MiSearch works with NCBI Entrez and your history of browsing to build a profile of your areas of interest, and uses this information to rank citations likely to be of most most information to you at the top of the list.
Getting started
Getting started is easy. Just launch the search window (http://misearch.ncibi.org) and follow this example. Your first search:
- type an identifier into the "Profile" text box
- type your query into the "Query" text box (for example, "gab2")
- click "Submit"
Your query will be sent to NCBI, and the results will be displayed in your browser. Now click on the checkboxes or use the underlined PMID to view articles that seem to be of interest to you. When you click on the hyperlinks, the browser will open a new window or tab with the NCBI Entrez page on each citation you click. Simultaneously, these citations are being added to your profile on MiSearch. The checkboxes allow you to add or remove a citation from your profile.
On the left you will also see lists of relevant authors, MeSH terms, and substances related to your query (click on show/hide to show). Click the 
to add the term to your query and refine your search results. Clicking on a MeSH term link will take you to the NCIBI Gene2MeSH browser and show the top genes associated with that MeSH term.
- After you have added a few articles to your profile and/or refined your query with relevant authors, terms, or substances, repeat your initial search (just click "Submit" again).
The citations that you selected as well as other articles with similar features will be ranked at the top of the result set. As you continue to browse and view articles of interest, the system will build a more extensive and refined profile of your interest areas and should do a better and better job of ranking articles according to your interests.
It is best to start with a query that you expect will return a dozen or so citations of interest; queries returning hundreds of citations can take a long time to process.
Background
MiSearch uses a classification algorithm based on MeSH term, substance names and author names associated with citations. Two sets are defined: one is the set of articles you have previously clicked on to view; the other is all of PubMed. For each citation in the retrieval set, the algorithm calculates the likelihood that the citation is a member of these two sets. Article having the highest likelihood of belonging to the set of articles you have viewed are ranked at the top of the list.
MiSearch uses an information based metric to rank terms and authors: It = fqt * log( fqt/ftdb) where fqt is the frequency of documents indexed with term t in the query result set and ftdb is the frequency of documents indexed with term t in the full database. This metric avoids terms like "human" that may be frequent in the query set but not particularly more so than in the database as a whole.
The "User" field is used as an identifier to track usage. If you do not provide a name, the IP address of your request will be used as a default. If you know you will be doing searches for different tasks with different subject areas, feel free to define a "User" for each task.
Status
MiSearch is updated daily with the latest PubMed data.
Privacy
MiSearch records both your queries and the citations that you click on to view. We will not share this information with third parties, but we may use this data to refine and improve MiSearch and other NCIBI applications.
Data Usage Policy
Access to this website is provided free of charge. Permission is granted to use this software and data internally only, so long as no fee is charged, usage of this website is cited in any resulting publications involving results from such use, and so long as the name of the University of Michigan is not used in any advertising or publicity pertaining to such use without specific, written prior authorization. Permission to redistribute this data in any form is specifically not granted.
Publications
States DJ, Ade AS, Wright ZC, Bookvich AV, Athey BD. MiSearch Adaptive PubMed Search Tool. Bioinformatics 2008, Mar 11; Accepted, In Press. [PubMed][PDF].
A note on the MiSearch name
Independent of this project, another adaptive search tool also called miSearch has been developed by Dr. Susan Gauch at the University of Kansas. That tool reranks Google searches based on user behavior and is available at http://www.ittc.ku.edu/~mirco/demo/search.php.
Acknowledgements
The MiSearch tools rely extensively on resources made available by the National Center for Biotechnology Information (NCBI). The statistics and algorithms supporting MiSearch were developed by David J. States supported in part by a grants from the National Library of Medicine (R01 LM008106) and National Center for Research Resources for the National Center for Pathways and Proteomics (P41 RR018627). The infrastructure for MiSearch is provided by the National Center for Integrative Biomedical Informatics and is supported by a grant from the National Institute on Drug Abuse (U54 DA021519).
