Subject indexing and authority control in archives: The need for subject indexing in archives and for an indexing policy using controlled language

Journal of the Society of Archivists

Volumn 17, Issue 1, 1996, Pages 27-47

  Ribeiro, Fernanda ^a  

^a UNIVERSITY OF PORTO   (Portugal)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 21344471589     PISSN: 00379816     EISSN: None     Source Type: Journal    
DOI: 10.1080/00379819609511787     Document Type: Article

References (71)

1. Fernanda Ribeiro, Indexação e Controlo de Autoridade em Arquivos (Porto, Faculdade de Letras da Universidade, 1992), 2 vols.
2. For further reading about IRS, see: Gerard Salton and Michael J. McGill, 'Information Retrieval: an Introduction', in Introduction to Modern Information Retrieval (international student edition, Auckland, etc, McGraw-Hill International Book Company, 1983), pp 1-23.
3. About the concept of performance, in the context of Information Sciences, see: Stuart Hannabus, 'The Concept of Performance: a Semantic Review', ASLIB Proceedings, London, vol 39 (May 1987), pp 149-158.
4. Richard H. Lytle, 'Intellectual Access to Archives', The American Archivist, vol 43, part 1 (Winter 1980), pp 64-74;
5. Richard H. Lytle, 'Intellectual Access to Archives', The American Archivist, vol 43, part 2 (Spring 1980), pp 191-207.
6. Lytle, 'Intellectual access', p 193.
7. Steven L. Hensen, Archives, Personal Papers, and Manuscripts. A Cataloging Manual for Archival Repositories, Historical Societies and Manuscript Libraries (2nd ed, Chicago, Society of American Archivists, 1990);
8. Rules for Archival Description, Ottawa, Bureau of Canadian Archivists, 1990;
9. Michael Cook and Margaret Procter, Manual of Archival Description (2nd edn, Aldershot, Gower, 1989);
10. International Council On Archives, Ad Hoc Commission on Descriptive Standards, Statement of Principles Regarding Archival Description (1st version rev, Madrid, ICA, 1992);
11. International Council On Archives, Ad Hoc Commission on Descriptive Standards, ISAD(G): International General Standards for Archival Description (Stockholm, ICA, 1993).
12. About this see: Richard P. Smiraglia, 'Subject Access to Archival Materials Using LCSH', Cataloging & Classification Quarterly, vol 11, no 3/4 (1990), pp 63-90. The author considers that indexing in archives is done at different levels, following the analysis of the 'collection'. He proposes three indexing levels: 'summary level' - 'would require a single phrase that summarises the entire subject content of an archival collection and identifies the predominant forms that appear in the collection'; 'depth level' - 'usually will provide a more meaningful approach to an archival collection. To achieve subject cataloging at the depth level we break a collection of records into appropriate components and summarise each component individually'; 'exhaustive level' - 'the analysis of every component of a collection is undertaken, yielding the subsequent provision of headings for each and every component'. Although we agree with this theory in general, we cannot follow it on the designation of subject indexing (or cataloguing) assigned to the first two levels. In particular, concerning 'depth level' - the one that relates to our study - we think that the content summary is not 'indexing' in a rigorous sense. The theory presumes that there is analysis of content and then synthesis, but does not suggest how concepts may be identified, and creating representations by indexing terms is not done. The summary can be called 'content description' or 'content synthesis' or something similar, but hardly 'subject indexing'. The same author proposes that access points based on the Library of Congress Subject Headings (LCSH) will be chosen from the content summary, in order to retrieve information by subject. He gives guidelines for the application of LCSH. This, in fact, is the operation that can properly be called 'subject indexing'.
13. Although archival classification does not have the aim of organising documents by subject, it is in practice often used for that purpose. In our opinion, to use classification in this way subverts the basic principles which underlie archival management. However, we do admit that it is possible to use classification to complement information retrieved through subject indexing. See: Richard H. Lytle, 'Intellectual access' part 1, The American Archivist, vol 43, no 1 (Winter 1980), pp 64-75;
14. Dale Reed, Some Considerations on Archival Subject Access (paper presented to the Annual Meeting of the Society of American Archivists, Aug. 1986 - manuscript, kindly given by the author).
15. Instituto Porguguês da Qualidade. Comissão Técnica 7, Norma Portuguesa NP 3715. Documentação. Método para a Análise de Documentos, Determinação do seu Conteúdo e Selecção de Termos de Indexação (Lisboa, IPQ, 1989); NP 3715, point 5.1, p 5.
16. Richard P. Smiraglia, 'Subject Access to Archival Materials Using LCSH', defines stages and methods of subject analysis, although what he is doing is far from the concept of 'indexing', as it is consensually accepted, because he does not limit the notion of 'subject' only to the concepts expressed in the content. When he talks of categories like 'provenance', 'intellectual forms', 'physical forms' and 'presentation forms', saying that they must be taken into account in analysis, he is not restricting indexing to its real purpose, identifying subjects. Smiraglia's method is valid for content analysis - meaning by that 'what is contained' - in a very broad sense. But as he does not say this, readers suppose that he is referring to subject indexing in a more restricted sense. Thus, we think that Smiraglia's method cannot be used for content analysis, because his categories do not relate to subjects.
17. Because we did not index these series, there was a lack of much relevant information when we came to search. The subject content of those series would be useful in most searches, because they reflect the City Council's central activities. In order to plug this gap, and as users do not tend to know the importance of such series, we decided to add to the finding aids a note explaining that they should always consult the series in question, using the appropriate catalogues and indexes.
18. For example, we could suggest some criteria that could be useful to establish the steps in a search, particularly for archives from city councils, where many of the users' questions concern aspects of local life: for example - does the series refer to occupations? - does the series contain geographic names? - in the case of a series dealing with trade or industry, are the products or merchandise specified? - are there references to buildings, monuments, bridges, fountains, etc? - are there any references to institutions? - is there mention of personalities prominent in art, science or culture?
19. Arlene Taylor in 'Authority Files in Online Catalogs: an Investigation of Their Value', Cataloging & Classification Quarterly, vol 4, no 3 (Spring 1984), pp 1-17 defines authority work as 'the process of determining the form of a name, title, or subject concept that will be used as a heading on a bibliographic record; determining cross references needed to that form; and determining relationships of this heading to other authoritative headings. She also defines authority control as 'the process of maintaining consistency in headings in a bibliographic file through reference to an authority file'. This authority file will be a group of authority records, that is 'the printed or machine readable unit that records the decisions made during the course of authority work'.
20. See also Robert H. Burger, Authority Control. The Creation, Use, Maintenance and Evaluation of Authority Records and Files (Littleton, Libraries Unlimited, 1985);
21. Doris Hargrett Clack, Authority Control. Principles, Applications and Instructions (Chicago, London, American Library Association, 1990).
22. Gerard Salton, 'A Comparison Between Manual and Automatic Indexing Methods', American Documentation, vol 20 (Jan 1959), pp 61-71;
23. Elaine F. Svenonius, 'An Experiment in Index Term Frequency', Journal of the American Society for Information Science, vol 23, no 2 (Mar-Apr 1972), pp 109-121;
24. Karen Sparck Jones, 'Automatic Indexing', Journal of Documentation, vol 30, no 4 (Dec 1974), pp 393-432;
25. Maurice J. Freedman, Michael S. Malinconico (eds) The Nature and Future of the Catalog. Proceedings of the ALA's Information Science and Automation Division's 1975 and 1977 Institutes on Catalog (Phoenix, Oryx Press, 1979);
26. Gerald Salton, 'Suggestions for Library Network Design', Journal of Library Automation, vol 12 (Mar 1979), pp 39-52.
27. Derek Austin, 'Vocabulary Control and Information Technology', ASLIB Proceedings, vol 38, no 1 (Jan 1986), pp 1-15;
28. Cordelia R. Cavalcanti, 'A Recuperação da Informação e as Linguagens Documentárias', Revista Latinoamericana de Documentación, vol 3, no 2 (Jul-Dic 1983), pp 24-27;
29. C. P. R. Dubois, 'The Use of Thesauri in Online Retrieval', Journal of Information Science, vol 8, no 2 (1984), pp 63-66;
30. Alexis J. Jamieson et al. Journal of Information Science, (op. cit.);
31. Jaana Kristensen, Kalervo Järvelin, Journal of Information Science, (op. cit.);
32. F. W. Lancaster, Vocabulary Control for Information Retrieval (1st edn, Washington, Information Resources Press, 1972);
33. Noelle Van Pulis, Lorene E. Ludy, 'Subject Searching in an Online Catalog with Authority Control', College & Research Libraries, vol 49, no 6 (Nov 1983), pp 523-533.
34. International Standards Organization, Documentation. Principes Directeurs pour l'Établissement et le Dévéloppement de Thésaurus Monolingues. Norme Internationale ISO 2788 (2ème édn, Genève, ISO, 1986).
35. Jean Aitchison (comp), Unesco Thesaurus. A Structured List of Descriptors for Indexing and Retrieving Literature in the Fields of Education, Science, Social Science, Culture and Communications, (Paris, UNESCO, 1977), 2 vols;
36. Secretariat Des Missions d'Urbanisme Et d'Habitat, Thesaurus (Paris, SMUH, 1975), 2 vols.
37. Michael Cook, 'Applying Automated Techniques to Archives Adminstration: a Commentary on the Present Situation and Areas of Likely Progress', Journal of Documentation, vol 39, no 2 (June 1983), p 81;
38. Harriet Ostroff, 'Subject Access to Archival and Manuscript Material', American Archivist, vol 53 (Winter 1989), pp 100-105;
39. Richard P. Smiraglia, 'Subject access to archival materials';
40. an example of a specialised vocabulary is in William H. McNitt, 'Development of the PRESNET Subject Descriptor Thesaurus', American Archivist, vol 52 (Summer 1989), pp 358-364.
41. More details are available in Ana Franqueira, 'ARQBASE, a Portuguese Project for Automatic Processing of Archival Data', Janus. Revue Archivistique, vol 1 (1992), pp 25-30.
42. Ana Franqueira and Madalena Garcia, ARQBASE. Metodologia de Descrição Arquivística para Tratamento Automatizado de Documentação Histórica (Usboa, Instituto Porguguês de Arquivos, 1991).
43. Setting up an indexing system requires, obviously, a previous analysis of different parameters, as for example, the type of documents being dealt with, the kind of institution that is running the system, the characteristics of the users, the type of indexing language to be used, equipment, technical and human resources, the resulting index. These factors have been studied in relation to the definition of an indexing policy for libraries or information services. (On this we have, for example, the following literature: Jean Aitchison and Alan Gilchrist, 'System Design', in Thesaurus Construction. A Practical Manual (London, ASLIB, 1972), pp 2-12;
44. Marcel Van Dijk and Georges Van Slype, Le Service de Documentation Face à l'Explosion de l'Information (Paris, Les Éditions d'Organisation; Bruxelles, Les Presses Universitaires, 1969);
45. Georges Van Slype, Conception et Gestion des Systèmes Documentaires (Paris, Les Éditions d'Organisation, 1979). As far as indexing policies for archives are concerned, we have no guidelines at all. In our research we considered three basic parameters, mentioned above; the level of technical treatment (this implies the end product that will be elaborated), the type of archives and groups and the users' profile. To these parameters we also joined the definition of an indexing language. Other factors can, of course, influence the decision, although in a less decisive way.
46. On the notion of relevance see D. A. Kemp, 'Relevance, Pertinence and Information System Development', Information Storage and Retrieval, vol 10, no 2 (1974), pp 37-47;
47. Tefko Saracevic, 'Relevance: a Review of and a Framework for Thinking on the Notion in Information Science', Journal of the American Society for Information Science, vol 26, no 6 (1975), pp 321-343.
48. This scale of levels of relevance has already been used by other authors. See, for example: Rao Aluri, 'Evaluation of Subject Retrieval in Online Catalogs', in Subject Analysis in Online Catalogs (Englewood, Libraries Unlimited, 1991), pp 273-297. The users mostly put questions related to their own research, which meant that they knew almost all the relevant documents existing in the archives. Because of that, it was easy for them to make a relevance judgment about the retrieved records, using only the information shown on the computer screen. In a few cases they needed to consult the documents themselves to give their opinion. It was pleasant to realise, in various cases, that the databases gave access to some information which the users could not have found through the manual finding aids, even though the access points were basically the same. However, the combination of search terms, made possible by the computer, enabled a more effective retrieval, whether in terms of recall or in terms of precision.
49. UNESCO, Division of Software Development and Applications. Office of Information Programmes and Services, Mini-Micro CDS/ISIS Reference Manual (version 2.3) (1st edn, Paris, UNESCO, 1989), pp 95-107.
50. a Versão-Janeiro de 1990) (Lisboa, Instituto Português de Arquivos, 1990), p 38.
51. The main guidelines for an evaluation experiment are in: F. W. Lancaster, Principes Directeurs pour l'Évaluation des Systèmes et Services d'Information (Paris, UNESCO, 1978, PGI/78/WS/18).
52. See also Paul B. Kantor, 'Evaluating of and Feedback in Information Storage and Retrieval Systems', Annual Review of Information Science and Technology, vol 17 (1982), pp 99-120;
53. Donald W. King and Edward C. Bryant, Evaluation of Information Systems (Washington, Information Resources Press, 1971);
54. F. W. Lancaster, Information Retrieval Systems. Characteristics, Testing and Evaluation (2nd edn, New York, etc, John Wiley & Sons, 1979);
55. John J. Regazzi, 'Evaluating Indexing Systems: a Review After Cranfield', The Indexer, vol 12, no 1 (Apr 1980), pp 14-21;
56. Gerard Salton and Michael J. McGill, 'Retrieval Evaluation'.
57. Besides these measurements, there are others that can be considered, in order to determine rigorously the effectiveness of an IRS. There are many studies on this subject. Salton's writings are particularly important because he defends the use of another measurement - fallout - as essential to achieve a correct evaluation result. See: Gerard Salton and Michael J. McGill, 'Retrieval Evaluation'.
58. This discussion relates to 'retrieved records' and not to 'retrieved documents'. Readers must bear in mind that most records represent series, and not individual documents.
59. For calculating precision and noise rates, only the questions for which some records were retrieved were considered. It was not possible to know exactly if the failure to retrieve records was due to problems of the indexing language or simply because there were not, in the databases, any records relevant to the question. In the 4 cases of non-retrieval in both databases it is almost certain that it was because there were no documents corresponding to the questions put by the users.
60. Knowing the number of documents that exist in an IRS that are relevant to a query, is a precondition for calculating the recall rate. But since in most cases - due to the quantity of documents - it is not possible to judge relevance through a document-by-document analysis, it is necessary to make an estimate. There are some statistical methods of doing this, that work by sampling. One of these has become known as the Cleverdon method and it is not very complicated. A description of it is in: Marcel van Dijk and Georges Van Slype, Le Service de Documentation Face à l'Explosion de l'Information (Paris, Les Éditions d'Organisation; Bruxelles, Les Presses Universitaires, 1969), pp 247-248.
61. Other methods can be used, which are described in the literature. (See, for example: F. W. Lancaster, Principes Directeurs;
62. W. A. Martin, 'Methods For Evaluating The Number of Relevant Documents In A Collection', Journal Of Information Science, vol 6 (1983), pp 173-177;
63. John Martyn and F. W. Lancaster, Investigative Methods In Library And Information Science. An Introduction (Arlington, Information Resources Press, 1981), pp 163-169;
64. Don Miller and Robert T. Dattola, 'Methods for Estimating the Number of Relevant Documents in a Collection', Information Storage Processing & Management, vol 18, no 4 (1982), pp 179-191. The way one might estimate the number of documents relevant to a query that might exist in an archival database, where the treatment is at series level (or other levels containing sets of documents and not individual documents) needs further research. It is not possible to apply methods devised for use in libraries, without changes.
65. C. W. Cleverdon, Progress in Documentation, p 63.
66. C. W. Cleverdon, 'On the Inverse Relationship of Recall and Precision', Journal of Documentation, vol 28, no 3 (Sep 1972), p 199.
67. Richard H. Lytle, 'Intellectual access', p 200.
68. F. W. Lancaster, Principes Directeurs, pp 64-76; Gerard McGill and Michael J. Salton, pp 186-191.
69. The time spent in searching was not taken into consideration, because it depends primarily on the number of queries required by each question. System characteristics - central processor speed and memory available - influence the time taken by searching. Delays are usually a problem in informatics, not in questions of indexing language.
70. Global costs should be taken into account. These include the costs of input, which were doubtless greater for database B, because the controlled indexing language demands more time and more effort in indexing. But database A also requires considerable effort, because of the elaboration of subject content abstracts. It is also necessary to do content analysis, as in database B. We can only save the time and effort demanded by the selection of concepts and their representation by indexing terms.
71. These measurements were also used by Richard Lytle in his experiment - see Richard H. Lytle, 'Intellectual access', pp 199-201.