Limits...
Ranking Medical Subject Headings using a factor graph model.

Wei W, Demner-Fushman D, Wang S, Jiang X, Ohno-Machado L - AMIA Jt Summits Transl Sci Proc (2015)

Bottom Line: Recent work demonstrated the feasibility of improving the existing automated Medical Text Indexer (MTI) system, developed at the National Library of Medicine (NLM).Specifically, we developed a graphical model to propagate belief through a citation network to provide robust MeSH main heading (MH) recommendation.Our preliminary results indicate that this approach can reach high Mean Average Precision (MAP) in some scenarios.

View Article: PubMed Central - PubMed

Affiliation: Division of Biomedical Informatics, University of California, San Diego, La Jolla, CA 92093 USA, Email: { w2wei@ucsd.edu , shw070@ucsd.edu , x1jiang@ucsd.edu , lohnomachado@ucsd.edu.

ABSTRACT
Automatically assigning MeSH (Medical Subject Headings) to articles is an active research topic. Recent work demonstrated the feasibility of improving the existing automated Medical Text Indexer (MTI) system, developed at the National Library of Medicine (NLM). Encouraged by this work, we propose a novel data-driven approach that uses semantic distances in the MeSH ontology for automated MeSH assignment. Specifically, we developed a graphical model to propagate belief through a citation network to provide robust MeSH main heading (MH) recommendation. Our preliminary results indicate that this approach can reach high Mean Average Precision (MAP) in some scenarios.

No MeSH data available.


An example of a factor graph with two layers of variable nodes. V1 is a variable node which represents the citing article. V2 to Vn are the variable nodes for cited articles. F1 to Fn are leaf factor nodes and they provide the prior probability distributions on MHs for their adjacent variable nodes. F12 to F1n are intermediary factor nodes.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4525219&req=5

f2-2092403: An example of a factor graph with two layers of variable nodes. V1 is a variable node which represents the citing article. V2 to Vn are the variable nodes for cited articles. F1 to Fn are leaf factor nodes and they provide the prior probability distributions on MHs for their adjacent variable nodes. F12 to F1n are intermediary factor nodes.

Mentions: Our two-layer factor graph model is illustrated in Figure 2.


Ranking Medical Subject Headings using a factor graph model.

Wei W, Demner-Fushman D, Wang S, Jiang X, Ohno-Machado L - AMIA Jt Summits Transl Sci Proc (2015)

An example of a factor graph with two layers of variable nodes. V1 is a variable node which represents the citing article. V2 to Vn are the variable nodes for cited articles. F1 to Fn are leaf factor nodes and they provide the prior probability distributions on MHs for their adjacent variable nodes. F12 to F1n are intermediary factor nodes.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4525219&req=5

f2-2092403: An example of a factor graph with two layers of variable nodes. V1 is a variable node which represents the citing article. V2 to Vn are the variable nodes for cited articles. F1 to Fn are leaf factor nodes and they provide the prior probability distributions on MHs for their adjacent variable nodes. F12 to F1n are intermediary factor nodes.
Mentions: Our two-layer factor graph model is illustrated in Figure 2.

Bottom Line: Recent work demonstrated the feasibility of improving the existing automated Medical Text Indexer (MTI) system, developed at the National Library of Medicine (NLM).Specifically, we developed a graphical model to propagate belief through a citation network to provide robust MeSH main heading (MH) recommendation.Our preliminary results indicate that this approach can reach high Mean Average Precision (MAP) in some scenarios.

View Article: PubMed Central - PubMed

Affiliation: Division of Biomedical Informatics, University of California, San Diego, La Jolla, CA 92093 USA, Email: { w2wei@ucsd.edu , shw070@ucsd.edu , x1jiang@ucsd.edu , lohnomachado@ucsd.edu.

ABSTRACT
Automatically assigning MeSH (Medical Subject Headings) to articles is an active research topic. Recent work demonstrated the feasibility of improving the existing automated Medical Text Indexer (MTI) system, developed at the National Library of Medicine (NLM). Encouraged by this work, we propose a novel data-driven approach that uses semantic distances in the MeSH ontology for automated MeSH assignment. Specifically, we developed a graphical model to propagate belief through a citation network to provide robust MeSH main heading (MH) recommendation. Our preliminary results indicate that this approach can reach high Mean Average Precision (MAP) in some scenarios.

No MeSH data available.