State Records Home
Personal tools
You are here :: Home Recordkeeping in the NSW public sector Keyword products Contemporary Recordkeeping - The Records Management Thesaurus - Response

Contemporary Recordkeeping - The Records Management Thesaurus - Response

Article in response to paper delivered at the 1997 Records Management Association of Australia's National Convention

Authors

Catherine Robinson B.A. (Hons.), Dip.IM. (Archiv.Admin.)
Janet Knight B.A., Dip.IM. (Archiv.Admin.), Dip.Loc.& App.Hist.

Abstract

At the Records Management Association of Australia's National Convention 1997, Dr Maggie Exon, Senior Lecturer, Department of Information Studies at Curtin University, delivered a paper called 'Contemporary Recordkeeping: The Records Management Thesaurus.' This paper extensively criticised the Keyword AAA thesaurus produced and marketed by the Archives Authority of New South Wales. In addition to appearing in the convention proceedings, her paper was published in the November 1997 edition of Informaa Quarterly.

This paper is the Archives Authority's public response to that criticism. The authors discuss the origins of Keyword AAA and, in particular, its departure from the tradition of subject-based records management thesauri. They then discuss Dr Exon's concerns about Keyword AAA's relationship with the International Standard ISO 2788, Documentation - Guidelines for the establishment and development of monolingual thesauri, and describe the impact of the Australian Standard AS 4390, Records Management, on the theory and practice of records management thesaurus development.

The authors describe the application of functional analysis to the development of business classification schemes, as advocated in the Australian Standard, how these schemes provide the basis for controlled language and the relationship of thesauri with other tools in a comprehensive and coherent regime for managing government records.

Finally the authors place Keyword AAA in the context of efforts, in which the Authority is playing an active role, in New South Wales and elsewhere to provide better access to government records and other forms of information.

Article

At the Association's National Convention in Perth in September 1997, Dr Maggie Exon presented a paper on the contemporary records management thesaurus which contained extensive criticism of the Keyword AAA thesaurus. As we were given no opportunity at the Convention to respond to Dr Exon's paper in a considered way, the editor of Informaa Quarterly agreed to publish this response. This article deals with the technical issues raised in Dr Exon's paper and is also an opportunity to place Keyword AAA within the context of modern recordkeeping theory and practice.

What is Keyword AAA?

Keyword AAA is a thesaurus of general terms designed for use in classifying, titling and indexing all types of records in any technological environment. It covers terminology common to business functions and activities in most organisations and is normally used in conjunction with a thesaurus of functionalterms, relating to the organisation's specific or core business functions, to provide comprehensive controlled vocabulary coverage.

Keyword AAA is a logical progression in thesaurus products developed by the Archives Authority of New South Wales, through its Records Management Office, over the last 20 years. The Thesaurus of General AdministrativeTerms, or the 'GADM' as it has been better known, was released in 1978 (not 1991, as Dr Exon states). The GADM Thesaurus became Australia's most popular and successful government records management thesaurus and in 1995 was being used by more than 140 public sector organisations at all three levels of government around Australia. The GADM Thesaurus was regularly revised and Keyword AAA, released in November 1995, is the product of the major 1994/1995 revision. The different name reflects the fact that our revision had involved a fundamental rethinking and had resulted in a new product, rather than just a new version of the old product.

Like its predecessor, Keyword AAA was designed specifically for the New South Wales Government. Because of the similarity in general administrative terminology used in other jurisdictions, it has been adopted as the standard general records management thesaurus in Western Australia, South Australia and Tasmania and is used extensively by Commonwealth Government agencies. It is currently being implemented and used by 125 organisations in Australia and Canada. It is also used extensively by educators in the recordkeeping discipline to help teach students around Australia about controlled language principles and tools. Curtin University of Technology has an educator's licence specifically for this purpose.

Keyword AAA is marketed on an organisation-wide licensing basis, enabling licensees to provide it to users in hard copy, on computer networks and Intranets and to incorporate it in records and document management software. It is an important source of revenue for the Authority's Government Recordkeeping program, helping to fund its research and development work in relation to records management standards and the management of electronic records.

While we must carefully protect our commercial interest in the intellectual property in Keyword AAA, we have not hidden its foundations from the records management community. We believe it is a significant contribution to records management practice and we have discussed this in the professional literature and in professional forums.

How does Keyword AAA work?

For readers unfamiliar with Keyword AAA, it will be helpful to describe how it is structured and works. Unlike GADM, which was subject based, Keyword AAA is constructed to reflect an organisation's business functions and activities as they are documented by records. Keyword AAA uses three levels of terms to do this:

  • keywords, which represent business functions of an agency
  • activity descriptors, which describe the more specific activities taking place, and
  • subject descriptors, which describe the more specific subjects or topics relating to the matter to be documented.

Thus, when used with structured file titling (this is a common, but not the sole, way of using Keyword AAA), we recommend assigning a keyword, an activity descriptor and a subject descriptor, in that order. We also provide guidance about using the names and organisations as part of this third level of description. Naturally it is at this level that most flexibility is required.

It is fundamentally important to an understanding of Keyword AAA and other modern records management thesauri to recognise that the act of assigning keywords and descriptors is not merely concerned with attaching 'labels' to records to aid their retrieval. It is an act of classification. By assigning a structured title to a file, we classify the file. By attaching individual records to the file, we classify those records. We will return to the crucial role of classification in records management later.

The GADM worked in quite a different way from Keyword AAA. It had only two levels, keywords and descriptors. The rule was that one keyword should be assigned in a file title (the GADM was designed specifically for file-based recordkeeping systems), followed by as many descriptors as were required to describe the matter to be documented by the file (this was termed the index depth or degree of specificity). The descriptors in the thesaurus were all of equal value and most descriptors could be used after many keywords. The keyword POLICY was an exception to the basic rule: it could be used with another keyword.

In its final stages the GADM was, like Elvis, bloated and cumbersome, as our users made very clear to us. It contained a vast array of descriptors (one of our favourites was Argentine ants) and forbidden terms, the majority of which our users indicated were never used. By 1994, at the very least, the GADM required a thorough spring clean.

The two levels and the relative lack of hierarchy, moreover, severely limited the GADM's value as a tool for classification. This was compounded by the fact that the descriptor level did not distinguish between activity-oriented terms and topic-oriented ones. To produce a thesaurus useful for classification, we had to separate the two and impose a stricter hierarchy than the GADM allowed. These were key factors in the design of what became Keyword AAA.

Technical issues

In her paper, Dr Exon raised several technical issues and touched on their broader implications. We will now deal with the technical issues, following which we will address the broader implications raised by her paper, notably regarding the application of functional analysis and the relationship between tools for records management and information retrieval.

For the purposes of this part of our paper in particular, we refer in the text to Dr Exon's paper, as published in the Perth convention proceedings, in the form (Exon, p. x) and to Keyword AAA in the form (Keyword AAA, Introduction, p. x). We deal with other references as endnotes.

Compliance with the ISO Standard 2788

A major issue which Dr Exon raises is the extent to which Keyword AAA is consistent with the International Standard ISO 2788, Documentation - Guidelines for the establishment and development of monolingual thesauri. Her criticism of Keyword AAA and its relationship to ISO 2788 focuses on three main areas:

  • Keyword AAA's use of codings
  • pre-coordinate indexing, and
  • hierarchical relationships between terms.

There is an implication in several places in Dr Exon's paper that Keyword AAA claims to comply with ISO 2788 to a greater extent than it does in fact. It is therefore important first of all to make clear to what extent and in what ways Keyword AAA complies with ISO 2788.

The introduction to Keyword AAA (Keyword AAA, Introduction, p. 2 and p. 5) makes it clear that it conforms to the conventions of the International Standard and that we have revised the rules and brought structure and codings into line with ISO 2788. Finally, the Codes section of the introduction indicates that we have used the codes in ISO 2788 (BT, NT etc.), noting that they are not abbreviated in Keyword AAA (Keyword AAA, Introduction, p. 12). This is the full extent of our claim to compliance with ISO 2788. Of more importance to us is Keyword AAA's compliance with the Australian Standard AS4390, Records Management, to which we will return later.

We note, however, that ISO 2788 is intended to be applied flexibly, as suggested in the Scope and field of application section of the standard. This states that:

 … the recommendations set out in this International Standard are intended to ensure consistent practice … they should not be regarded, however, as mandatory instructions … the choice of procedure will vary from one indexing agency to another, depending on management decisions that fall outside the scope of this International Standard."

A key example of the flexibility required when applying the standard to a function-based thesaurus is in the hierarchical relationship of terms, which Dr Exon discusses at length (Exon, pp. 101-102). We address this issue specifically later in this paper.

Another example is our spelling out of 'Broader Term', 'Narrower Term' and 'Related Term' instead of their abbreviated codes (BT, NT and RT) prescribed in the standard. Dr Exon refers to this in passing (Exon, p. 100) without indicating whether or not she approves. It is, however, an important point. Thesauri envisaged by ISO 2788 are designed primarily for use by librarians and other information professionals. With the devolution of many operational tasks, especially in the electronic environment, a records management thesaurus is as likely to be used by an action officer as by a records manager. Ease of use for all types of users, not just those familiar with other thesauri, was a priority in the design of Keyword AAA. Complying with the letter of the standard would not have helped here.

Dr Exon argues that

'the change to a format based on ISO 2788-1986(E) … has artificially disguised the fact that keyword file titling has a very different purpose from post-coordinate indexing'.
(Exon, p. 103)

We used the structure, layout and coding conventions because we believe that a records management thesaurus should comply with national and international standards to the fullest extent that they can be applied. ISO 2788 is highly applicable to Keyword AAA in terms of the structure, codings and layout that it advocates. Keyword AAA really would have ' … cut itself off from the approaches used by other sectors of the information industry … ' if we had decided, as she suggests, to ' … structure [indexing languages] in ways which are clearly distinguished.' (Exon, p. 103) Keyword AAA is, after all, a monolingual thesaurus.

In one respect, Keyword AAA cannot, and was never intended to, comply with ISO 2788. Keyword AAA is a function-based thesaurus and ISO 2788 is concerned with subject-based thesauri. Function-based thesauri are a phenomenon of records management, not of library science, which is the tradition in which ISO 2788 was developed.

We do not accept that anyone will take the use of the structure, coding and layout conventions in ISO 2788 to imply that Keyword AAA complies with the standard in ways that it clearly cannot and should not. This is an important issue, not just for Keyword AAA, but for any records management thesaurus using function-based terms, which means any records management thesaurus seeking to comply with AS 4390.

Pre-coordinate and post-coordinate indexing

An important part of Dr Exon's arguments about the relationship of Keyword AAA with ISO 2788 centre around her assertion that Keyword AAA is a pre-coordinate system. She states:

'The ISO standard definitely states that it is a standard for the construction of post-coordinate indexing languages.'
(Exon, p. 100).

To coin a phrase, there is a real mystery here. Clause 1.2 of ISO 2788 states that the techniques described ' … are not limited to a particular method of indexing, whether post-coordinate or pre-coordinate.' (our emphasis)

To support her argument Dr Exon presents an extract from the introduction to ISO 2788 which discusses a posteriori relationships

(' … relationships between the terms which together summarise the subject of a document … [but] … are not normally associated according to common frames of reference … ')

and a priori relationships

(' … thesaural relationships between terms assigned to documents and other terms which, because they form part of common and shared frames of reference, are present by implication'). (Exon, p. 106)

She asserts that the standard is

' … in no way concerned with the problems of citation order in pre-coordinate languages.' (Exon, p. 100)

Unfortunately for you, dear reader, this requires closer examination.

The crux of Dr Exon's argument seems to be that Keyword AAA is a pre-coordinate language (it relies on citation order) with a posteriori relationships between its terms, while the codings BT, NT etc. concern thesaural or a priori relationships. Therefore, Keyword AAA should not use these codings to describe the relationship between terms at its different levels.

Keyword AAA is actually by this description both pre-coordinate and a priori (Dr Exon seems to believe that the two are mutually exclusive). Keyword AAA does prescribe the relationship between terms but it does so in a hierarchical fashion with terms that form part of a common and shared frame of reference. The introduction to ISO 2788 notes:

The International Standard is especially concerned with those priori relationships which can be displayed in a thesaurus, where they then, in effect, add a second dimension to an indexing language".

This is exactly what Keyword AAA achieves with the application of keywords and activities to a file or document titles.

A close reading of other parts of the Standard also reveals that it can cover more than post-coordinate a priori relationships. For example, Section 3.3 of ISO 2788 defines a thesaurus as a priori but does not state that it has to be post coordinate and Section 6.3.1 discusses the role of pre coordinate methods in the choice of singular and plural terminology. It states that

 … terms selected from a thesaurus are organised into index entries in such a way that the entry as a whole expresses a subject in summary form. Relationships between terms may be conveyed in various ways, for example by word order and/or by the choice of special typography and punctuation."

The use of pre-coordinate indexing in Keyword AAA instead of post-coordinate indexing was deliberately adopted so that the file title would represent the business classification scheme. The first term is always a keyword (function), the second an activity descriptor (activity) and at the third level subject descriptors or free text which can be added to represent subjects (for further explanation on the use of hierarchies within Keyword AAA see below). Therefore, all titles move from the broad business function to the specific topic/subject. By retaining this order, the classification scheme can be used for other recordkeeping management activities such as the appraisal of records. If terms were able to be mixed and matched in a post-coordinate way the classification scheme would no longer be visible, and the benefits for other recordkeeping activities nullified. Keyword AAA would no longer be valuable for use with directory structures for electronic records either, one of its current advantages.

The disadvantages of pre-coordinate indexing for retrieval in comparison to post-coordinate models are largely overcome with modern records management software. Records management software packages index all terms used within record titles, so that records users are able to search on any of the terms to retrieve the required document and record. Therefore it does not matter that the title's order is specified. This is also the situation with document management software, where increasingly full text retrieval mechanisms such as ISYS are being used for the retrieval of documents regardless of classification schemes.

Hierarchies

Dr Exon criticises our description of Keyword AAA as having a hierarchical relationship between keywords and activity descriptors, in one place describing it as 'misleading' to describe the relationship between a keyword and an activity descriptor (the example she uses is Travel and Procedures) as hierarchical. (Exon, p. 102)

Dr Exon starts by noting that

' … Hierarchies lie at the heart of the guidelines embodied in the International Standards Organisation (ISO) standard referred to in Keyword AAA … so these two statements [i.e. bringing the structure and codings in Keyword AAA in line with ISO 2788 and Keyword AAA having a hierarchical relationship] are linked in meaning and import.' (Exon, p. 100)

This connection is misconceived. We do not claim that the hierarchy in Keyword AAA derives from ISO 2788. On the contrary the introduction to Keyword AAA states that 'Classification schemes are normally hierarchical in nature, with a tree like structure moving from the broadest to the narrowest aspect. This is the essence of the principle of hierarchy.' (Keyword AAA, Introduction, p. 3) The hierarchy in Keyword AAA is based on the concept of the business classification scheme, as described in Part 4 of the Australian Standard.

There is hierarchy from the broad to the specific in Keyword AAA. The Collins Dictionary simply defines a hierarchy as 'a system of persons or things arranged in a graded order.'

In records management, a given activity can be identified as a part of a function: they are clearly not on the same level in the business classification scheme. If we turn to Schellenberg's model, we see that a function normally is comprised of a number of activities. An activity, in turn, is comprised of a number of types of recurring transactions relating to different topics, people or things. We describe the business classification scheme as hierarchical and the relationship between the terms in a records management thesaurus reflecting that scheme as equally hierarchical.

In many cases, this relationship is 'one-to-many'. In such cases, for example, a function has a number of activities, but an activity belongs to only one function. For example the activity descriptor ACCOUNTING is linked only to the function of FINANCIAL MANAGEMENT, but FINANCIAL MANAGEMENT has a number of activities. This corresponds most closely to the 'whole-part' form of hierarchical relationship described in ISO 2788.

In some cases, however, the relationship is necessarily a 'many-to-many' one. Policy development, for example, is an activity involved in many broader functions. In Keyword AAA this is reflected in a many-to-many relationship between the terms: POLICY as an activity descriptor is not limited to one keyword. Here the relationship reflects that described in another part of ISO 2788. These more complex hierarchies present in Keyword AAA reflect the realities of business activity and the records produced in the process. We believe it is more important to reflect these realities than to impose limitations simply in order to conform to ISO 2788. We discuss the business classification scheme further in the following section.

It is clear that a fundamental cause of difference between us is the use of functional analysis to produce a business classification scheme. This colours the argument about hierarchies in Keyword AAA. To return to the example that Dr Exon describes as misleading: the a priori indexer in Dr Exon says ' … procedures are neither a kind of travel nor a part of travel … '; the records classifier in us says 'Travel is a business function and the activity of developing procedures for administering travel is a part of that function.' We will now examine why this perspective is so important in records management.

Functional analysis

Dr Exon was correct to note that Keyword AAA is different to the GADM Thesaurus. It has a very different basis. Keyword AAA is based on a classification scheme which reflects a hierarchy of business functions, activities and transactions. This allowing records, through the recordkeeping system, to be linked directly to the business functions and activities which resulted in those records. The GADM, as we have noted, was a subject-based thesaurus. Its terminology, at the keyword as well as descriptor level, described subjects or 'things' rather than representing concepts of functions and activities undertaken within government agencies. It was not based on a classification scheme, nor did it have a close relationship to business functions and activities.

How did this fundamental difference between the GADM and Keyword AAA come about? The GADM Thesaurus had been regularly revised since it was first published in 1978. The revision undertaken in 1994/1995, however, was influenced by developments taking place in recordkeeping theory and practice, notably the use of functional analysis, the development of the Australian Standard on records management and a fundamental rethinking of how recordkeeping systems are designed and implemented.

Australian recordkeeping professionals, that is, both records managers and archivists, have been strongly influenced in recent years by our colleagues in North America, who have been examining recordkeeping systems and articulating the use of functional analysis for the appraisal of records, and specifically the management of electronic records. A leading commentator noted in 1993 that the

advent of electronic records which are not susceptible to ready examination … has led archivists to seek alternative approaches to appraisal. It was soon realized that if archivists could make such decisions on the basis of analysis of the business functions and the need for evidence of these functions, they could avoid trying to assess records themselves."

At the same time, at the University of Pittsburgh, a project established in 1991 to study 'recordkeeping functional requirements for electronic information systems' clearly identified the nexus between recordkeeping systems and the business functions of an organisation and articulated a methodology for identifying business functions.

Business classification schemes

In Australia, early discussion of functional analysis related particularly to its use in the appraisal and descriptive control of records. The cumulative experience of recordkeeping professionals in its use found expression in the Australian Standard on records management, released in 1996, representing current best practice in records management. The standard uses functional analysis methodologies, in particular in Part 3: Strategies, Part 4: Control and Part 5: Appraisal and Disposal. In Part 3, Clause 6.2 outlines a methodology for designing and implementing recordkeeping systems. It uses an analysis of business functions and activities to ensure that the records captured and maintained by the recordkeeping system are 'linked to the activities that they [the records] document.' The standard recommends that the first three steps undertaken in this methodology can be used to 'serve a range of records management purposes' including the development of business classification schemes.

The standard provides further guidance on the development of business classification schemes in Part 4: Control. It recommends that classification schemes used to manage records should not be based on the organisational structure, but rather on an analysis of business functions and activities. The standard notes that functions and activities are more stable than organisational structures. An example of this type of classification scheme is that developed by the Roads and Traffic Authority of NSW to control its records. The Authority has built a business classification scheme on functional analysis undertaken for corporate planning, using the 76 activities identified for its activity-based management system.

The Australian Standard also recommends that the business classification scheme should be hierarchical, with the top level representing broad business functions, lower levels representing the activities undertaken to accomplish those functions, and ending with the bottom level to represent the recurring transactions that take place within these activities. This is nothing novel. It was described with admirable clarity by T.R. Schellenberg forty years ago. Schellenberg represents the classification scheme thus:

How do business classification schemes relate to records management thesauri? As the Australian Standard notes, the

 … business classification scheme reflects the terminology used to describe functions, activities and transactions in use within the specific organization. This terminology can be applied to the naming of records within the organization."

Thus, a records management thesaurus is a product of a classification scheme, rather than a basis for a classification scheme. As part of the development cycle for Keyword AAA, we devised a generic classification scheme of general administrative functions, activities and recurring transactions that are common across the NSW Government.

Part of Dr Exon's concern about function-based thesauri is that they are difficult to merge with subject-based thesauri. To pursue this argument she notes that

' … any organisation wishing to use this thesaurus [Keyword AAA] for records management must add to it terms to cover the functional aspects of the organisation. It might be considered that such terms could be derived from a published thesaurus in the relevant area … .' (Exon, p. 102)

As students in our keyword training courses learn very early, a functional thesaurus is derived first and foremost from a business classification scheme, as described in the Australian Standard and developed through an analysis of business activity. A functional thesaurus developed using the methods described in Part 4 of the Australian Standard is easily merged with Keyword AAA.

It is, of course, possible to add terms from a subject-based thesaurus, but only after the completion of the analysis of the organisation's functions. Terms taken from published subject-based thesauri will generally be useful at the subject/topic level of the functional thesaurus. This reflects Schellenberg's notion that this level of classification relates to persons, corporate bodies, places and topics - facts, events, ideas.

Retrieval using functional analysis

Dr Exon argues that

'the current emphasis on functional analysis has been to the detriment of efficient retrieval'. (Exon, p. 103)

Chris Hurley, on the other hand, has observed that retrieval via function could assist records users far more than retrieval using provenance, as functions appear to users to be more subject oriented than administrative structures as context. We reserve judgement on this issue, as we believe that there is as yet too little evidence to form a conclusion. Clearly we need empirical evidence, making this is a topic to add to the recordkeeping community's research agenda.

It may also be that views about the impact of functional analysis on information retrieval come from a limited perspective on the ways in which functional language is being used by recordkeeping professionals to describe records. To help put the discussion into perspective, it is worth noting some of this work.

There are many notable examples of functional analysis being used for the contextual control of records. These include:

  • the Australian Archives' thesaurus of functional terms, used to develop a functional index for retrieval in its databases of record holdings
  • the PIVOT project at the National Archives of the Netherlands
  • functional description in the control systems of the Public Record Office of Victoria
  • the incorporation of functions and activities by Mark Stevens into the series system of control at the Sydney City Council.

At the Archives Authority of New South Wales we have embarked on a major context project to document functions across the NSW Government. The project uses a three level model of functions:

  • government functions, reflecting the most ambient of functions such as law and order, health, and education
  • agency functions, reflecting how those high level functions are broken down and pursued as large organisational functions, for example, the government function of law and order is carried out through the agency functions of corrective services, law enforcement, courts administration etc.
  • business functions, conducted within agencies in the pursuit of the goals and strategies of the organisation and its agency functions.

These business functions represent the top level functions identified by agencies when developing their business classification scheme using the methods described in the Australian Standard. As we have already noted, these kinds of functions are reflected in Keyword AAA and agencies' functional thesauri. This model provides, for the first time, the opportunity to combine a government-wide business classification scheme with those developed within individual agencies, permitting, in theory at least, searching for records using functional terms across the entire documentary output of a government.

Recordkeeping professionals need to examine how we manage records across time with the necessary information relating to their context, particularly in the electronic environment. There are obvious benefits to be drawn from linking records directly to functions. Functions are regarded as more stable than organisational structures. Hurley notes that

when functions are treated as units of description to establish relationships they operate more like provenance statements than indexable headings."

Thus, if records are encapsulated with metadata (information about records) relating to business functions through the application of particular titling regimes, then this encapsulation process would result in greater contextual control of records now and into the future. As Hurley notes, understanding records now and in the future (and as electronic records) requires contextual control and knowledge, that is, to understand a record, we need to ensure that it has the appropriate contextual control.

Managing records using functional analysis

The Australian Standard on records management makes clear what records managers already know: that functional analysis provides an invaluable foundation for a wide range of records management functions. Dr Exon recognises this. (Exon, p. 103) Where we differ is in how important this is and in the extent to which records management and information retrieval needs are mutually exclusive. We do not believe that they are.

In our case the use of functional analysis extends to the management of records across government, in addition to agencies using it for the management of their own records. The functional structure in Keyword AAA has been used in the most recent edition of the Authority's General Records Disposal Schedule - Administrative Records, issued in June 1996, which applies to all agencies governed by our current legislation.

In addition, the identification of government-wide functions will enable us to take more of a strategic approach to appraisal across our jurisdiction. The end result of our functional analysis work will be better management, as well as better understanding and control, of records across the NSW Government.

Activity descriptors

Dr Exon was critical of the second level descriptor used in Keyword AAA,

' … as it places in an important position in the file title terms which are often not helpful for retrieval purposes and which add very little to the total effective meaning of the file title as a description of the content of the file.' (Exon, p. 104)

This second ('activity') level is important in the classification scheme and consequently in the thesaurus. To remove the activity level within the classification scheme would mean removing a critical part of the hierarchy of business functions, activities and transactions. If this level is removed, not only the hierarchy no longer be representative of the functions and activities taking place in an organisation, but there also is no clear understanding of what activity was being accomplished within the function. Thus, trying to understand records within the context of business functions and activities is made more difficult.

In our experience the activity descriptor is useful for retrieval purposes, as users will often remember the business activity that they have been involved in documenting and will retrieve records on this basis. We also find that the use of the activity descriptor in file titles lessens the tendency towards 'bag files'.

We disagree with Dr Exon that

' … concrete terms which are likely to be sought are relegated to a lesser role at the end of the file title.' (Exon, p. 104)

The subject descriptor does not play a lesser role in the Keyword AAA model, merely a different one. Dr Exon's problem with printouts of file titles in alphabetical order is a problem with this kind of report for any structured titling system. It is solved by producing KWOC indexes.

Appropriate tools for managing records

Dr Exon contends that

' … records managers may find themselves on such a different track from other information professionals in the tools which they use for indexing that their ability to respond to increasingly integrated information handling will be reduced.' (Exon, p. 96)

Later she notes that

'We live in a converging world … .There needs to be increasing cooperation between different types of information professionals.' (Exon, p. 103)

The sub-text here is the relationship of records management and information management. Here and elsewhere in her paper, Dr Exon seems in danger of taking what Sue McKemmish has termed the unitary (as opposed to pluralist)view of information management.

Briefly, the unitary view of information is based on definitions of recorded information directly relating to format, with the distinction between records and other forms of information limited to whether they have been published or not. Crudely, records are seen as just another kind of 'non-book material'. The pluralist viewpoint, on the other hand, perceives records as the by-product of business activity, as opposed to 'consciously authored materials'. Within this view there are separate and meaningful roles to be played by both the information community in general and the recordkeeping community in particular in ensuring that all forms of recorded information, and records specifically, are managed and made accessible in their respective, and overlapping, regimes. These perspectives colour much of the discussion about the relationship of records management and information management.

The Archives Authority supports a pluralist view of information management. We recognise that there are unique and differing roles to be played within information management by librarians, recordkeeping professionals, data administrators and IT professionals. These unique and differing roles are necessary within information management because of the requirement to manage different types of information. At the same time, we fully support calls such as Dr Exon's for people playing these distinct but complementary roles to work together. Convergence means collaboration.

Crucial to the unitary/pluralist debate as it affects records management is an understanding of what records are and do. In essence, the unitary view sees records merely as an information resource. The pluralist view sees them also as evidence, in the broad sense of the term, of business activities.

The definition provided in the Australian Standard on records management recognises this evidential quality of records. Records are

recorded information, in any form, including data in the computer systems, created or received and maintained by an organization or person in the transaction of business or the conduct of affairs and kept as evidence of such activity."

Equally the Australian Standard emphasises the information role of records, viewing a key function of records management as being

 … Managing records as an asset and information resource … ."

This understanding of the dual role of records has an impact not only on how we retrieve records, but on why. As David Roberts noted in a recent review of proposals for an information management framework in the Commonwealth Government,

… the recommendations arguing for information retrieval tools providing integrated, seamless access to information resources, including records, deserve the strongest support. But when it is records, capable of functioning effectively as evidence, that people need, they must be able to retrieve records."

We need a new model for records retrieval recognising the duality of records as evidence and information. Among leading commentators, Sue McKemmish of MonashUniversity, in particular, has started the necessary dialogue.

Records are more than an information resource. They support business activities within the 'business domain', they 'are an indispensable ingredient in organisational accountability', and are a resource available to society to account for collective behaviour within the 'cultural domain'.

Thus any management regime for records needs to recognise that there are specific requirements for managing evidence of business activities over time to ensure that it is useable, reliable, authentic and accessible. For this reason we need to ensure that the tools we employ to manage records allow us to ensure that records retain their 'recordness' and integrity over time. A functional approach to classification schemes and thesauri is an important tool to help manage records as evidence, not just as information resources.

A continuum approach

Dr Exon discusses Keyword AAA in the context of titling records within an filing system for active records. While Keyword AAA is used for titling active records, it is designed to be used to assist in the management of records across the records continuum. Thus, for example, by assigning a record to a classification, it is possible to make decisions about how the record should be handled and stored throughout their existence and who should have access to them. This kind of tool helps us manage records better and manage them, where necessary, as archives better.

We need to view records management and archives management as an integrated recordkeeping process, thus activities undertaken within records management will and do have an impact upon management strategies for records through the rest of their existence. Thus using the Keyword AAA thesaurus to title records, regardless of format or technological environment, establishes a regime of classification and control which will assist in the management of those records as they are organised and pluralised through the records continuum. By establishing such a regime of classification and control from the design of recordkeeping systems, we can ensure that the metadata required to understand records can be migrated with the records.

Metadata

Dr Exon criticises Keyword AAA for taking a different approach to indexing compared with the rest of the information community. We believe that tools like Keyword AAA can be married to the subject-based approaches taken by the rest of the information community to produce useful metadata for records and other kinds of information. An example is the Dublin Core metadata set, featuring thirteen metadata elements designed to assist in retrieving document-like objects in networked environments, particularly the Internet.

The Dublin Core metadata set is designed primarily for authored materials but it is possible for particular elements within the set to be used for function-based retrieval of records. For example, free text or keywords may be used with the Subject/Keyword Element and it is possible 'to flag that you are using authorised subject terms or codes from a particular scheme....' For records, this can be terms from a function-based thesaurus.

If a common thesaurus across government is used to identify government business functions and activities, not only are records pluralised and accessible within the collective memory along with other types of recorded information, but, also, the metadata required for contextual control of the record is also present.

We see the incorporation of function-based searching into government information locators and similar tools, using these approaches to defining metadata, as one of the most exciting and substantial benefits of functional analysis. The Commonwealth Government has already identified the need for a common core thesaurus to form

… part of the infrastructure ensuring quality metadata supporting the proposed AusGILS development …"

and has identified Keyword AAA as the likely source.

Working together

In state and federal government jurisdictions around Australia collaboration between government records authorities and central agencies concerned with government information management as a whole is increasingly important. In the NSW Government, the Archives Authority is committed to collaborative efforts with the NSW Office of Information Technology so that better use of all government information, including records, is achieved. Among other things, this includes active involvement in the Office's Information Management Working Party, significant efforts to align the Archives Authority's government-wide records management policies, standards and guidelines with the Office's information management policies and guidelines, and promoting a regular and fruitful dialogue between government records and information technology managers.

We see this as a tangible example of working in a pluralist view of information management: we are working with others in the information community and ensuring that recordkeeping requirements are addressed and that recordkeeping tools such as Keyword AAA are linked to government-wide endeavours to provide better access to government information.

Conclusion

Dr Exon's paper raised many issues and we have welcomed the opportunity to discuss some of them, despite the manner in which that opportunity arose. The technical issues can be complex and difficult to grasp. We hope, however, that we have demonstrated that, from the perspective of records management theory and practice, Dr Exon's criticisms of Keyword AAA are without substance. At the same time, we hope that we have shed more light on why Keyword AAA developed as it did and how it fits into our broader strategies for improving government recordkeeping and contributing to better management and use of government information of all kinds. Finally we hope that we have shown that, far from cutting ourselves off from the rest of the information community, we have a vision for collaboration which recognises the nature and purpose of records as evidence and information and makes the best use of our diverse strengths and skills.