RECORDS MANAGEMENT INSTITUTE
MARCH 07, 2001
PAUL SABOURIN

FUNCTION-BASED (RECORDS) CLASSIFICATION SYSTEM

Good morning. Today I want to discuss the design and development of a function-based file classification system.

I'll go quickly through the early slides as these are usually presented by Dr. Richard Brown, who wanted to give some historical perspective. Basically, the classification systems design that you are currently using is about a hundred and seventy years old. Prior to Confederation, under the British Colonial Office, people would send messages and dispatches and so forth in dockets. The dockets were essentially documents, folded into three, almost cylindrical, and put upright in boxes. They had a docket number attached to them and you had to know what that number was to access a particular dispatch or document. The Great Docket Unfolding happened at a certain stage when there were so many boxes and dockets standing upright without any organization that they decided to literally unfold the documents and lay them out flat. But outside of the box, you couldn't organize unfolded dockets so then they invented file jackets, which we're still using today, standing upright on file shelves.

The last significant evolutionary stage was roughly at the turn of the century, and that came out of problems with record keeping. The Post Office, Department of Railways, Canals, Marine and Fisheries, etc. including Indian and Northern Affairs, designed a subject based classification system, based on themes, subjects and general events. In 1968, the National Archives of Canada had a series of publications on Records Subject Classification; Records Organization and Operations, and so on. They addressed subject classification conventions, but that was the last great advance in record classification in the Government of Canada.

The record keeping crisis at the beginning of the twentieth century is similar to the current record keeping crisis today. If you look at substituting the computer for the typewriter at the turn of the last century, digitization for lithography, comparatively complex government functions today compared to what they were a hundred years ago during the building of confederation, we're in the same type of records management crisis but now we're in an electronic work environment.

The Victorian solution was essentially to de-contextualize the information or the items. Essentially, the British Empire extended through all the continents and what they would do in the, for example, in the sciences like biology and anthropology and so on, is they would go into a particular area or continent, get a leaf, plant, or animal, etc., dry it out, bring it back to a museum in Great Britain, put it under glass and label it under a particular class or category of entity. What they essentially did was they de-contextualized the item, removing it from its natural habitat, its origins, its location of creation, and created artificial classification systems and categorizations for all elements.

There is a tale of two contexts here which means essentially that we are stuck with the same problem the Victorians faced in that with subject classification we removed the documents that are created as by-products of business activity and that document evidence of business transactions. We then create artificial subjects and hierarchies which try to capture the essence of groupings (not what the information is about) by subject, and then re-arrange everything in alphabetical order thus completely destroying the natural relationship between activities, events and so forth. For the archivist this would be known as destroying principles of "provenance" and "respect des fonds".

The repair that we see is required, needs to reconnect the business enterprise with information management: to fully integrate and load IM into the operational programs and service delivery in the most completely transparent and seamless way possible. Knowledge Management is the key to building this bridge and re-directing information traffic into program areas from the current status as "hors context".

The solution is to go functional. A lot of people are going functional, and a lot of archives and records management businesses throughout Europe and Australia are going this route. So it's not something new. As a matter of fact, I see someone here from Statistics Canada. They developed a functions-based classification system close to sixteen years ago for all of their survey activities. So it's nothing really new except that we're only recently starting to talk about it worldwide.

Now I'm getting into the functions based classification systems, which is why you came here. I want to start out with some basic notions and what I'm building or what we want to build, first of all, is a theoretical structure upon which to develop a methodology and then build systems, business rules, etc. for such a classification system.

What we are projecting to develop in the National Archives is what we call a Business Activity Structure Classification System. What we meant by "classify" was simply, according to the dictionary, "to arrange in classes, assign to a class, or rank or order,"of something. And what we meant by a "class", where we find the origins of our notion to classify, was (again going back to the old Victorian sciences) at the highest division you started with a class, then order, family, genus, species and so on, for any entity broken down in a hierarchical arrangement.

The subject based classification system is, essentially, a system that arranges records into hierarchies of classes and sub-classes of subjects. Yet the problem with a lot of the records management literature is that it never really defines what a "subject" is. The Oxford Dictionary describes or defines a subject as "any theme, of discussion, of a description or representation, a matter to be treated or dealt with, a theme, a leading phrase, motif". Even the National Archives Subject Classification Guide was not very helpful in that it tried to explain subject files by redefining it in terms of subject.

For example, it defined subject files as a "collection of papers on a specific subject or sub-subjects within a single file."

As for a subject classification system, it is stated that "the logical arrangement of individual subject files within a filing system constitutes a subject file classification system." In other words, subjects are anything and everything that you can think of. It wasn't very clear. If it's about everything possible then it's not about anything in particular. Current standard texts are not much better in coming up with and analysing the theoretical and methodological foundations upon which to build an information classification logic.

In the book titled 'Information and Records Management: Document Based Information Systems', by Robek, Brown and Stevens, what they describe is that classification is the art of putting like-things or records of a similar characteristic together. For most systems, the entire scope of record subject matter is divided into seven to ten major divisions (in Canada these would be called Blocks). I won't read the rest of the slide except for the last bullet where they say 'Describing the records series by the function they perform is called functional filing.' That's not what the National Archives considers functional filing. What they mean there is the series of records are created pursuant to a function carried out by the institution. I think that's where part of the confusion comes from in determining what is a functional filing classification system.

The notion of functional filing in Robek, Brown and Stevens is actually subject based filing or arranging of file thematic titles in a hierarchical arrangement, alphabetical order, of subject titles. It has nothing to do with analysing the function, sub-functions, activities, transactions. The example that is given in the book for a functional file classification system shows they have entities and end products, listed conveniently in alphabetical order under Human Resources, starting with E for employees, F [for Fringe Benefits], J [Job Description], T [Training], etc. On the other part of the slide, we find under Property subjects in alphabetical order again [Buildings, Equipment, Land, Patents]. Patents is there though nobody is really sure what Patents is doing under Property unless it's intellectual Property, but that is not what most people would think of under movable or immovable assets. Except for Training, all the file titles are subject titles, denoting objects or end products listed in alphabetical order.

The perspective we are going to take in "function-based classification systems" (that's just the generic word we're using) is that we have to go deeper into analysing the functional components, sub-functions, actions, transactions, and business processes in order to develop the theoretical and methodological foundations for arranging, i.e. classifying, information according to a structured business process. Because it's the business process which actually creates the records and uses the records while doing its business. In other words, what we need to do is create a new records management orthodoxy for developing file classification systems that are not subject or object based but rather based on a design structured according to the sequence of activities of a business process which creates and uses the records to be classified in the first place.

We conduct the business process analysis in order to discover the underpinning business activity structure and the methodology for this type of system analysis for record keeping. As far as we know, this methodology for business process modelling to an information classification system does not exist in published form. So if you wanted me to send you to a web site or publication that would give more information on this, I believe there is done, especially providing you with a Canadian context. As far as we know, there's no published material for function-based classification systems that reflects the functional analysis in the MIDA's for the common administrative functions of the Government of Canada (GoC) issued by the National Archives.

On a number of issues what we are proposing is different from what the Australians are doing as described on the Australian DIRKS site. The DIRKS modelling of their function based systems has 3 tiers; first at the highest level you have a function, next is an activity, then a transaction. However, in their model, there are no apparent linkages, no natural relationship between the activities that are mapped out. It looks as though the activities are all stove-piped. So you don't know why, for example in their DIRKS manual, Activity 3 is the third activity. Why isn't it first, or why isn't it fifth? They don't look at a function in terms of a sequential activity structure. DIRKS only lists supporting activities in alphabetical order. They talk about relationships but they don't explain what they are.

The key strategy to the methodology to be used in these types of structural-functional sequencing of activities would structure the classification system according to the blocks, primaries, secondary file titles. We didn't want to change the language now. That might change next week, I don't know. It would normally be at the secondary but especially at the tertiary transactional level that subjects, objects, and cases would be identified. Now, on the two previous occasions that I've given this, people have come up and said "Well, we still need subject files!" I'm not saying you don't need subject files. What I'm saying is that there is a superstructure and the records in a subject file must relate to the function performed by an individual or institution. A superstructure is made up of the blocks and primaries because records don't exist at the block and primary level, they only exist at a secondary or tertiary level. So the superstructure, which is the block and primary, is the function and sub-function structure respectively of your business process. It's at the secondary, but especially at the tertiary level that you will actually see the files and information classified.

But the problem with the subject file classification system was that there was no linkage between the subject file or the information documenting the business line or service line performed. And it certainly was not obvious to the end user how to link the manner she/he performed the work to a subject title. There wasn't a direct linkage and so when we had to do an archival appraisal, it then became very difficult to make the link as to what activity do these sets or series of records support, or what is being evidenced in the records that you are creating and classifying.

Another twist is that if you are systems designers who, during your preliminary research, ask in your user areas or program areas, "What is it that you do?" then that is of secondary importance now. You ask the user/program areas what it is that they do but more importantly, how do they conduct their business? If you are running a project or program, how do you go from the start/initiation/conception phase of operations to the delivery of a final product or delivery of goods and services to Canadians? What is the business process? These are the questions we must ask today.

Definitions are helpful so lets look at another. This is the working definition the National Archives is using for a function. A function is, 1 - any purpose, responsibility, task or activity which is assigned to the accountability agenda of an institution by legislation, policy or mandate; 2 - typically common administrative or operational functions of policy development and program and/or delivery of goods or services; 3 - a set or series of activities (broadly speaking a business process) which when carried out according to a prescribed sequence will result in an institution or individual producing the expected results in terms of goods or services it is mandated or delegated to provide.

What we're most interested in for the design of a function-based classification system is the set or series of activities which, when carried out according to a prescribed sequence (either legislation, policy or mandate), will enable the institution or individual to produce the type of goods or services it is mandated or delegated to deliver.

Now, given that understanding of 'what is a function', you can probably see how the GRDS will be completely replaced by going functional with the MIDAs. There are no subject lists in MIDAs, much to the chagrin of some people, and we're not going back to subject hierarchies. We will use a high level functional approach, including new retention guidelines for government records that are of a common administrative nature.

In this system's design methodology, first we look at functions. What we're saying is that within the total mandate of an institution there is only one functional universe. The functional universe is composed of all activities performed by an accountable institution in the Government of Canada. No activity performed by an accountable institution can exist outside of its functional universe. That idea will help position people when they continue to ask us what we do with the United Way records for disposal. If it is not part of the accountability agenda of the your institution, then it's not part of your universe. The functional universe is divided into two parts: one, representing operational activities and, two, non-operational activities, the later of which is mostly common administrative activities. The argument that all records are either common administrative or operational in nature is logically invalid. You may have records that are neither common administrative or operational. But it necessarily is the case that all records must be either operational (and we should all know the definition of that by now from the MIDAs), or non-operational in nature. There can be no exceptions.

 

The records universe is composed of all records, under the control of an institution, relating to the activities for which it is accountable. No record under the control of an accountable institution exists outside of it's records universe. The records universe is divided into only two parts: one represents the operational records, and the second non-operational records; the latter of which is composed mostly of common administrative records. Now at some point in time, the records universe is going to collide with the functional universe. How it does that is rather interesting. What we needed was a new definition of a record. I think the issue of the definition of a record was part of the agenda for today's meeting. I'm looking at the definition that was composed by the International Council of Archives in the publication called 'Guide for Managing Electronic Records From an Archival Perspective.' Catherine Zongora told me earlier today that, in fact, the ISO Standard is looking more and more towards this definition as the working definition of a record. That definition is also in John MacDonald's report of last year on a Situational Analysis of IM in the GoC. So this new definition of a record does not contradict or run in conflict with the definition of a record in the National Archives Act. Definition - "A record is recorded information produced or received in the initiation, conduct, or completion of an institutional or individual activity and that comprises content, context, and structure sufficient to provide evidence of the activity regardless of the form or medium."

What this definition does is point out the characteristics which are shared by things like a map, a book, a photograph, etc., that would allow you to say "Ah, they share common characteristics of recordness, therefore they ought all to be grouped under this universal idea called a Record." What is important to note here is that records are essentially information about activities. They are not about subjects, not about hierarchies or motifs, or leading phrases, they are about activities.

The second major point that is made here by this definition is that it describes the sequential nature of activities and information about those activities in the sense that the information is produced or received in the initiation, conduct or completion of an activity, which means that it has a beginning, a middle and an end. It is sequential in nature because all activities have a beginning, a middle and an end. You don't find that sense of sequence with subjects. And you don't find that theory as part of the Australian literature on functional analysis and functional classification.

Looking at this working definition which describes the essence of a record, we can describe how the two universes can collide and merge. The records universe and its parts is produced, derived and directly linked to the functional universe and the activities which create and use the records. Since the records evidence and document activities of an accountable institution, the nature of the functional universe i.e. either operational or non-operational activities, directly determines the nature of the records it creates and uses in the records universe, i.e. making them either operational or non-operational records.

What we are creating is a Business Activity Structured Classification System which "is a logical structure for the classification of records which treats the information contained in records as by-products and evidence of institutional functions, and whose logical sequential structure is predetermined by the natural spatial, temporal, and causal relationships which exist between the activities, or set or series of activities, which compose the function". We conclude, that if information is about activities, and we know that every activity has a beginning, a middle and an end then there must be a real sequence of events and real sequence to the records which evidence the activities. The sequence exists because we all conduct activities at a specific moment in time, or over a particular duration of time, and usually in particular locations. And normally you will see there is a cause/effect relationship between events or activities we carry out, such as, when you turn a key in your car's ignition it then caused the engine to turn on, and when you slipped the gear into reverse you backed out of the driveway starting perhaps from the top of your driveway to the street curb before proceeding onto the road way to work after slipping the gears into forward. There is a sequence and structure to all activity, even unplanned activity.

Now we are exposing and creating the methodology. I've given some rudimentary axioms on how we can position the functional universe of an institution within its operational mandate or common administrative delegated mandates, and we've looked at the records universe. What we're looking for is the business activity structure of the institution's functions which is discovered through an analysis of the linear or life-cycle business processes supporting particular functions and of the subsequent sequence of business activities. In a particular project you may have an anticipated end product that will not be reoccurring cyclically, it will be a one time event. Some projects or activities like materiel management or real property management adopt a life-cycle approach to it's business functions. All these activities may be linear or cyclical in nature but they are all composed sequential as well, while allowing for concurrent activities to take place towards the final end product. The business activity structure to this sequence is determined by and constructed according to the natural spacial, temporal, and causal relationships between activities composing the function. What we're doing is mapping out, either in the MIDA's or in other program analyses, the structure of your activities and transactions, mirroring that structure which is innate and obvious, into a structure for the classification of all information created by those activities and supporting or evidencing those activities.

The inherent logic to the structure of a business activity classification system is not hierarchical groupings but rather is sequential in nature. The sequential components are mapped out using various business process analysis techniques such as program review and evaluation, workflow charting, process modelling, or critical path mapping. That's how you get to it. We're saying here, the days of using subject themes and the alphabet to structure how your records classification systems are designed, i.e. especially in the electronic work environment, are over at the superstructure block and primary level.

The classification system on the right hand side gets mixed reactions. Some people look at that and have said "My gosh, that is so different from what we are used to seeing. How did we move and shift so far?" And I had someone recently send me an e-mail writing: "Excellent presentation and I'm so glad that we haven't really changed much from the classification systems based on subject." So maybe your reaction depends on how far away from the slides you're sitting, I don't know. I'm thinking this is pretty radical.

Subject-based vs. Function Based

 

Equipment and Supplies

Materiel Management Function

700-849

700-849

700 

Equipment and supplies - General

700

Materiel management - General

715

Building materials

710

Assessing materiel requirements

720

Catalogues, manuals, price lists

720

Planning materiel requirements

725

Clothing

730

Acquiring materiel assets & related services

750

Foods

790

Operating materiel assets

754

Forms

800

Using materiel assets

758

Fuels

820

Maintaining materiel assets

764

Furniture and furnishings

830

Replacing materiel assets

795

Office appliances

840

Disposing of materiel assets

830

Stationary

 

 

840

Vehicles

 

 

What does this really look and feel like? You may recall when we rolled out the Materiel Management MIDA last year that I had delivered to all participants or the orientation sessions, a classification prototype shown here on the right hand side. Basically it's the breakdown of the function called Materiel Management in the government of Canada as it is laid out, described, and prescribed by Treasury Board policy. There are eight and only eight sub-functions. The policy describes it as a cyclical, sequential, business activity or function. When you look at the left hand side, i.e., the old subject classification hierarchy, what you see is Equipment and Supplies as the block subject followed by itemized entities in alphabetical order: building materiel, catalogue, clothing, food, etc., down to "V" for vehicles. That worked well enough in the past because, after all, if you wanted to look immediately for records about vehicles, and you had a couple of pages to look through in a file manual, you wanted to quickly and visually move down to the bottom of the list because you knew the "Vs" are at the end of the alphabet. That efficiency trick is irrelevant in the electronic work environment where you just put in a keyword to find any related files to your search. However, if you are creating and want to retrieve information about the types of policies under Crown assets disposal, for example, you would know that it's at the very end of the life cycle of managing materiel assets using a functional approach, and where all associated evidential records (that perhaps contain none of the keywords you choose) for that sub-function can be located and accessed.

 

What are we looking at in terms of a numbering structure? This is maybe what some of you are interested in. I can't change the sequence of activities because they are described or prescribed by policy, and it makes sense (for example under materiel management) that you can't use materiel assets prior to acquiring them, and it's not logical that you would start acquiring materiel assets before you have an inventory and assess what you have in relation to what your program needs are. That's the logic of the sequence in terms of what it is that you need, when over time, for how long, where will they be located, whether you're putting them in warehouses, whether you're shipping them out to regional operations, and whether there is a cause and effect relationship so that you cannot acquire or procure items until certain approvals and signatures are obtained and so forth.

Business Activity Structure Classification System
 

Materiel Management Function X.3

X.3.0

Materiel management - General

X.3.1

Assessing materiel requirements

X.3.2

Planning materiel requirements

X.3.3

Acquiring materiel assets & related services

X.3.4

Operating materiel assets

X.3.5

Using materiel assets

X.3.6

Maintaining materiel assets

X.3.7

Replacing materiel assets

X.3.8

Disposing of materiel assets

The numbering convention proposed here is only a suggestion. Essentially the "X", which is the first digit, should denote the identifiable business line within an institution's mandate relating to administrative services, corporate services, or management services, whatever you call that broad function which is common administrative in nature. Some institutions have four business lines here while others may have six or less. Usually one business line within an institution pertains to everything dealing with corporate services, i.e. the common administrative services. That "X" in the file number could be, if the business line is number 1 for corporate services, simply 1. Everyone in the institution may know or can be easily trained to recognize that 1 in the first position of a file number represents the first business line of their (or any) institution. For those of you who haven't done the research, the business lines are available and published in the Main Estimates and are found in the Treasury Board website. All institutional business lines and service lines are there, are published and they are becoming more and more stable. What I mean by stable is an institution bases the business lines on its legislation and it's pretty safe that the legislation will remain that way for at least five to ten years. The chances of changing classification systems every time you decide to change the name of your organizational structure is not going to happen if your functional system is based on business lines derived from legislation.

The second positioned digit in the file number (X.3.0) denotes the number of the Records Disposition Authority that covers a particular common administrative function profiled in the MIDA. How convenient for the Archives and all federal institutions to link file numbers to functions reflected by the number of an Authority covering the same function! The Authority is 99/003 that covers Materiel Management and of course prior to that it was Schedule 3. So for the records practitioner, records user, or classification system designers building the systems they know that the second digit would be the same as the Authority and function number for common administrative functions. For example, "X.4" would be for the Comptrollership function [Records Disposition Authority (RDA) 99/004] and "X.5" would be for the Human Resources Management function [RDA 98/005]. We believe this numbering convention makes sense to everyone and helps to base classification structures on something that everyone is used to, that is almost transparent to classification systems analysts, and designers. This is something you can use, especially if you are still having trouble trying to decipher what Authority applies to what records. This is one way that can make the link between the Authority for the destruction of records and your block functional classification system.

The "0" at the third digit position (X.3.0) would indicate the general series for the function because we know there is always a requirement for a general series of information that crosses a multiplicity of activities or transactions within a function. The "1" to "8" digits simply denote the number of the particular sub-function within the sequence of sub-functions that was previously identified, mapped out (according to the particular Treasury Board policy in this case or legislation) and published in the 99/003 MIDA. The underpinning idea here is that people want a number that can relate to a function or how a business activity is structured, something that is common knowledge and almost intuitive for all employees within that particular institutional accountability domain. So I think we must build that idea into functional classification systems design methodology, something that is meaningful and useful to the users as well as records and information management professionals.

Using this particular structures approach, we are hoping to create some sort of classification system prototype and model, but we have to get the theories correct first. For example, there's a lot of people running in different directions today describing the records universe, looking at categories of records and other things. Those people here who took the old Records Management course might know how many categories of records are there? How many? There's no wrong answer - any answer will do. Forty-two categories someone has just said. That could be true because we've never defined what a category of record was. Normally we said, in the old publications, that there were only two categories of records, i.e., common administrative, which we called housekeeping then, and operational records. Then the Archives came out with other publications in the 1990s that said there was three categories including ministerial records. Some people also claimed there were really more categories, including a fourth category which is for transitory records. But when you read the Transitory Records Authority, it never claimed this was a category of record.

This is why I say there's much groundwork to be done before we start confusing ourselves and we must become more rigorous in our definitions, concepts, and discourse. We must start asking ourselves what records belong under what types of systems, belonging and supporting what kind of common administrative activities or business activities, and which are operational in nature and which ones are non-operational? We haven't done that on a consistent basis yet. But when we do, I think we can roll this function-based classification system out into the operational program areas, the business areas, and construct it according to the business lines, functions, and in sequential structures. Every manager who has taken a project management course will understand our approach. This will not be foreign to them. In fact, I think it will make more sense then trying to describe why something like "Patents" is under "Property", or how their entire business process model is rearranged according to the alphabet.

Business Activity Structure Classification System - Materiel Management Function
 

Materiel Management - General

700- 0

Policy

       1

General

       2

Departmental Materiel Management Committee

       3

Complaints of defective materiel

 

Business Activity Structure Classification System - Materiel Management Function
 

Acquiring materiel assets and related services - General

730- 0

Policy

       1

General

       2

Purchasing procedures

       3

Catalogues, manuals, price lists

       4

Fuels

       5

Acquisition cards

       6

Office appliances

       7

Stationary

This classification structure is part of what I handed out last year when we disseminated the MIDA for Materiel Management, and what you see in the general series is again, information pertaining to a multiplicity of actions or transactions under the function Materiel Management. So that's where you would classify such files as your committees related to Materiel Management because your committee would likely, at the departmental level, look at all aspects of the entire life cycle of managing materiel assets.

At the secondary level for example, you can see files on office appliances or stationary. These could be subject driven without listing the types of stationary and so on, but at the "-2" level, purchasing procedures, that might be guidelines and procedures that describes exactly what you must do in order to obtain the type of materiel assets over time required for your business processes and programs. So at the secondary level that's where you can have the mix of titles for activities, subjects, case files, projects, etc. And for the individuals who have told me "my users like project files. You're not going to tell me that if I have a project file on project "X" that I'm going to have to take out all the records apart and place them in sequence?". No, that's not what we're saying. Under the "740-2-98/123456" file, you can have a project, a case file, a contract file, etc., and the contract file could be several volumes thick.

Business Activity Structure Classification System - Materiel Management function

Acquiring material assets and related services - Contracts

740-


1
2
2-98/123456
3


General
Contracts
Specific contract case files
Lease agreements

We're not asking you to completely throw out your good records management practices or ignore user requirements, but what you have to consider in the classification system design at the "-2" secondary level is the adoption of identification nomenclatures to link the case files or project files that support the sub-activities of acquisition, purchasing, or procurement, of materiel assets to the real sub-functions the records evidence. That's the type of link and type of system design we must look to build.

The contents of the following slide has been provided to me by my friends and former colleagues at Statistics Canada (STC) who designed this function-based classification system close to sixteen years ago. I was involved in that project and we had headaches trying to design it. We held the view then that we were records management professionals and we knew best what to do. The users kept saying "You're just making this up. You're just throwing numbers at us. Give us something that makes sense to us the end user." So what we agreed to do collectively, with the highest levels of management support, was to construct a records classification system based on a departmental accepted survey methodology of only six phases (in sequence) from the survey initiation/conception to the final marketing, publication, and disseminating phase. What they've done is, where you see four "X's" for the primary number, (they call it a SDDS number) this is assigned and managed across the department by a program area for each survey. Every one of the over 400 surveys is assigned a specific four digit number by the institution as a whole, and any of the thousands of employees at STC can log into a departmental site and locate any survey by name and be provided with the SDDS number or visa versa. Use of these SDDS numbers, rather than blocks of numbers as per the old Subject Block Numeric methodology, makes sense, is immensely user friendly (the true criterion of an effective classification system), and is meaningful throughout the institution.

STC Records Classification System

Based on survey activity methodology

SDDS

Survey

XXXX-0
XXXX-1
XXXX-2
XXXX-3
XXXX-4
XXXX-5
XXXX-6
XXXX-7
XXXX-8
XXXX-9

Policy
General
Development
Collection
Processing
Research and analysis
Marketing and dissemination
Budget
Contracts and agreements
Methodology

At the secondary level in the STC example, you start out with Policy, represented by "-0", because everything derives from policy or legislation. You are not likely to start activities if it's not in your mandate or accountability agenda to do such functions. The General is represented by "-1" denoting the first phase (normally called survey initiation/conception phase) of carrying out a survey. The second phase is called survey development. Third is collection. You can't collect the data if you haven't created the survey. And you can't process data , which is the next sub-function in sequence, before you've collected the information. And this goes right down to "-6", Marketing and Dissemination. This is the structure of their business activity but we didn't call it that then. We called all of this subjects because we didn't have the terminology developed then. It was based on a methodology that was acceptable throughout the whole department, and, when you look at it, it makes sense today to recognize that every survey project proceeds through the initiation/conduct/completion process and this view is today supported by the new working definition of what is a record.

For the last three file titles we see there is "Budget" which pertains to financial arrangements and transactions in each one of the phases, "Contracts and Agreements", and the last one, "Methodology". The users wanted specific files for these activities and subjects. Close to four hundred and fifty surveys replicate these secondaries and except for the largest survey (i.e., the Census) there is no deviation from the schema. This function-based classification system has withstood the test of time and has changed relatively little over the last sixteen years.

I wanted to leave this presentation with a provocative statement. When you think that all the information in records you ever read, classified, put into files, and have managed or disposed of in the past have all used (what some call) the technology of narrative. When you think that all your email, reports, letters, correspondence, spreadsheets, etc., describes a chronology of events that have or will tale place over time, space, and be either the cause or effect of some thing - Why is it that we destroy the very chronology described in narrative which forms the basis for understanding the content and context of every document by then grouping these records into subject hierarchies and re-arranging the whole lot into an artificial alphabetical order? Why do we do it?

Just to give you some idea what other people are thinking and doing, this next slide is from a publication 'Archivaria'. 'Archivaria' is a scholarly publication, produced here at the National Archives. There you will find some serious leading edge work done internationally by the authors of the articles. This article one comes from an author in Australia. It's called the "Australian Record Keeping Metadata Schema". What they are doing is they are looking at information within a functional context depicted visually as a triangle of relationships. At the top we have your business lines, your business activities, or your functions. Moving down to the right corner, you have your business activities as documented in the records which are used and created by people and agents. What they mean by "agents" here are the employees of the institution, employees of other institutions, or other private or public institutions that share and access their information during the course of doing business. And the third corner depicts what the agents do and how their activities are documented through the records.

This schema is driven by business activity, not subjects, objects, or themes. What I understand may be taking place in the GoC is that the new revised version for MGIH talks specifically about and uses the words 'business activities', and the responsibility of every employee will be to document their business activities. So when we say that we are developing a Business Activity Structure Classification System, I think we're right on the mark.

Here is a little more elaboration on what the Australians are proposing. What you see depicted here is that the records that document the business activities are managed within a business record-keeping arrangement, a system, an infrastructure of policy, procedures, and people who take accountability for record-keeping. It views record keeping as part of the integrated business activity. It's not an adjunct, it's not about the (so called) people in the basement. It's about the business activity and supporting the normal course of business of an institution or an enterprise. That idea is the additional component in the slide where the function-based classification comes into play. It depicts within the triangle what people or institutions do, how they document what people do with records, and what the business activities are.

The next slide is a little more elaboration on the same schema. Ultimately, records-keeping and records classification systems in this schema form an integral part of any business activity starting with the legislated mandate (in the top down view) governing what and how you conduct business. Again, with our MIDAs we have always said how you do business is proscribed according to legislation, policy, regulations, statutes. There is more elaboration in this slide in terms of authentication of the records which (in that triangle) is part of the record-keeping system, part of the collective memory of the institution which documents how you conduct your business and, in terms of accountability, how you delivered what you were mandated to deliver with the quality and quantity, etc. necessary.

I'm now going to show you the organization chart for CCRA. What I want to point out here is that they have listed in the organization chart all the business lines for which they are accountable. Going back to how we define the records universe it's their accountability agenda and they have six business lines. CCRA has mapped out which areas or organizational structures are responsible for which business lines within the department. If you're designing a classification system based on business lines that support a business function, then you should be looking at the organizational structures that are responsible for the different business lines and this is a way to do that. This is the type of analysis and mapping that is really interesting, that grounds how you are going to propose a classification system to users who may be a little sceptical about adopting this functional route, who love their precious project files and want to keep them in their "C" drives or worse places. You can see that the sixth business line is common administration and in all the regions they do business line six for their own common administrative functions. This is a good tool in doing a structural-functional analysis.

You may notice some differences between what we're going to try to develop here at the National Archives of Canada and what the Australians are doing. One example is that when we do functional analysis, we deconstruct function at two levels. One level is the organizational structures of the government, the organizational boxes and portfolios, the Offices of Primary Interest and non-OPI's. We analyse functions in terms of the organizational structure and then we look at another level, i.e., the functions themselves. What is the structure of activities, the sequencing? You won't see anything in the Australian literature about Office of Primary Interest or Office of Collateral Interest. That's peculiar to Canada.

The other difference which I have not fully examined is that the Australians do not accept a life-cycle approach to records management as most people do. They look at records in terms of a records continuum theory. It's not a life cycle approach. Will that make a difference in the structure of their classification systems and ours, I don't know. But that's okay because there was never one way, and only one way, of designing even subject classification systems in the past. You had the duplex numeric, subject block and a few other variations. But everyone seems to be moving in the same direction towards a functional approach because of the vast amount and volume of information being produced, the complexity of the information, and how the Government of Canada and others now operate within an increased climate of comptrollership and accountability. And the only way that you can be accountable for what you do and demonstrate responsible governance is through the records that you produce within your business activity and which document what was done.

Basically, that's the end of the presentation today.

*********************************************************************************

 [Questions and Answers]

Who's going to pay for this?

You are going to have to build a business case when you redesign your classification systems, perhaps as you did after program review. And the business case should be based on the premise that this will be easier for everyone. The advantage that we found in developing function based classification systems in other departments is that you don't have to spend a lot of time training the users, because they know what business they are in and how they carry it out. They create the records that support their particular activity. So there was one hundred percent buy-in. I think you can sell it in terms that it makes sense to the users. It would require a lot less training of the users, and records management staff. It would require initially understanding of how business systems operate and business activity sequencing occurs. But once you start down that path, it's based on things like your legislation or the business lines that won't alter very much" There would be very little upkeep and maintenance. You wouldn't have to reconvert file titles at the change of a name of a branch or division. You would undertake changes probably just for additions of new programs but the old legislated mandatory services that you have to deliver would remain the same. Not one hundred percent perhaps but ninety-five percent of the time the system would remain the same.

In one of the presentations we had, which was internal to the archives, an archivist was saying that they always had a subject classification system which sometimes mapped out the functions and sub-functions and so on. The response is, if you did, that's great, don't change it. But by and large all the examples that we have ever produced including the last one, very recently for Y2K records, is by subject; it's in alphabetical order. There's no link to the business function or common administrative function. You must mirror back the innate structure of the activity into a file classification system. If you are a large institution like Public Works who has to manage all the records of Real Property across the country, you are going to need another design for sure instead of subject filing. Putting everything under "B" for Buildings is not going to work for you.

What about classifying things like the United Way campaign?

For subjects like the United Way Campaign, you have to change your perspective to the functional system. One of the comments of an archivist was: we used to have subject filing and subject numeric systems, we lost the context of the function. What did it actually support in terms of a common administrative or operational function? Now if we go functional we may lose the context of the subject. I don't think the latter will happen. But it's true that if you look at things like the government charitable donation subject, the United Way Campaign, it's a project file and people wanted everything in the same file. The problem is it doesn't document or evidence one function. You have documents that, I think eighty percent of it, has to do with pay deductions, removing from your salary a certain amount of contributions for the United Way, which is a Human Resources sub-function under Pay and Salaries. It documents a Human Resource activity. Or your reserving a room or a park and having a Bar-B-Que for the campaign and you gain money that way, and that's time off work for employees. Or rental of office space which is Real Property or again Human Resources for leave from work. Or you are acquiring flip charts, pens and pamphlets to distribute to advertise that the United Way Campaign is taking place and so on, that's purchasing and Materiel Management. It's what is left over, since I define the universe very tightly as either operation or non-operational, after the Human Resources, Materiel Management and all other MIDA's are used, that the government is probably not accountable for. We're never going to give you a Disposition Authority for what the government is not accountable for. It's outside of our jurisdiction. But under the subject classification you had an "Everything Goes" attitude as long as the subject was about any campaign. Well, not any more. That's going to be tricky for some, but on the other hand, there's very few examples like that. Once we get our mind around the functional approach and sequencing of business activities, it should be a lot clearer to everybody. After all, we've been working at subject filing for a hundred and seventy years. We're not going to change everything by next year.

Your definition of operational and non-operational. There's one category that I think you are missing and that is, if, as an institution, you receive records from a non-institution, such as a non-government organization and non-solicited information, you still have to manage that through it's life-cycle. You've still got responsibility for that information.

Well you'd have to define that. Why are you getting it if it's none of your business?

You have the legal responsibility for managing that information.

Why would you be getting records you have no business receiving. You must have some reason why you're acquiring them. It's either common administrative or operational for your institution.

But it could be attached to a business line for which you are responsible.

But then it's because you have a mandate to acquire them. Otherwise the Privacy Commission will say "What the heck are you doing with this person's information? You're not supposed to have that." So it's either part of your common administrative or your operational mandate to receive the information in the first place. If you are sharing information with others you should have a contract or MOU stating when the information will be returned to the originator or how it should be kept and disposed of by you. You may be merely a records storage facility and not have legal control over the records.