformulate a clinical note. Both approaches were evaluated on the 2014 and 2015 TREC CDS datasets; in our tests, they outperform the previously proposed method by up to 28% in inferred NDCG; furthermore, they are competitive with the state of the art, achieving up to 8% improvement in inferred NDCG.

show abstract

“…Several works explore the query expansion strategy in the medical domain to improve the precision and recall of queries. Zhu et al [20], and more recently Thesprasith and Jaruskulchai [21], first identify medical terms in the query using a basic lexical tool and match them to MeSH ontology concepts. Then, they expand the found concepts by adding UMLS co-concepts, i.e.…”

Section: State Of the Artmentioning

confidence: 99%

Ontology-Based Search of Genomic Metadata

Fernández

Lenzerini

Masseroli

et al. 2016

IEEE/ACM Trans. Comput. Biol. and Bioinf.

View full text Add to dashboard Cite

The Encyclopedia of DNA Elements (ENCODE) is a huge and still expanding public repository of more than 4,000 experiments and 25,000 data files, assembled by a large international consortium since 2007; unknown biological knowledge can be extracted from these huge and largely unexplored data, leading to data-driven genomic, transcriptomic, and epigenomic discoveries. Yet, search of relevant datasets for knowledge discovery is limitedly supported: metadata describing ENCODE datasets are quite simple and incomplete, and not described by a coherent underlying ontology. Here, we show how to overcome this limitation, by adopting an ENCODE metadata searching approach which uses high-quality ontological knowledge and state-of-the-art indexing technologies. Specifically, we developed S.O.S. GeM (http://www.bioinformatics.deib.polimi.it/SOSGeM/), a system supporting effective semantic search and retrieval of ENCODE datasets. First, we constructed a Semantic Knowledge Base by starting with concepts extracted from ENCODE metadata, matched to and expanded on biomedical ontologies integrated in the well-established Unified Medical Language System. We prove that this inference method is sound and complete. Then, we leveraged the Semantic Knowledge Base to semantically search ENCODE data from arbitrary biologists' queries. This allows correctly finding more datasets than those extracted by a purely syntactic search, as supported by the other available systems. We empirically show the relevance of found datasets to the biologists' queries.

show abstract

Query Expansion Using Medical Subject Headings Terms in the Biomedical Documents

Cited by 12 publications

References 9 publications

Query expansion using MeSH terms for dataset retrieval: OHSU at the bioCADDIE 2016 dataset retrieval challenge

Query expansion using MeSH terms for dataset retrieval: OHSU at the bioCADDIE 2016 dataset retrieval challenge

Learning to reformulate long queries for clinical decision support

Ontology-Based Search of Genomic Metadata

Contact Info

Product

Resources

About