Semantic Searching within National Library of Medicine Databases: NLMPlus

What is the most effective way to search for biomedical information? You can search through a wide range of interfaces within the databases of the NCBI Network: PubMed offers an advanced search form with a variety of limiters and expert command line searching, PubChem offers a chemical structure search option, NCBI BLAST offers genetic sequence searching, and NCBI Entrez offers a federated search portal for all NCBI databases.  One type of search seems to be missing, though.  Where is the semantic search option?

Users now have the option to search biomedical data and services of the National Library of Medicine (NLM) through NLMPlus, a “semantic search engine and biomedical knowledge base application” created by WebLib LLC.  The application was a winner of the NLM Show Off Your Apps Challenge in November 2011, and the president of WebLib confirmed that the company “plans to maintain and enrich the product indefinitely,” according to a recent article from Information Today.

Continue reading to learn about the semantic search features of NLMPlus!

The Databases and Technology Behind NLMPlus

The About Us section of NLMPlus states that over 60 databases are searched by the application:

  • NCBI All Databases (38 databases)
  • TOXNET (15 databases)
  • Medlineplus (1 databases, 5 DB subsets)
  • Drug Information Portal (1 database)
  • ClinicalTrials.gov (1 database)
  • Images from the History of Medicine (1 database)
  • DailyMed (1 database)
  • AIDSInfo (1 database)
  • Dirline (1 database)
A detailed review of the technology used to create NLMPlus is also available on the “About Us” page.  The section lists the following tools that were used to create the semantic search system:
  •  WebLib’s Biomedical Knowledge Base (BKB) – More than four million BKB concepts are automatically generated from trusted biomedical content on the Web.
  • Semantic Indexing of the PubMed Reviews database – NLMplus uses the Solr/Lucene open source enterprise search engine with WebLib’s proprietary semantic indexing technology for the automatic identification of Key Title phrases, Key Abstract phrases, and MESH Qualifier terms. The key phrases are mapped to UMLS concepts and other concepts in the Biomedical Knowledge Base.
  • Semantic Search – The WebLib Semantic Search Engine integrates the Solr/Lucene open source enterprise search engine with WebLib’s Semantic Query Translator and Biomedical Knowledge Base web services.
  • Applied Natural Language Processing Tools and Applications:
    • Part-of-speech Tagger
    • Noun Phrase Chunker
    • Semantic Parser
    • A-Z-Search (Knowledge Search Engine)
    • KnowledgeMap (Biomedical Knowledge Base and UMLS semantic annotation web service)
    • PolySpell (Scientific and Medical SpellChecker)
    • AutoComplete (Query concepts look-ahead display)
  • Federated Search – WebLib’s PolyMeta Distributed Meta-Search and Discovery Engine simultaneously searches 60 NLM databases and on several different NLM search platforms.

Searching on NLMPlus

Despite the complex technology behind the semantic application, the search interface for the site is incredibly simple to use.  From the NLMPlus homepage, the user can enter any term or concept into the search form (e.g. “stent”) and select “Search.”

NLMPlus has a simple keyword search form that includes an auto-suggest function.

The result page will generate lists of Related Concepts, Health Concerns, Tests and Treatments, Drugs and Substances, and Alternative Medicine, which are displayed in expandable lists at the top of the results page.  Mouse over the related concepts in the list to view a definition, related terms, and the option to add the concept to the current query.  Below the expandable lists of related concepts, the results are displayed, with search terms bolded in the excerpts. A magnifying glass icon beside each search results allows the user to view a preview of that result directly in the hit list window.

Lists of related concepts are displayed above the search results.

Above the expandable lists and search form, the user can select to view results according to specific topics, sources, and document types:

  • Health topics – Default search results page, shown in the screenshot above.
  • NLM Databases – This results page displays a chart that breaks down the number of results in each specific NLM database. Selecting a database name will open a link to the search results listed on the database website.

View a breakdown of how many results were found for each database.

  • PubMed Semantic Search – Normal search results display.
  • PubMed Reviews Semantic Search – Normal search results display.
  • Drugs – Normal search results display.
  • PubMed -Normal search results display.
  • News – Normal search results display.
  • Images – Image thumbnails are displayed instead of text excerpts.
  • Videos – Normal search results display.

On the right hand of every results page (except the NLM Databases page), a series of boxes display sample results for all other search types. The user can select the “More” option at the bottom of each box to go directly to that search type’s page of results.

Selecting any result from the result page from will open the link in a new tab of the browser at the original NLM or NCBI database website.

Conclusion

NLMPlus is an excellent alternative to the NCBI Entrez portal for cross-searching multiple NCBI databases at one time.  Users can utilize natural language in their query to retrieve both relevant results and lists of additional concepts to add to their query.  The organization of results into document types (news, images, videos) and by source through the NLM Databases chart can aid the user in quickly identifying which results are relevant.  The preview option also provides a quicker way to judge relevance, since the user doesn’t need to open a separate tab to check each search result.

A feature that would improve the functionality of the application would be the addition of a saved results and concepts list.  Currently, there is no way to select or save results or relevant concepts, so the user must manually copy and past relevant links and concepts into a separate Word document.  The ability to re-rank search results would also add to the functionality of application, since users may want to view the most recent results in the list.  Finally, some type of help-form or guide would add to the utility of the application, since no help features seem to be available for the site.

Despite these issues, I definitely recommend that prior art searchers, especially those in the biomedical field who commonly search on NCBI databases for prior art, try the semantic search features of NLMPlus.  At the very least, a user may find new terms to add to their query.

Are there other semantic search tools that may be useful to biomedical prior art searchers?  Let us know in the comments!

Free Webinar

On January 18, 2012 from 2:00 to 3:00pm EST, Landon IP will be hosting a FREE WEBINAR providing a demo of the features and functionality of Patent Workbench®, a productivity tool that helps you work faster, smarter, and more efficiently whenever you need to review and analyze patent file histories and understand the scope of a patent’s claims.

Patent Workbench™ from Landon IP

This post was contributed by Joelle Mornini. The Intellogist blog is provided for free by Intellogist’s parent company Landon IP, a major provider of patent searches, trademark searches, technical translations, and information retrieval services.

About these ads

4 Responses

  1. Thanks for this great post. There is a certain number of other semantic alternatives for biomedical literature (GoPubMed, etc.) that I have mentioned in my post: How many Medline platfoms on the web?
    http://wp.me/psYf1-pq
    Hope it helps
    Hervé

  2. Thanks Joelle, stay up to date following
    http://scienceintelligence.wordpress.com/tag/medline/

  3. [...] Semantic Searching within National Library of Medicine Databases: NLMPlus (The Intellogist) [...]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 742 other followers

%d bloggers like this: