Search (1 results, page 1 of 1)

0.011551105 = product of:
  0.02310221 = sum of:
    0.02310221 = product of:
      0.04620442 = sum of:
        0.04620442 = weight(_text_:systems in 3640) [ClassicSimilarity], result of:
          0.04620442 = score(doc=3640,freq=16.0), product of:
            0.16037072 = queryWeight, product of:
              3.0731742 = idf(docFreq=5561, maxDocs=44218)
              0.052184064 = queryNorm
            0.28811008 = fieldWeight in 3640, product of:
              4.0 = tf(freq=16.0), with freq of:
                16.0 = termFreq=16.0
              3.0731742 = idf(docFreq=5561, maxDocs=44218)
              0.0234375 = fieldNorm(doc=3640)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

This classic paper presents the reasoning behind the research undertaken by the Classification Research Group in London, beginning in 1952 and producing, by 1955, the direction in which the Group's efforts were to go in the next thirty years. The Group's original purpose was to review the basic principles of indexing and classification without committing itself to any existing system. It began by uncovering - among existing systems such as indexes, classifications, automatic selectors, and other information retrieval systems - the steps in the process by means of which a search was performed. The Group went over this very carefully, identifying parts of the process each step of the way. At the time this work was performed systems such as UNITERMS, which did not survive, and other alphabetical coordinated indexes, mainly experimental, were a dime a dozen. Classification to most librarians meant Dewey and the Library of Congress systems, both of which have very serious shortcomings from an intellectual point of view. The Group finally came to the conclusion that a classification of knowledge was necessary for constructing any successful retrieval system. The question then became one of deciding which kind of classification system. The members identified ten unsatisfactory features of existing systems. In fact, they could not find any general class schedule that either was satisfactory or could be made satisfactory. Obviously then, a new system had to be made. The question became one of how to do this. Existing theories did not conform to the theory of logical division; dividing and subdividing an the basis of a single characteristic was not followed. Thus logical division, which is a "top down" method of analysis, was rejected. Generic relationships would have to be made by some other methodology. A "bottom-up" or inductive, as opposed to deductive, method was a possibility. The Group actually decided to use a system in which a given genus could be subdivided in more than one way, thus "yielding a homogeneous group of collateral species."
The technique chosen was S. R. Ranganathan's facet analysis (q.v.). This method works from the bottom up: a term is categorized according to its parent class, as a kind, state, property, action, operation upon something, result of an Operation, agent, and so on. These modes of definition represent characteristics of division. Following the publication of this paper, the group worked for over ten years developing systems following this general pattern with various changes and experimental arrangements. Ranganathan's Colon Classification was the original of this type of method, but the Group rejected his contention that there are only five fundamental categories to be found in the knowledge base. They did, in fact, end up with varying numbers of categories in the experimental systems which they ultimately were to make. Notation was also recognized as a problem, being complex, illogical, lengthy, obscure and hard to understand. The Group tried to develop a rationale for notation, both as an ordering and as a finding device. To describe and represent a class, a notation could be long, but as a finding device, brevity would be preferable. The Group was to experiment with this aspect of classification and produce a number of interesting results. The Classification Research Group began meeting informally to discuss classification matters in 1952 and continues to meet, usually in London, to the present day. Most of the British authors whose work is presented in these pages have been members for most of the Group's life and continue in it. The Group maintains the basic position outlined in this paper to the present day. Its experimental approach has resulted in much more information about the nature and functions of classification systems. The ideal system has yet to be found. Classification research is still a promising area. The future calls for more experimentation based an reasoned approaches, following the example set by the Classification Research Group.