topic maps applied to pubmed
DESCRIPTION
Topic Maps applied to PubMed. Giovani Rubert Librelotto Mirkos Martins Henrique Machado Franciscan University UNIFRA Brazil. José Carlos Ramalho Pedro Gabriel Ferreira Pedro R. Henriques Department of Informatics University of Minho Portugal. Outline. Motivation PubMed Topic Maps - PowerPoint PPT PresentationTRANSCRIPT
Topic Maps applied to PubMed, Extreme’07, August 2007 1
Topic Maps applied to Topic Maps applied to PubMedPubMed
Giovani Rubert Librelotto Mirkos Martins
Henrique MachadoFranciscan University
UNIFRABrazil
José Carlos RamalhoPedro Gabriel FerreiraPedro R. Henriques
Department of InformaticsUniversity of Minho
Portugal
Topic Maps applied to PubMed, Extreme’07, August 2007 2
Outline
• Motivation
• PubMed
• Topic Maps
• Metamorphosis
• PubMed knowledge representation
• Conclusion
Topic Maps applied to PubMed, Extreme’07, August 2007 3
Motivation
• Write down what you want to find!
• Underline the key concepts
• Consider how best to combine and refine the concepts
Topic Maps applied to PubMed, Extreme’07, August 2007 4
Identify Key Concepts
Is it risky for pregnant women to take antidepressants?
Risky – (Synonyms) Harm, adverse effects
Anti-depressants – Name of drug or drug(s)?
Topic Maps applied to PubMed, Extreme’07, August 2007 5
PubMed
• PubMed is a free search engine that provides very full coverage of the related biomedical sciences, such as biochemistry and cell biology.
• It also offers access to the MEDLINE database with citations and abstracts of biomedical research articles.
Topic Maps applied to PubMed, Extreme’07, August 2007 6
PubMed (2)
• The PubMed core subject is medicine and its related fields.
• It is offered by the United States National Library of Medicine as part of the Entrez information retrieval system.
Topic Maps applied to PubMed, Extreme’07, August 2007 7
PubMed (3)
• The inclusion of an article in PubMed does not endorse the article's contents, as other indexes.
• Nevertheless, many PubMed citations contain links to full text articles which are freely available, often in the PubMed Central digital library.
Topic Maps applied to PubMed, Extreme’07, August 2007 8
MEDLINE
• MEDLINE database covers over 4.900 journals published around the world primarily from 1966 to the present and is composed of more than 17 millions of citations.
Topic Maps applied to PubMed, Extreme’07, August 2007 9
Milestones
• MEDLINE 35 years young– 15.8 million records
• PubMed approximately 10 years old– 17 million records
• MEDLINE database covers over 4.900 journals
Topic Maps applied to PubMed, Extreme’07, August 2007 10
PubMed architecture
Topic Maps applied to PubMed, Extreme’07, August 2007 11
PubMed web site
Topic Maps applied to PubMed, Extreme’07, August 2007 12
Querying PubMed
Topic Maps applied to PubMed, Extreme’07, August 2007 13
Query Results Formats
Summary
Basic Bibliographic Citation (default)
Abstract
Includes abstract if available
Brief
First author, first few words, PMID
Citation
Citation, Abstract & Subject Headings
Topic Maps applied to PubMed, Extreme’07, August 2007 14
Changing Results Format
• Pull down menu on display bar
• Select desired format
• Change occurs automatically
Topic Maps applied to PubMed, Extreme’07, August 2007 15
Information indexed in PubMed
• Subject or journal title
• Title Abbreviation
• NLM ID (NLM'sunique journal identifier)
• ISO abbreviation
• Both the print and electronic International Standard Serial Numbers (pISSN and eISSN).
Topic Maps applied to PubMed, Extreme’07, August 2007 16
PubMed data structure - CFG
MedlineCitation ==> PMID, DateCreated, DateCompleted, Article, MedlineJournalInfo, ChemicalList, CitationSubset, MeshHeadingList
Article ==> Journal, ArticleTitle, Pagination, Abstract, Affiliation, AuthorList, Language, PublicationTypeList
Journal ==> ISSN, JournalIssue, TitleJournalIssue ==> Volume, Issue, PubDatePubDate ==> Year, Month, Day, Hour?, Minute?, Second?MedlineJournalInfo ==> Country, MedlineTA, NlmUniqueIDChemicalList ==> Chemical+Chemical ==> RegistryNumber, NameOfSubstanceMeshHeadingList ==> MeshHeading+MeshHeading ==> DescriptorName, QualifierName?AuthorList ==> Author+Author ==> LastName, ForeName, InitialsPublicationTypeList ==> PublicationType+
Topic Maps applied to PubMed, Extreme’07, August 2007 17
PubMedXML Schema
Topic Maps applied to PubMed, Extreme’07, August 2007 18
Available PubMed Concepts
• Publication Date• Language• Literature Type• Journal
• Author• Gender• Age• Journal Subset
Topic Maps applied to PubMed, Extreme’07, August 2007 19
Available PubMed Association
• Author writes article;• Keyword describes article;• Article was published in an year;• Article is published in a journal;• Article is written in a language;• Article refers to chemical substances; • Author publishes in an year;• Author writes paper in a language;• Journal refers to the keywords;
Topic Maps applied to PubMed, Extreme’07, August 2007 20
What’s the idea?
1.To create an ontology based on PubMed concepts and associations
2.To use that ontology to present query results.
Topic Maps applied to PubMed, Extreme’07, August 2007 21
How?
Structure level
Catalog level
1. Prebuild ontology’s structure level
2. Populate the ontology with query results (occurrences)
Topic Maps applied to PubMed, Extreme’07, August 2007 22
Metamorphosis
Topic Maps applied to PubMed, Extreme’07, August 2007 23
Metamorphosis
Topic Maps applied to PubMed, Extreme’07, August 2007 24
XSTM editor
Topic Maps applied to PubMed, Extreme’07, August 2007 25
Structure level through XSTM
<?xml version="1.0" encoding="ISO-8859-1"?><xstm> <topicTypes> <topicType> <id>Article</id> <instanceOf>Article</instanceOf> <name>Article</name> </topicType> <topicType> <id>Year</id> <instanceOf>Year</instanceOf> <name>Year</name> </topicType> ...
<topics> <topic> <xpath id="PMID” name="Article/ArticleTitle"> //MedlineCitation</xpath> <type>Article</type> <resourceData scope="Abstract"> Article/Abstract </resourceData> <resourceData scope="PMID"> PMID</resourceData> <resourceData scope="Pagination"> Article/Pagination/MedlinePgn</resourceData> </topic>
Topic Maps applied to PubMed, Extreme’07, August 2007 26
Topic Maps applied to PubMed
Topic Maps applied to PubMed, Extreme’07, August 2007 27
New architecture
Metamorphosis
Extracting the base ontology
Qresults intersection
Semantic Browser
Topic Maps applied to PubMed, Extreme’07, August 2007 28
• One of the advantages of this approach is that no new database is created and no redundant data is produced;
• Information is interconnected within a huge knowledge network navigable in any direction.
Topic Maps applied to PubMed, Extreme’07, August 2007 29
Future work
• Integration of Topic Maps and MeSH headings
• Minimizing “false hits” and saving time
• Identifying other useful – but frequently overlooked – features of the PubMed database.
Topic Maps applied to PubMed, Extreme’07, August 2007 30
MeSH
• Medical Subject Headings (MeSH) is a controlled vocabulary designed by the National Library of Medicine to search MEDLINE and other health sciences databases
Topic Maps applied to PubMed, Extreme’07, August 2007 31
MeSH example by Wikipedia
Topic Maps applied to PubMed, Extreme’07, August 2007 32
MeSH qualifiers
• In addition to the descriptor hierarchy, MeSH contains a small number of standard qualifiers (also known as subheadings), which can be added to descriptors to narrow down the topic.
Topic Maps applied to PubMed, Extreme’07, August 2007 33
MeSH qualifiers example
• For example, "Measles" is a descriptor and "epidemiology" is a qualifier; "Measles/epidemiology" describes the subheading of epidemiological articles about Measles.
• The "epidemiology" qualifier can be added to all other disease descriptors. Not all descriptor/qualifier combinations are allowed since some of them may be meaningless. In all there are 83 different qualifiers
Topic Maps applied to PubMed, Extreme’07, August 2007 34
Conclusion
• This paper described the integration of data from PubMed information system using the ontology paradigm, in order to generate an homogeneous view of this resources.
• The proposal uses an environment, called Metamorphosis, for the automatic construction of Topic Maps with data extracted from the various data sources, and a semantic browser to navigate among the information resources.
Topic Maps applied to PubMed, Extreme’07, August 2007 35
Questions?
Short announcement:
XATA2008: Portuguese XML conference, Feb. 2008
• Becoming international
• English will be the language
• We are seeking people for PC commitee...