NLM TEST COLLECTION PERSON TO CONTACT: Dr. Alexa T. McCray National Library of Medicine Bethesda, Maryland 20894 (301) 496-9300 mccray@csb1.nlm.nih.gov COPYRIGHT CONSTRAINTS: 1. Some NLM databases carry an international copyright when distributed outside the United States, its Territories or Possessions. 2. The following must be displayed in full in order that its users be made aware of the copyright constraints. "Some material in the NLM databases is from copyrighted publications of the respective copyright claimants. Users of the NLM databases are solely responsible for compliance with any copyright restrictions and are referred to the publication data appearing in the bibliographic citations, as well as to the copyright notices appearing in the original publications, all of which are hereby incorporated by reference." (In the United States, Paragraph 2. does not apply to Technical Services databases.) 3. As to the re-use of NLM databases, the Licensee and/or users shall be solely responsible for compliance with any copyright restrictions; NLM assumes no responsibility or liability associated with the Licensee's (or any of the Licensee's users') use and/or reproduction of copyrighted material. Anyone contemplating reproduction of all or any portion of any of the NLM databases should consult legal counsel. 4. The producers of some of the data (such as Chemical Abstracts Service (CAS), BioScience Information Service (BIOSIS), and the American Society of Hospital Pharmacists (ASHP) contained in TOXLIT and CHEMLINE, require that separate royalty agreements be negotiated in advance with these producers. DESCRIPTION OF TEST COLLECTION: The test collection consists of retrieved MEDLINE citation records, with relevancy judgments, for questions asked of the MEDLINE database at either the NLM library or the NIH library during 1985. The questions were asked by health professionals, e.g., physicians, researchers in the life sciences or health care delivery, etc., and were selected from a larger set of library request forms. They represent three major content areas: Health Services Research, Clinical Medicine Research, and Basic Science Research. The questions were searched on a subset of MEDLINE (year86) by an expert NLM searcher. The search strategy emphasized recall over precision, and NLM's controlled vocabulary for indexing and retrieval (MeSH) was used whenever possible. Occasionally the strategy was further augmented with text-word searching. The retrieval results were evaluated by a subject matter expert. The collection consists of 155 queries (transcribed from library search request forms) and 3,078 MEDLINE citation records. Each citation record is marked as either relevant or non-relevant in relation to the query. The citation record consists of a unique identifier (8-digit number), author name(s), title, abstract, a collection of MeSH headings, and journal source. The records were judged for relevancy to the query based on the title and abstract only. The subset of MEDLINE (year86) consists of 167,000 citations. This represents a portion of the citations added to MEDLINE which have a 1986 publication date. All citation records are to articles written in English and all records include an abstract. If you would like to obtain the year86 file, which is also available for research use only, write to: Lois Ann Colaianni Associate Director of Library Operations National Library of Medicine Bethesda, Maryland 20894 In your letter include a description of the research use to which you intend to put the file. You will be asked to provide a written report on the results of your research with this test database. DESCRIPTION OF FILES IN THE NLM TEST COLLECTION: Character Format: ASCII. Record Format: Variable. Number of Files: 310 separate ASCII text files. Number of Directories: 3 with 2 subdirectories each* - health queries (queries with retrieved citation records attached) relevncy (relevancy judgments) medicine queries (queries with retrieved citation records attached) relevncy (relevancy judgments) science queries (queries with retrieved citation records attached) relevncy (relevancy judgments) *If you receive the collection on diskettes, the logical organization of the files holds, but the query directories may have been split due to space constraints. File Description: A. There are three sets of files, one set for each of the domains listed above: Basic Science Research (BSR), Clinical Medicine Research (CMR), and Health Services Research (HSR). The first line of each file gives the file name surrounded by a set of equal signs; e.g., the file CMR01 has as its first line: "==========================CMR01=========================" B. Files with the '.ui' extension: Each numbered file has a list of MEDLINE unique identifiers which reflect the citations which were retrieved for that query; e.g., the file named 'CMR01.ui' has the unique identifiers for clinical medicine question number 1. Following each unique identifier is either 'y' (citation is relevant to the query) or 'n' (citation is not relevant to the query). C. Files without the '.ui' extension: Each numbered file begins with the actual query; e.g., the file named 'CMR01' contains clinical medicine question number 1. The query is immediately followed by the retrieved citation records. Each citation record has the labeled fields, UI (MEDLINE unique identifier), AU (one or more authors), TI (title), AB (abstract), MH (list of MeSH terms), SO (journal), in that order.