Search (38 results, page 1 of 2)

0.049055643 = product of:
  0.09811129 = sum of:
    0.09811129 = sum of:
      0.05608411 = weight(_text_:searching in 131) [ClassicSimilarity], result of:
        0.05608411 = score(doc=131,freq=2.0), product of:
          0.2091384 = queryWeight, product of:
            4.0452914 = idf(docFreq=2103, maxDocs=44218)
            0.051699217 = queryNorm
          0.26816747 = fieldWeight in 131, product of:
            1.4142135 = tf(freq=2.0), with freq of:
              2.0 = termFreq=2.0
            4.0452914 = idf(docFreq=2103, maxDocs=44218)
            0.046875 = fieldNorm(doc=131)
      0.04202718 = weight(_text_:22 in 131) [ClassicSimilarity], result of:
        0.04202718 = score(doc=131,freq=2.0), product of:
          0.18104185 = queryWeight, product of:
            3.5018296 = idf(docFreq=3622, maxDocs=44218)
            0.051699217 = queryNorm
          0.23214069 = fieldWeight in 131, product of:
            1.4142135 = tf(freq=2.0), with freq of:
              2.0 = termFreq=2.0
            3.5018296 = idf(docFreq=3622, maxDocs=44218)
            0.046875 = fieldNorm(doc=131)
  0.5 = coord(1/2)

Abstract

After a fundamental examination of the phenomenon of time, this paper presents the history, authority, and structure of period subdivisions and chronological terms in the three subject heading languages LCSH (Library of Congress Subject Headings), RAMEAU (Répertoire d'Autorité Matière Encyclopédique et Alphabétique Unifié), and RSWK (Regeln für den Schlagwortkatalog). Their usefulness in online searching is demonstrated using the online catalogs of the Library of Congress, the Bibliothèque nationale de France, and the Deutsche Bibliothek and is compared to the search options in selected digital encyclopedias (Encyclopaedia Britannica, Encarta, Brockhaus-Enzyklopädie). The author develops a model for common time retrieval across all three online catalogs, outlines the conditions for that model (time period code, chronological code, and chronology authority file), and proposes a search interface.

Date

10. 9.2000 17:38:22

0.03271573 = product of:
  0.06543146 = sum of:
    0.06543146 = product of:
      0.13086292 = sum of:
        0.13086292 = weight(_text_:searching in 1330) [ClassicSimilarity], result of:
          0.13086292 = score(doc=1330,freq=2.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.6257241 = fieldWeight in 1330, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.109375 = fieldNorm(doc=1330)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

0.03238018 = product of:
  0.06476036 = sum of:
    0.06476036 = product of:
      0.12952071 = sum of:
        0.12952071 = weight(_text_:searching in 5169) [ClassicSimilarity], result of:
          0.12952071 = score(doc=5169,freq=6.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.61930627 = fieldWeight in 5169, product of:
              2.4494898 = tf(freq=6.0), with freq of:
                6.0 = termFreq=6.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0625 = fieldNorm(doc=5169)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Open system protocols for search and retrieval have not provided explicit ways in which to implement thesaurus-aided searching. A number of different approaches within the existing protocols, as well as a proposed service, are evaluated. A general approach to implementing thesaurus-aided searching, particularly during consultation of a thesaurus, requires an entirely new service, whose main features are described

0.028042056 = product of:
  0.05608411 = sum of:
    0.05608411 = product of:
      0.11216822 = sum of:
        0.11216822 = weight(_text_:searching in 2098) [ClassicSimilarity], result of:
          0.11216822 = score(doc=2098,freq=2.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.53633493 = fieldWeight in 2098, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.09375 = fieldNorm(doc=2098)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

0.028042056 = product of:
  0.05608411 = sum of:
    0.05608411 = product of:
      0.11216822 = sum of:
        0.11216822 = weight(_text_:searching in 5277) [ClassicSimilarity], result of:
          0.11216822 = score(doc=5277,freq=2.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.53633493 = fieldWeight in 5277, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.09375 = fieldNorm(doc=5277)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

0.026438305 = product of:
  0.05287661 = sum of:
    0.05287661 = product of:
      0.10575322 = sum of:
        0.10575322 = weight(_text_:searching in 3349) [ClassicSimilarity], result of:
          0.10575322 = score(doc=3349,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.5056614 = fieldWeight in 3349, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0625 = fieldNorm(doc=3349)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Attempts to gauge whether syndetic structure, the cross reference system long regarded as an integral component of the controlled vocabulary which predominates as the indexing framework for online subject searching, still has a role to play in assisting end users in their information quests. To this end, examines a number of existing and proposed enhancements to OPACs by means of a literatur review. These include: Boolean and keyword searching; online browsing; artificial intelligence; classification clustering; and hypertext. Discusses Dervin's sensemaking theories and methodologies

0.024285134 = product of:
  0.048570268 = sum of:
    0.048570268 = product of:
      0.097140536 = sum of:
        0.097140536 = weight(_text_:searching in 3351) [ClassicSimilarity], result of:
          0.097140536 = score(doc=3351,freq=6.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.46447968 = fieldWeight in 3351, product of:
              2.4494898 = tf(freq=6.0), with freq of:
                6.0 = termFreq=6.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.046875 = fieldNorm(doc=3351)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Investigates the extent to which title keywords convey subject content and compares the relative effectiveness of searching title keywords using 2 search strategies to examine whether adjacency operators in title keyword searches are effective in improving recall and precision of online searching. Title keywords from a random sample of titles in the field of economics were searched on FirstSearch, using the WorldCat database, which is equivalent in coverage to the OCLC OLUC, with and without adjacency of the keywords specified. The LCSH of the items retrieved were compared with the sample title subject headings to determine the degree of match or relevance and the values for precision and recall were calculated. Results indicated that, when keywords were discipline specific, adjacency operators improved precision with little degradation of recall. Systems that allow positional operators or rank output by proximity of terms may increase search success

0.024285134 = product of:
  0.048570268 = sum of:
    0.048570268 = product of:
      0.097140536 = sum of:
        0.097140536 = weight(_text_:searching in 5698) [ClassicSimilarity], result of:
          0.097140536 = score(doc=5698,freq=6.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.46447968 = fieldWeight in 5698, product of:
              2.4494898 = tf(freq=6.0), with freq of:
                6.0 = termFreq=6.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.046875 = fieldNorm(doc=5698)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Describes the shortcomings of search engines for the WWW comparing their current capabilities to those of the first generation CD-ROM products. Some allow phrase searching and most are improving their Boolean searching. Few allow truncation, wild cards or nested logic. They are stateless, losing previous search criteria. Unlike the indexing and classification systems for today's CD-ROMs, those for Web pages are random, unstructured and of variable quality. Considers that at best Web search engines can only offer free text searching. Discusses whether automatic data classification systems such as Infoseek Ultra can overcome the haphazard nature of the Web with neural network technology, and whether Boolean search techniques may be redundant when replaced by technology such as the Euroferret search engine. However, artificial intelligence is rarely successful on huge, varied databases. Relevance ranking and automatic query expansion still use the same simple inverted indexes. Most Web search engines do nothing more than word counting. Further complications arise with foreign languages

0.02336838 = product of:
  0.04673676 = sum of:
    0.04673676 = product of:
      0.09347352 = sum of:
        0.09347352 = weight(_text_:searching in 2306) [ClassicSimilarity], result of:
          0.09347352 = score(doc=2306,freq=2.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.44694576 = fieldWeight in 2306, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.078125 = fieldNorm(doc=2306)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

0.023133516 = product of:
  0.046267033 = sum of:
    0.046267033 = product of:
      0.092534065 = sum of:
        0.092534065 = weight(_text_:searching in 2640) [ClassicSimilarity], result of:
          0.092534065 = score(doc=2640,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.44245374 = fieldWeight in 2640, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0546875 = fieldNorm(doc=2640)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

For many years, searching medical and health-related topics has caused thousands of searchers to consult the well-known Medical Subject Headings (MeSH) controlled vocabulary. This plan works well for searching MEDLINE, but what about the MEDLARS databases such as TOXLIT, BIOETHICSLINE and POPLINE? Several printed vocabularies are available for searchers who wish to benefit from the consistency afforded by consulting established term lists. Occasionally, use of keywords from vocabularies other than MeSH is recommended even when information in the database is also indexed to MeSH. Search techniques available for use with several of the controlled vocabularies which are discussed vary, but built-in consistency of term use is a continuing theme from one controlled vocabulary to another

0.023133516 = product of:
  0.046267033 = sum of:
    0.046267033 = product of:
      0.092534065 = sum of:
        0.092534065 = weight(_text_:searching in 2883) [ClassicSimilarity], result of:
          0.092534065 = score(doc=2883,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.44245374 = fieldWeight in 2883, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0546875 = fieldNorm(doc=2883)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Optimal retrieval in on-line searching can be achieved through combined use of both natural language and controlled vocabularies. However, there is a large variety of types of controlled vocabulary in data bases and often more than one in a single data base. Optimal use of these vocabularies requires understanding what types of languages are involved, and taking advantage of the particular mix of vocabularies in a given data base. Examples 4 major types of indexing and classification used in data bases and puts these 4 in the context of 3 other approaches to subject access. Discusses how to evaluate a new data base for various forms of subject access.

0.023133516 = product of:
  0.046267033 = sum of:
    0.046267033 = product of:
      0.092534065 = sum of:
        0.092534065 = weight(_text_:searching in 6930) [ClassicSimilarity], result of:
          0.092534065 = score(doc=6930,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.44245374 = fieldWeight in 6930, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0546875 = fieldNorm(doc=6930)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Introducess a new approach to display retrieved subject headings in subject searching designed to encourage users to browse bibliographic information. Emphasizes the importance of the exact display approach by showing how many user queries could be candidates for this approach, demonstrates an implementation of the exact-display approach in an experimental online catalogue. End user experiences gives opportunities to make recommendations for enhancing the original design of the exact display approach so that future implementations of this approach in operational online catalogues are responsive to the needs of online catalogue users

0.023133516 = product of:
  0.046267033 = sum of:
    0.046267033 = product of:
      0.092534065 = sum of:
        0.092534065 = weight(_text_:searching in 207) [ClassicSimilarity], result of:
          0.092534065 = score(doc=207,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.44245374 = fieldWeight in 207, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0546875 = fieldNorm(doc=207)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Optimal retrieval in on-line searching can be achieved through combined use of both natural language and controlled vocabularies. However, there is a large variety of types of controlled vocabulary in data bases and often more than one in a single data base. Optimal use of these vocabularies requires understanding what types of languages are involved, and taking advantage of the particular mix of vocabularies in a given data base. Examples 4 major types of indexing and classification used in data bases and puts these 4 in the context of 3 other approaches to subject access. Discusses how to evaluate a new data base for various forms of subject access.

0.023133516 = product of:
  0.046267033 = sum of:
    0.046267033 = product of:
      0.092534065 = sum of:
        0.092534065 = weight(_text_:searching in 2775) [ClassicSimilarity], result of:
          0.092534065 = score(doc=2775,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.44245374 = fieldWeight in 2775, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0546875 = fieldNorm(doc=2775)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Browse searching of online catalogs and electronic indexes is a popular means of retrieving information, but some subject heading lists are so extensive that browsing them can be difficult and time-consuming. This resaerch tested one way of presenting a browse interface to reduce the amount of scanning of subject headings required. It found that a hierarchical presentation of headings reduced the amount of scanning required by more than 50% without any reduction in the effectiveness of the search. Better browsable displays of this sort can improve the efficiency of searching but appear not to alter effectivenenss

0.020237612 = product of:
  0.040475223 = sum of:
    0.040475223 = product of:
      0.08095045 = sum of:
        0.08095045 = weight(_text_:searching in 5567) [ClassicSimilarity], result of:
          0.08095045 = score(doc=5567,freq=6.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.38706642 = fieldWeight in 5567, product of:
              2.4494898 = tf(freq=6.0), with freq of:
                6.0 = termFreq=6.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0390625 = fieldNorm(doc=5567)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Subject searching is difficult with traditional text-based online public access library catalogues (OPACs), and the next-generation discovery layers are keyword searching and result filtering tools that offer little support for subject browsing. Next-generation OPACs ignore the rich network of relations offered by controlled subject vocabulary, which can facilitate subject browsing. A new generation of OPACs could leverage existing information-organization investments and offer online searchers a novel browsing and searching environment. This is a case study of the design and development of a virtual reality subject browsing and information retrieval tool. The functional prototype shows that the Library of Congress subject headings (LCSH) can be shaped into a useful and usable tree structure serving as a visual metaphor that contains a real world collection from the domain of science and engineering. Formative tests show that users can effectively browse the LCSH tree and carve it up based on their keyword search queries. This study uses a complex information-organization structure as a defining characteristic of an OPAC that goes beyond the standard keyword search model, toward the cutting edge of online search tools.

0.018694704 = product of:
  0.03738941 = sum of:
    0.03738941 = product of:
      0.07477882 = sum of:
        0.07477882 = weight(_text_:searching in 4459) [ClassicSimilarity], result of:
          0.07477882 = score(doc=4459,freq=2.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.3575566 = fieldWeight in 4459, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0625 = fieldNorm(doc=4459)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Discusses established approaches to subject access using subject headings (such as LCSH), and the demands for improved subject access associated with the widespread availability of OPACs and the evidence of problems experienced by catalogue users in subject searching

0.018694704 = product of:
  0.03738941 = sum of:
    0.03738941 = product of:
      0.07477882 = sum of:
        0.07477882 = weight(_text_:searching in 6579) [ClassicSimilarity], result of:
          0.07477882 = score(doc=6579,freq=2.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.3575566 = fieldWeight in 6579, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0625 = fieldNorm(doc=6579)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Considers the future of LCSH and whether their viability can be maintained through a written standardization code. Until there is an online substitute for LC subject headings which has been tested on a large scale over a period of time, a successor to the present combination of LC's controlled vocabulary headings plus computerized techniques such as keyword searching is unlikely to emerge

0.01652394 = product of:
  0.03304788 = sum of:
    0.03304788 = product of:
      0.06609576 = sum of:
        0.06609576 = weight(_text_:searching in 96) [ClassicSimilarity], result of:
          0.06609576 = score(doc=96,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.3160384 = fieldWeight in 96, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0390625 = fieldNorm(doc=96)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

The paper looks at ways in which traditional classification and indexing tools have dealt with the relationships between constituent terms; variations in these are examined and compared with the methods used in machine searching, particularly of the Internet and World Wide Web. Apparent correspondences with features of index languages are identified, and further methods of applying classification and indexing theory to machine retrieval are proposed. There are various ways in which indexing and retrieval systems, both conventional and electronic, deal with the problem of searching for documents on a subject basis, and various approaches to the analysis and processing of a query. There appear to be three basic models; the taxonomic or structural system, in which the user is offered a map of the `universe of knowledge'; the language based system, which offers a vocabulary of the subject and a grammar for dealing with compound statements; and the mathematical model using the language of symbolic logic or the algebra of set theory

0.01652394 = product of:
  0.03304788 = sum of:
    0.03304788 = product of:
      0.06609576 = sum of:
        0.06609576 = weight(_text_:searching in 1970) [ClassicSimilarity], result of:
          0.06609576 = score(doc=1970,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.3160384 = fieldWeight in 1970, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0390625 = fieldNorm(doc=1970)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Authority files have played an important role in improving the quality of indexing and subject cataloging. Although authorities can significantly improve searching by increasing the number of access points, they are rarely an integral part of the information retrieval process, particularly end-users' searches. A retrieval prototype, searchFAST, was developed to test the feasibility of using an authority file as an index to bibliographic records. searchFAST uses Faceted Application of Subject Terminology (FAST) as an index to OCLC's WorldCat.org bibliographic database. The searchFAST prototype complements, rather than replaces, existing WorldCat.org access. The bibliographic file is searched indirectly; first the authority file is searched to identify appropriate subject headings, then the headings are used to retrieve the matching bibliographic records. The prototype demonstrates the effectiveness and practicality of using an authority file as an index. Searching the authority file leverages authority control work by increasing the number of access points while supporting a simple interface designed for end-users.

0.01652394 = product of:
  0.03304788 = sum of:
    0.03304788 = product of:
      0.06609576 = sum of:
        0.06609576 = weight(_text_:searching in 2703) [ClassicSimilarity], result of:
          0.06609576 = score(doc=2703,freq=4.0), product of:
            0.2091384 = queryWeight, product of:
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.051699217 = queryNorm
            0.3160384 = fieldWeight in 2703, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              4.0452914 = idf(docFreq=2103, maxDocs=44218)
              0.0390625 = fieldNorm(doc=2703)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

When consumers search for health information, a major obstacle is their unfamiliarity with the medical terminology. Even though medical thesauri such as the Medical Subject Headings (MeSH) and related tools (e.g., the MeSH Browser) were created to help consumers find medical term definitions, the lack of direct and explicit integration of these help tools into a health retrieval system prevented them from effectively achieving their objectives. To explore this issue, we conducted an empirical study with two systems: One is a simple interface system supporting query-based searching; the other is an augmented system with two new components supporting MeSH term searching and MeSH tree browsing. A total of 45 subjects were recruited to participate in the study. The results indicated that the augmented system is more effective than the simple system in terms of improving user-perceived topic familiarity and question-answer performance, even though we did not find users spend more time on the augmented system. The two new MeSH help components played a critical role in participants' health information retrieval and were found to allow them to develop new search strategies. The findings of the study enhanced our understanding of consumers' search behaviors and shed light on the design of future health information retrieval systems.