Information Management Classification Guideline

Department of Commerce Guidelines Information Management - Classification Guideline Issue No: 1.0 Issue Date: Sept 1997 Review Date: May 2002 Table of Contents 1. 2. 3. INTRODUCTION ................................................................................................2 OBJECTIVES AND SCOPE ...............................................................................3 WHAT IS INFORMATION CLASSIFICATION? .................................................4 3.1 3.2 3.3 3.4 Purpose of Information Classification.................................................................................. 4 Definition of Information Classification ................................................................................ 4 Formal and Informal Classifications .................................................................................... 5 Scope of Information Classification ..................................................................................... 5 4. INFORMATION CLASSIFICATION PRACTICES ..............................................7 4.1 4.2 4.3 4.4 Dimensions of Classification ............................................................................................... 7 Designing Classification Models.......................................................................................... 7 Managing Classification Models.......................................................................................... 8 Applying Classification Schemes ........................................................................................ 9 5. INFORMATION CLASSIFICATION CHECKLIST ............................................10 5.1 5.2 Pre-design: ........................................................................................................................ 10 Post-design:....................................................................................................................... 10 APPENDIX 1 .................................................................................................................12 A B FORMAL INFORMATION CLASSIFICATION ................................................................. 12 INFORMAL INFORMATION CLASSIFICATION.............................................................. 13 APPENDIX 2 – THE DIMENSIONS OF CLASSIFICATION..........................................14 A B C RANGE.............................................................................................................................. 14 SPECIFICITY .................................................................................................................... 14 CONSISTENCY................................................................................................................. 15 APPENDIX 3 – DESIGNING CLASSIFICATION MODELS ..........................................16 A B C DESIGN STEPS ................................................................................................................ 16 CLASSIFICATION SCHEMES FROM DATA MODELS .................................................. 17 CLASSIFICATION SCHEMES AND METADATA ........................................................... 17 Table of Figures Figure 1: The Information Management Framework ............................................................................ 2 Figure 2: Fundamental Characteristics of Information and supporting Information Management ............. 3 Information Management - Classification Guideline Page 1 of 18 1. Introduction This Guideline is one in a series on the management of Information Management and Technology (IM&T) within Government agencies in New South Wales. It is conceived and developed as a fundamental Guideline for managing information as a strategic resource, in line with NSW IM&T Blueprint Strategy. Effective information management ensures that the value of an agency’s information is identified and exploited to its fullest extent, to serve the needs of the agency’s business, both strategic and operational. It also ensures that agencies address information management planning at the earliest point in the business planning cycle, to provide a more strategic and coordinated approach to corporate decision making for whole-ofGovernment outcomes. This Guideline takes as its starting point, and its context, the Information Management Framework, which describes the lifecycle functions and characteristics of Governmentheld information. Figure 1 shows the scope of this Framework. Figure 1: The Information Management Framework Information Management - Classification Guideline Page 2 of 18 2. Objectives and Scope The purpose of this Guideline is to advise agencies on the design and application of Information Classification to their agency. Categorising of information for storage and retrieval purposes as practised by Records Management, Library Management and indexing professionals are well understood and well managed procedures. This guideline aims to extend these practices to the design and application of classification schemes for information holdings within all business systems of an agency. This Guideline provides direction for business units when planning their business systems to include planning and design of classification schemes that will provide flexible and meaningful information based on business needs. Traditional information management disciplines, such as library management, effectively use classification schemes to categorise both electronic and non-electronic information holdings. For example the Dewey Decimal and other classification schemes (at the physical level) and the use of controlled subject headings (at the conceptual level), are two distinct activities practiced by librarians. The use of classification schemes such as in assigning subject headings to documents using standard indexing methods eg, Medical Subject Headings (MeSH) are addressed in the Indexing Guideline. This Guideline is part of the set of Guidelines to do with the Definition of agency information. See Figure 2. It is best read in conjunction with the Information Management Inventory Guideline, which describes how information will be identified, based on a set of classifications relevant to an agency’s business processes. Inventory Classification DEFINITION Management Quality Custodianship Copyright QUALITY Management OWNERSHIP Management Integrity Liability Licensing ACCESSIBILITY Management SENSITIVITY Management Pricing Indexing Privacy Confidentiality Figure 2: Fundamental Characteristics of Information and supporting Information Management Information Management - Classification Guideline Page 3 of 18 3. 3.1 What is Information Classification? Purpose of Information Classification The purpose of classifying information is to enable it to be stored in an organised manner to facilitate timely, easier access to the right information required to support a business process. More generally, the purpose is to enable information to be managed in a manner that is relevant to the agency’s business goals and objectives. Information classification makes it easier for users and business units to manage large volumes of information, both electronic and non-electronic. The categories into which information is classified are dependent on the business need for grouping the information. Traditionally, physical holdings of information such as files, books and journals, have been classified by functional and hierarchical schemes developed by records and libraries management specialists. Powerful locating and searching technologies have since emerged to assist users in more efficiently accessing large volumes of electronic information. Consistent classification and proper indexing of both electronic and non-electronic information and its intellectual content more efficiently and effectively support the business processes of the agency. Textual electronic storehouses of information can be searched using search engines which match (to a greater or lesser extent of sophistication) search terms to text strings in the storehouse. This will recover information items having that text string, but it will not recover items having the same conceptual information, that is, on the same subject, that do not contain that search term. For this to occur a classification scheme containing controlled terms, usually referred to as "keywords", is required. Examples of these schemes include the thesauri used with bibliographic databases such as Engineering Index (EI) and Medical Subject Headings (MeSH). Classification is of critical importance in determining the value of a piece of information or an information holding to an agency’s business. Classifications can be applied to data items within a storehouse (especially applicable to computer databases) and to the storehouse (database, holding, document, record; especially applicable to archives, record management and libraries). 3.2 Definition of Information Classification A formal definition of classification comes from Harrod’s Librarians’ Glossary: “Any method of recognising relations, generic or other, between items of information, regardless of the degree of hierarchy used and of whether these methods are applied in connection with traditional or computerised information systems”. This Guideline proposes: Information Management - Classification Guideline Page 4 of 18 “Information Classification is the act of grouping related information, determining the values that the grouping may have and assigning standardised descriptions to the values for practicable storage, retrieval and analysis”. 3.3 Formal and Informal Classifications Information classifications may be informal, relying on descriptive text which has no pre-determined or limited content, or may be formalised as “codesets” or “classification schemes”. Put simply, informal classifications are descriptive; formal classifications are prescriptive. Formal classification schemes are common in all information-related disciplines, and are required for successful automated processing of information. A classification scheme is a finite range of pre-determined, defined logical values (or meanings, or descriptions), that may be assigned to a conceptual or physical item. To each of these allowable logical values, an arbitrary distinguishing code is assigned which represents the logical value. This distinguishing code consists of letters and / or numbers arranged in some manner convenient to the agency, discipline and / or technology used (for example the number 540 for Chemistry in the Dewey Decimal Classification). In simple schemes the code may be the text of the logical value itself (for example, "sedan" in a classification of vehicles), but this would be unusual. The effect of a formal classification scheme is that exactly the same descriptive text is always used to refer to the same underlying concept (for example, a Name Authority File, where BHP is always "BHP" and not "B.H.P."). Appendix 1 describes formal and informal classification schemes in more detail. In this Guideline, the term “classification scheme” is used to refer to a formal classification scheme. The guideline is concerned with the design and management of formal classification schemes. The totality of classification schemes within an agency, and their relationships with each other, is a classification model. A library subject catalogue corresponds somewhat to a Classification Model. In structure a subject catalogue is similar to a hierarchical database schema. 3.4 Scope of Information Classification The scope of this Guideline is formal classification systems, as described in Appendix 1.1. The Guideline does not, however, attempt to cover the content of formal classifications as practised by librarians and records managers, who have their own professional standards and procedures. However, much can be learnt from these areas. There are three aspects of information classification covered by the Guideline: • Design, that is the decision as to what classification schemes will be used; Information Management - Classification Guideline Page 5 of 18 • Managing and administering classification schemes; • Usage, or the act of classification, of a particular piece of information. This is addressed more fully in the Indexing Guideline. These aspects are described in Section 4 below. The Inventory Guideline contains a description of the Conceptual Inventory of information for an agency. It outlines a structure for conceptual information, where information is classified as belonging to different organisational levels, referred to as Corporate, Strategic and Tactical. Information Classification schemes and structures may exist at all of these levels. The Classification Model in an agency exists within this context. To make sense of the information collected in an information inventory, a set of classifications concerning the location and description of the items in the inventory should be undertaken as described in the Inventory Guideline. Information Management - Classification Guideline Page 6 of 18 4. 4.1 Information Classification Practices Dimensions of Classification The design of classification schemes occurs during Requirements Definition planning, part of the IM&T strategic planning cycle. Classification activities can be performed during collection, storage, access, use and disposal. “Classification dimensions” are an aid to developing or assessing classification schemes. They provide a conceptual framework for making judgements or tradeoffs on the size and complexity of a classification scheme, or a collection of related classification schemes (a classification model). They may be used to assist with the design of classification schemes and models to optimise their use. Appendix 2 describes the three dimensions of classification: • Range; • Specificity; • Consistency. 4.2 Designing Classification Models The importance of the design of classification schemes and classification models is often not appreciated, and little time and other resources are allocated to it. Yet the impact of decisions on information classification may be immense, long lasting, and some times ineradicable. A good example of a successful classification model, which has lasted for hundreds of years, is the Linnaean scheme for classifying the biological world. In current computerised systems, storehouses consisting of millions of records may become useless, their information effectively lost, because the way the information is classified is inappropriate, or because classification schemes that should have been applied were not. The World Wide Web is presently an example of an information resource suffering from the absence of a classification model (and a technology to apply one consistently). On the World Wide Web, search tools such as Yahoo, Lycos and Alta Vista have, until recently, worked by indexing the contents of all web pages. A civil engineer searching on, say, the word ‘Roads’ would find all web pages which contained “roads”, however incidentally (such as “all roads lead to Rome”) or whatever the relevance (for example the text of songs which contain the word “roads”). Yet the search would fail to find possibly relevant pages which contained the words, “highway”, “motorway” or “autobahn”. Without a classification scheme for web pages, too much irrelevant information is found, yet the results of a search may still be incomplete. In principle, the web is like a library of information, but lacks the standard search and access tools, based on classification methodologies long familiar to librarians. At present, there Information Management - Classification Guideline Page 7 of 18 is evidence of some limited attempts to group web pages into broad subject categories for searching. (The dynamic, highly volatile nature of the web is a major barrier to successful classification.) Agencies wishing to publish on the web, for example making available technical or scientific papers and reports, should classify them in a manner that facilitates access, rather than relying only on web text searching facilities. (For example, document classification subject headings could be set up as ‘html’ hyperlinks which lead to the relevant documents.) The classifications should be designed according to well-established principles of librarianship. For large agencies, such documents may already have been correctly classified internally, in which case these classifications should be used. Appendix 3 describes the design steps and the relationship of classification schemes and data models and metadata. 4.3 Managing Classification Models Every classification scheme should have a documented business purpose for the context in which it is applied, since collecting the information and classifying it has a cost. For example, if an industry classification scheme exists to classify business organisations in various ways, its use in any particular agency system should have a particular purpose, such as to enable statistics to be derived on how many organisations are operating in particular industries at particular locations. Every classification scheme should have a custodian, or controlling body, which has the sole power to modify the range of allowable values in the scheme. Some particular classification schemes will be more important than others, depending on the agency. Information should be classified in relation to the Information Characteristics defined in the Information Management Framework. These are Definition, Ownership, Sensitivity, Quality and Accessibility. Critical classification schemes should be designed and managed with the close attention and support of senior agency management. These will include classification schemes designed to support Information Inventory and Information Audit. Potentially, every item of information identified by the Information Inventory should have at least one classification scheme. Business rules concerning relationships between classification schemes should be documented. For example, the fact a particular value is selected from one classification scheme restricts the range of values that may be selected from another, or controls which of several alternate schemes are used. The occurrence of duplicate, overlapping, incomplete and incompatible classification schemes, or codesets, in an agency is a frequent source of problems in information management. Information Management - Classification Guideline Page 8 of 18 Note that it is not the fact that particular symbols, or codes, are the same in different classification schemes, but that the meanings, or descriptions, which these symbols represent may overlap or conflict. In such cases, valid comparison or translation of data is very difficult or even impossible. As pointed out in Appendix 3.2, no new business system should be established without a survey of existing classification schemes, to determine if appropriate classifications exist which can be re-used. In the Information Audit, special emphasis should be made on identification, evaluation and control of classification schemes. 4.4 Applying Classification Schemes Apart from some automated processes, the application of a classification scheme always requires some judgement by a human user. Success in classification therefore depends on the user being both willing and able to use it. The culture of the agency needs to be acknowledged. In the context of Records Management and Library Management, the thesaurus is a well established form of classification scheme usually established to assist in the retrieval of records and library material as part of the agency functions of Records Management, Document Management or Library Management. The library thesaurus may be industry specific (for example an engineering, administrative or legal thesaurus) or may be developed as a corporate thesaurus for an agency, usually under the custodianship of the agency’s Library manager. The records thesaurus may be a subset of the corporate thesaurus or may be based on a function analysis of the agency, usually under the custodianship of the agency’s Records manager. For general business and information systems purposes, a classification scheme may be implemented by automatic procedures, for example, where information is classified by its source, or is in a range of values to which classifications can be assigned (an order of more than 20 tonnes of sand is classified as “bulk” and attracts a discount, for example). Typically, information classification in a computerised business system prompts the user to select from a list of allowable meanings, that is, the classification values, and not the actual code symbols that represent these values. In practical terms, this is often displayed in a “pull down” list in graphical user interface environments. Where a classification scheme is implemented by automatic procedures, a general proviso applies: it is highly preferable that the contents of “pull down” lists are initialised from separately maintained parameter files, not from the source code of the software itself. Information Management - Classification Guideline Page 9 of 18 5. 5.1 Information Classification Checklist Pre-design: Has a survey been performed of the data to be classified, to ascertain its required range, the members of the range, and required specificity? Does your agency have to exchange data with another Government agency, or non-Government organisation? What special requirements does this impose on your classification scheme in terms of cross-agency data compatibility and reporting? Does your agency document all its existing classification schemes? Does a formal classification scheme that matches your requirements already exist in your organisation? Is a suitable classification scheme matching your requirements available in some other Government agency? eg, the Australian Bureau of Statistics, or other Federal Government agency? Is there an industry standard classification scheme which meets your requirements (eg, defined by IEEE)? Is there an international (ISO) standard classification scheme which meets your requirements? 5.2 Post-design: Does your documentation for the classification scheme, contain the following items: A definition, or overall description and purpose, of the scheme? A nominated custodian or controlling body for the scheme? An outline of the sensitivity of the information classified by the scheme (in terms of privacy, security, etc.)? • Quality controls which exist in relation to classification and data entry procedures for the scheme? • The access controls which are relevant to data classified by the scheme? • The business rules which are associated with the scheme? Is each value of the classification scheme unique? Does your classification scheme require detailed usage instructions or guidelines to be written for it? Have they been completed? Have you documented any relationships with other classification schemes? (For example, are elements in this scheme extended by other sub-schemes, or is this scheme an extension of an element in another scheme?) • • • Information Management - Classification Guideline Page 10 of 18 Do the users have to be familiar with the underlying codes of the classification scheme, or can they work, (enter data, search and report), solely using the displayed descriptions? Is the classification scheme extensible? Does the classification scheme have a “null value” item? Has the classification scheme been pilot tested in a suitable environment? If the classification scheme is changed in any way, is there a single computer database to be updated, (accessed by all relevant software), or does every business system using the classifications scheme have to be updated? The Office of Information and Communications Technology is part of the Department of Commerce Address: Level 21 McKell Building 2-24 Rawson Place Sydney NSW 2000 T: (61 2) 9372 8877 F: (61 2) 9372 8299 E:mailto:oict@commerce.nsw.gov.au Page 11 of 18 Information Management - Classification Guideline APPENDIX 1 A FORMAL INFORMATION CLASSIFICATION An example of a formal information classification for motor vehicle shape might be: Allowable logical values, or description Articulated vehicle Bus Hatch-Back Sedan Truck Utility Van Distinguishing Code 1 2 3 4 5 6 7 In a formal classification, only one value may be selected from the allowable set for each particular use of the scheme. In database terms it is a table which has only two columns and all the rows are unique. The distinguishing code is stored in the data record, and a representation of the above table is held for display, selection and presentation purposes. The distinguishing code can be effectively used by computerised systems. Generally, the user of a computerised business system should not have to see the code, but should be able to work only with the logical values, or descriptions. The same formal classification scheme may be used in several different business systems. Many classification schemes subdivide into further classification schemes, or levels of specificity. Formal classifications can still be set up or designed in a manner which is inconsistent, or overlapping, or with any of the other problems outlined below for informal classification systems - but if set up in a computer application, their use will always be consistent with their design. A formal classification scheme as used in computerised applications is similar to, but not quite the same as, controlled subject headings used by librarians. A library subject catalogue does not require distinguishing codes to represent every subject. In a library the subject heading structures are set up once, and are slowly refined and extended over a period of years. In IM&T planning, classification schemes are frequently redesigned with each new business system, and completely new ones are frequently developed by individuals without reference to the activities of each other. Classification schemes cover a much wider range of computerised applications and information types than a library subject catalogue, and often in much more detail. The range and variety of classification schemes used in computerised applications presents Information Management - Classification Guideline – Appendix 1 Page 12 of 18 many design, administrative and usage problems. A person may be classified by sex, language, physique, eye-colour, education and many others. The “ICD 10” medical problems classification scheme used by health agencies is many thick volumes, and includes, for example, injury types such as those caused by prolonged exposure to zero gravity and space craft launchpad accidents. B INFORMAL INFORMATION CLASSIFICATION Examples of informal information classifications for motor vehicles might be: • Tidy, Untidy, Neat, etc.; • Green, Red, Bright Blue, Yellowish, Light-yellow, Yelow, Yellow, White, Offwhite, etc.; • Sedan, Hatch-Back, Car, Truck, Lorry, Semi, People-mover, Van, Ute, Flattop, Four-wheel, etc. If the values in each of these examples was the allowable set, and had codes allocated to them, then they would be formal classifications, albeit badly structured. Informal classifications are more likely to be used in a manner that is inconsistent, overlapping, fuzzy, and mis-spelt. An agency possessing thousands of records containing informal classifications will have difficulty in searching and analysing these records, and will have to expend manual effort to do it properly. Information Management - Classification Guideline – Appendix 1 Page 13 of 18 APPENDIX 2 – THE DIMENSIONS OF CLASSIFICATION A RANGE Range indicates the breadth of the classifications applied to information. Range is the number of different classification schemes that are assigned to a particular item of information. As an example, a storehouse containing information about real estate property might have the following three classification schemes pertinent to property: Structure Type, Usage Type and Finance Method. These may be represented as three separate fields in a record or database table row. The storehouse therefore has a “range” of three schemes. Other fields will hold non-classifying data, eg, the purchase details of the property. It is also possible that two simultaneous Usage Types may be applied to every property, eg, residential usage and business usage. If this were the case, the file would contain two Usage Type fields, (say, “Primary Usage”, and “Secondary Usage”), both of which use the same Usage Type classification scheme. As a second example, a Credit database field might have a range of two classification schemes applicable to it, being Confidentiality and Credit Rating. Both could have the values High, Medium and Low. In principle, it is not possible for an item of information to have a classification range of zero. In practice, however, judgements are made that some classification schemes are not useful, too expensive, or do not meet the needs of the agency. Range determination is a management decision. It represents the breadth within which a given piece of information is analysed. A large classification range will be more effective but will make the classification process longer. B SPECIFICITY Specificity is the level of detail of the classification scheme. It allows a user to be more precise when searching for information. It is a reflection of the number of hierarchies and values allowed within a classification scheme. Every classification scheme has some degree of specificity. The number of different values, which a particular classification scheme offers, indicates the degree of specificity. That is, for a particular classification, the number of different values from which a user may choose only one. In the example of vehicle type codes given in Section 1B above, there are seven values. The greater the specificity of information classification, the greater will be the precision and degree of relevance attainable in the results of information searching, and the more fine grained will be the statistical analysis which is possible. Several classification schemes may be related to form a hierarchy, for example, within a classification scheme for motor vehicles, each type of motor vehicle may also be Information Management - Classification Guideline – Appendix 2 Page 14 of 18 classified by its engine size and whether it is a front-wheel, rear-wheel or four-wheel drive. Thus, each record would contain a field holding a value or code for vehicle type, engine size, and drive type. This is better design than merely expanding the set of values for the ‘vehicle type’ classification scheme, that is, trying to accommodate type, engine size and drive type into a single classification scheme. Hierarchical classification schemes, which work to extend classification, amplify specificity. Even so, problems arise. For example, in the Dewey Decimal Classification system used in libraries the concept of Engineering was originally more simple. Then subdivisions such as electronic engineering and aviation needed to be developed to encompass new disciplines which developed with new technology. The Dewey Engineering classification area has now become crowded, with some very long numbers. (Librarians might not regard this example as having much to do with the subject oriented approach of classification schemes.) In a well designed or appropriately selected corporate classification model, their levels may broadly correspond to the levels of the Conceptual Information Inventory described in the Information Inventory Guideline. C CONSISTENCY Consistency is the uniformity with which information is classified, and may be measured by the degree to which the same document is classified the same way by different users. If three individuals were to classify the same vehicle using the table in Section 1A above, and each select three different choices, then the consistency of the classifying operation is low. If they all agree that it is a Utility, then the consistency is high. An example of low consistency as a consequence of classification scheme design occurs in the Dewey Decimal Classification system in the subject of Information Management where there is no consistent agreement on the Dewey numbers to be assigned. Information Management as a concept did not exist at the time the Dewey system was designed. It is better handled in library subject catalogues. (This is an example of where descriptive text is more useful than a code, as the text can be more responsive to changes at the conceptual level, whilst the code is fixed.) A good classification scheme makes it easier for users to achieve high levels of consistency. Information Management - Classification Guideline – Appendix 2 Page 15 of 18 APPENDIX 3 – DESIGNING CLASSIFICATION MODELS A DESIGN STEPS In considering a new classification scheme, the first step should be to see if a useable scheme already exists which can be applied to the information to be collected. Candidates may exist both internally and externally, such as in other agencies, defined by standards bodies or the Australian Bureau of Statistics (ABS). The ABS has defined many useful classifications schemes, pertaining to Industry type, Education and other fields. Use of standard classifications schemes enables cross agency information sharing and access. ISO/IEC 11179 Information Technology Specification and standardisation of Data Elements - should be adopted by agencies to manage relationships between classification schemes as a basis for Whole of Government agreements on common attributes. The next step is to perform a survey of the data to be classified, to ascertain its required range, the members of the range, and required specificity. Range and specificity, and the logical values, or descriptions to be assigned to each classification item should be developed in consensus with those business users in the field of interest who are familiar with the data and will have to use the classification scheme once completed. The design of distinguishing codes to be used in a classification scheme is a technical process, dependent on the computer technologies in use. It nevertheless requires input from practitioners to optimise the ease of use of the codes. A set of distinguishing codes for a classification scheme should have the following characteristics: • The code set should be able to be completely changed, requiring only a mechanical translation process, without affecting the use of the classification scheme itself. Changes and additions to the logical values are much more difficult, however, and may have wide impact. • Within a classification scheme, distinguishing codes should be mutually exclusive, otherwise the implication is that two different logical values mean the same thing. Similarly, no two descriptions should be identical, yet have different distinguishing codes. • The structure of the codes or symbols which represent the classification meanings or descriptions should be designed to be extensible, so that as the real world objects which are the subject matter of the classification scheme change or grow, new elements can be easily added. A particular example of this is in the next paragraph. In general, it is preferable that the presentation sequence of the elements in the classification scheme, if some sequence is necessary, be based on the values (meanings or descriptions) of the scheme, not the assigned code symbols. Otherwise, insertion of new classification elements may be difficult. In the example of a formal classification scheme in Appendix 1A, the descriptions are sorted alphabetically, and the table rows are numbered sequentially. If this correspondence is made use of, say by assuming that accessing the table sequentially by distinguishing code would always give an alphabetic sequence of descriptions, then there is a problem. If we now wish to Information Management - Classification Guideline – Appendix 3 Page 16 of 18 insert a new item, say, “camper van”, there is no gap between the numbers used as the code. Distinguishing codes should be entirely arbitrary. It is advisable that every classification scheme has a ‘null value’ item. This is the “no classification” or “undefined” classification. A classification item exists with a value or description which says something like, “undecided” or “no selection”, with a corresponding distinguishing code. This feature is necessary where the agency’s business rules allow that, under certain circumstances, no classification needs to be made. This applies even to apparently simple schemes. The classification of “sex”, for example, as defined by ABS, goes further than “male” and “female”. It has the possible values, “Male”, “Female”, “Not known” and “Not stated”. Often, null values will be used because a particular choice in one classification scheme determines which of several other possible classification schemes are then applied. Complex classification schemes and their codes should be tested in a pilot scheme before full operational use. B CLASSIFICATION SCHEMES FROM DATA MODELS Information Management professionals are coming to recognise that the management of data classifications is as important to an agency as management of its databases and data models. Increasingly, the focus of information management projects is moving from the development of data models to clarification and utilisation of the classification schemes inherent in the data model. This Guideline recognises that there is a difference between classification scheme values and the codes that are assigned to them, and that getting the scheme right is more important than the codes allocated to the values of the scheme. It also recognises that for classification schemes to be both comprehensive and specific, it is necessary to allow a number of schemes to be applied to an item of information and for schemes to be hierarchically embedded. Agencies should use the Enterprise Information Model and Business System Model to identify classification schemes pertinent to the agency. The creation of classification hierarchies from entities in a data model is straightforward. Firstly, the different ways in which stakeholders would wish to classify the entity are addressed. This enables all stakeholder views and uses of the classification to be made explicit and thus the possible range of classification schemes to be identified. Secondly, the subtypes of the entity based on these classification schemes are developed, giving the values for the schemes. The classification hierarchy is then extended to further subclassify the subtypes in the structure, giving greater specificity. Once in place, codes equating to these values can be designated (or the codes may be the values themselves). C CLASSIFICATION SCHEMES AND METADATA Classification schemes may represent information at a highly detailed level, for example, a set of ‘Customer Type’ codes. Classification schemes may also represent Information Management - Classification Guideline – Appendix 3 Page 17 of 18 information about other classification schemes, for example a scheme which classifies information, including classifications, by Security, Quality, Reliability and Value. Standard classification schemes exist for these, such as the Admiralty codes used in military and police intelligence operations. The latter are examples of data which describes other data, commonly known as “metadata” (or meta information). Metadata in the context of classification schemes tends to be administrative in nature, which is to say it is mostly concerned with the administration, rather than the intellectual content, of the information being classified. Other examples of such classification schemes are Purpose, Form, Source, Usage, Value, Confidentiality, Custodianship, Ownership. Other examples of metadata are library catalogues, data dictionaries and database schemas or data models. Information Management - Classification Guideline – Appendix 3 Page 18 of 18