Supporting document management in complex multitask
environments
Citation for published version (APA):
Bondarenko, O. Y. (2010). Supporting document management in complex multitask environments. Technische Universiteit Eindhoven. https://doi.org/10.6100/IR657436
DOI:
10.6100/IR657436
Document status and date: Published: 01/01/2010 Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
providing details and we will investigate your claim.
Supporting Document
Management
in Complex Multitask
Environments
The work described in this thesis has been sponsored by Océ-Technologies B.V.
The work described in this thesis has been supported by Royal Philips
Electronics, Novay and RUNET Software & Expert Systems.
Cover design: G. Metaxas, A. Sokurova.
A catalogue record is available from the Eindhoven University of Technology Library ISBN: 978-90-386-2153-1
out under the auspices of the J.F. Schouten Graduate
School of User-System Interaction Research.
in Complex Multitask Environments
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus, prof.dr.ir. C.J. van Duijn, voor een
commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op donderdag 11 februari 2010 om 16.00 uur
door
Olha Yuriivna Bondarenko
prof.dr. D.G. Bouwhuis en
prof.dr. J.F. Juola Copromotor:
Contents
Chapter 1. Introduction
7
1.1. Challenges in the support of information workers 8 1.2. The aim and position of the current research 13
1.3. Related work 17
1.4. Overview of this thesis 23
Chapter 2. Documents at hand
25
2.1. Introduction 25
2.2. Related work 26
2.3. Approach 27
2.4. Method and participants 29
2.5. Results: differences 32
2.6. Results: commonalities 36
2.7. Discussion 44
2.8. Conclusions 47
Chapter 3. Task switching in detail
49
3.1. Introduction 49 3.2. Related work 50 3.3. Methodology 52 3.4. Results 55 3.5. Discussion 60 3.6. Conclusions 66
4.1. Introduction 68
4.2. Methodology 73
4.3. Results 82
4.4. Discussion 91
4.5. Implications of the results and further research 96
4.6. Conclusions 101
Chapter 5. Requirements for the design of a digital
document management system
103
5.1. Introduction 103
5.2. Related work 106
5.3. Requirements for the design of a digital
document management system 109
5.4. Discussion 124
5.5. Conclusions 127
Chapter 6. Conclusions
129
6.1. Contribution of this research 131
6.2. Methodology 132
6.3. Limitations and future research 133
6.4. The future of the paperless office 134
Appendixes
137
Appendix A 137 Appendix B 141References
143
Summary
153
Biography
155
Acknowledgements
157
Introduction
The term “knowledge worker” has been introduced by P.F. Drucker in 1959 (Drucker, 1959, p.71) and became commonly used along with the development of information technology at the end of the XX century. Forty years later, Drucker named knowledge workers “the most valuable asset of a 21st century institution” (Drucker, 1999, p. 79). At the beginning of the XXI century, knowledge workers and their work have indeed become essential for the economies of developed world.
The rapid expansion of knowledge work would have been impossible without the technological progress of the last decades. Firstly, the development of electronic computers, started at mid-20th century, progressed to the widespread use of per-sonal computers (PC). Decreasing size and costs and increasing functionality and accessibility made personal computers nowadays available in vast majority of offices and households of the developed countries. A wide range of software tools offered for computer users simplified production of digital documents as means of com-municating information (Bouwhuis, 1989). Secondly, development of global net-works, such as Internet, and tools for communication, such as email, has sped up information sharing and dramatically increased information accessibility. The afore-mentioned technological developments have enabled the defining role of knowledge work in the modern society.
Rapid technological progress has made personal computers the key instrument for knowledge work. Using computers has shifted from being a profession for a few specialists to daily routine for all knowledge workers, most of them not receiving special education in the field of computer science. Along with the development of information technology and increasing number of computer users, a new discipline, human-computer interaction (HCI), has emerged, investigating various aspects of
interaction among machines and human users. Due to the significant role of com-puters in knowledge work, support of knowledge workers has received significant attention within human-computer interaction discipline. In her study, that later be-came one of the corner stones for HCI, Kidd has described one of the first specific requirements for the support of knowledge workers, defining them as those “being themselves changed by the information they produce”(Kidd, 1994, p.186). Kidd also specified two other types of office workers, namely communication workers, whose primary value is communicating of information, and clerical workers, who manage information produced by others. Changes in organizations, occurring in the past decade, let us now speak of “information worker”, in whom all three types of office workers, defined by Kidd, became integrated. The term information workers will be used further on in this thesis.
1.1. Challenges in the support of information workers
The need to cope with rapidly developing technology, together with the increased demands on information workers’ performance, result in a conclusion that informa-tion workers require dedicated support, such as specially designed tools and strate-gies. Overload of information, fragmented across different environments, multitask-ing within an unstructured intermittent workflow, and increased mobility demand can be considered the three major challenges for the support of information work-ers.
Challenge 1. Information overload and fragmentation
The development of information technology has resulted in vast volumes of in-formation becoming instantly available. The number of communication channels, as well as amounts of information passing through these channels, have also dramati-cally increased. Due to these developments, a problem of information overload arose that has become widely recognized in the last years (Edmunds & Morris, 2000). Information overload negatively affects individual and organizational performance (Ho & Tang, 2001) and has been acknowledged as a major challenge for the support of information workers (Kirsh, 2000).
Already in 2006, the number of websites in the World Wide Web has grown to over 100 million (Nielsen, 2006). Many libraries started with digitizing their content and making it available on the Internet, scientific and public press began offering digital editions, companies and institutions progressed in publishing information externally and on internal networks (intranets). Although increased availability of retrievable information is crucial for the performance of information workers, over-supply of retrievable information has become a major source of cognitive overload
(Kirsh, 2000). In addition, with the increasing ease of making information publicly available, it became more difficult to validate its quality, thus resulting in the decrease of information validity and trust (Bouwhuis, 2006; Ho & Tang, 2001).
The increasing number of communication channels has resulted in oversupply of pushed information. The development of electronic mail went as far as multiple email accounts being associated with a single person. Besides that, other communication channels, such as instant messaging, mobile telephony, and communication through various social network tools have also become widely used. Although communicat-ing information might be identified as one of the core tasks of information workers, active use of multiple communication channels that can be poorly controlled on the receiver’s side and social pressure for using these channels, such as an obligation to re-spond to messages, also became significant factors contributing to cognitive overload of information workers (Kirsh, 2000). It is especially hindering the performance of those information workers whose primary tasks are communicating information and coordinating activities, such as managers (Farhoomand & Drury, 2002).
Due to the increased number of information sources and communication chan-nels, information has become fragmented across different places. Information frag-mentation has been recognized as one of the biggest challenges information workers have to deal with (Jones, 2007). Firstly, information workers have to use both paper and digital documents. Although most of information is available digitally, printing volumes are actually increasing over time (Greengard, 1999). Secondly, digital infor-mation is stored in a variety of formats in different places within the digital domain: local file systems, email clients, network places and information sharing facilities. Finally, technology offers a variety of mobile digital devices, starting from portable memory cards and ending up with miniature computers such as smart phones. Vari-ous devices where digital information can be stored are difficult to keep synchro-nized. In addition, they often impose a different organization of information, such as folder structures, that also vary per storage. As a result, information workers have to deal with information spread over multiple places, which all have to be regularly accessed and maintained. Maintenance of multiple storage places and difficulties in retrieving information are only a few factors that contribute to information overload and hinder information workers’ performance.
As a result of information fragmentation, the tasks of information workers have also become fragmented across different environments. A modern information worker has to perform various activities and operate various tools in multiple envi-ronments to achieve his or her goals (Karger, 2007; Kirsh, 2000). For example, to produce a report, one is using a word processor, saves the file somewhere on the PC, seeks information on the Internet and on-line libraries, uses paper notes and library books, and communicates to others face-to-face and via various electronic commu-nication channels. Each of the environments and tools require different interaction
methods and impose restrictions on how things should be done. In many of the environments distracting factors are present (for example, when sending an email, one is most likely to read new messages as well), which drive attention away from the main task. In addition, modern tools often require special skills, such as layout for-matting, to be operated efficiently. One may therefore expect that switching among environments and tools is a major distracting aspect. Drucker suggested that one of the key factors of the productivity of knowledge workers is the ability to concentrate on the main task (Drucker, 1999). One may therefore expect that frequent changes among tools and environments have a negative influence on information workers’ performance and contribute to information fragmentation.
Concluding, overload of information, fragmented over different storage places and communication channels, has been identified as a major challenge for the sup-port of information workers that negatively affects performance and causes stress in the workplace.
Challenge 2. Multitasking within an unstructured, frequently interrupted workflow
The fragmented and hectic nature of the task flow is one of the major challenges for the support of the information workers. Most of them have to deal with a num-ber of tasks simultaneously, all of a different nature, duration and priority. Multi-tasking is now accepted as being common among information workers (González & Mark, 2004). Multitasking causes frequent task switching with, on average, 11 minutes spent on a task before switching to another one. As a result of increased demand to deal with more tasks, on the one side, and increased accessibility through various communication channels, causing interruptions, on the other side, informa-tion workers are most likely to be interrupted numerous times before they complete a task.
For supporting information workers, it is important to understand the nature of their tasks and structure of their workflow. The fundamental differences between manual work, that has been dominant in most of the economies up to the last decades of XX century, and knowledge work, influenced the understanding of the structure of tasks of information workers. For example, assembling a car for a manual worker is a pre-planned set of activities, that is hierarchically decomposed into specified sub-tasks by someone else and is getting to a pre-specified result. Compare it to a routine task of an information worker, such as writing a report, consisting of a succession of largely unpredictable steps, often re-iterated, getting to a result with limited pre-defined requirements. Drucker (1999) suggested that the key factor of productive knowledge work is the autonomy of the worker, in particular, in identifying his or her tasks. “Knowledge work, unlike manual work, does not program the worker” (Drucker, 1999, p.85). Information workers, therefore, are to a large extent freely
defining the way they achieve their goals and accordingly the structure of their task flow. This results in an unstructured and unpredictable workflow, where the notion of task is vaguely identified (Kirsh, 2000).
There is an ongoing discussion in the HCI field with regard to defining the tasks of information workers (Jones, 2007, p.48). Jones proposed the notion of “projects”, representing high-level goals (e.g., buy a house) consisting of several sub-projects and tasks, execution of which is necessary to complete projects. In the recently expanded field of personal information management (PIM), that addresses support of informa-tion workers, the definiinforma-tion of a task used by Bellotti et al (Bellotti, et al., 2004) is widely accepted. The authors proposed to view a task as “something that one can put on a to-do list”. Naturally, the entries on a to-do list can also represent various levels of abstractions, from higher-level activities (e.g., work on a presentation) to more specific actions (e.g., schedule a date for the presentation).
Identifying tasks on varying levels of abstraction may be explained by the Action Identification theory (Vallacher & Wegner, 1987). This theory suggests that same sets of actions can be cognitively viewed at different perspectives by different actors, depending on the complexity of actions, context and actor’s experience. Sequences of actions that are often repeated, become automated, and can be executed subcon-sciously. This allows an actor to move to a higher level of identification (e.g. “get-ting home” instead of “driving a car”) and free up cognitive resources for more ab-stract level of activity. An experienced actor, therefore, executes the sets of automated chunks from the lower levels while keeping only the higher level goal active. Driving a car can be one of the examples of automated chunks: an inexperienced driver has to think about every action, however, in a while, pressing pedals and shifting gears does not require attention anymore. The theory also suggests that 1) people always tend to the higher level of identification of their actions and 2) if interrupted by an unexpected event, an automated set breaks down to the conscious level where every action requires attention.
The phenomenon of interruptions has attracted attention of researchers since the beginning of the past century. It is still being discussed whether interruptions have a negative effect on task performance. For example, the first experiments of Zeigar-nik (ZeigarZeigar-nik, 1927) suggested that interrupted tasks are remembered better than non-interrupted ones. With regard to information workers’ routine, however, the negative effect of interruptions is expected: Kirsh (2000) named interruptions as one of the factors that causes cognitive overload of information workers and is directly related to stress and work burden. In contrast, some laboratory studies showed an increase in performance of an interrupted task (Speier, Valacich, & Vessey, 1997; Zijlstra, Roe, Leonova, & Krediet, 1999).
The latest research puts some light on the influence of interruptions on task per-formance. It has been suggested that complexity is one of the influencing factors
with regard to both main and interrupting tasks, affecting the performance nega-tively (Burmistrov & Leonova, 2003). In the case of simpler tasks, compensatory strategies can be used, reducing total time of the task (TOT) and thus leading to a false impression of increasing performance. Information workers, however, mostly have to deal with complex and unstructured tasks, as discussed above. It could there-fore not be expected that interruptions will have an overall positive effect on their performance.
Herbert Simon (Simon, 1996) proposed that the influence of interruptions on performance depends on the moment they occur within the hierarchical tree of sub-tasks. He implied that an interruption occurring during the moment in between subtasks (so-called “stable state”) causes less damage than an interruption occurring in the middle of a subtask. Interrupted in the middle of a subtask, the actor has to restart the subtask and reconstruct the initial hierarchy. This theory was further supported by later research in HCI (Iqbal, Adamczyk, Zheng, & Bailey, 2005). The described effect was related to cognitive load that drops down during the process of evaluating the result of the subtask and setting up the next step. One may conclude that the unstructured task flow of information workers consisting of multiple activi-ties with vaguely defined tasks and subtasks stimulates negative effect of interrup-tions on the performance.
Concluding, information workers operate in an unstructured workflow, have to work on multiple tasks simultaneously and are frequently interrupted before they complete a task. Multitasking in an unstructured workflow with multiple interrup-tions and task switches during task execution has therefore become another major challenge for the support of information workers.
Challenge 3. Mobility and nomadic work
There is also an increased demand for mobility of information workers. Many of them have to work at multiple work places, which became standard practice for some professions, such as consultants who work for multiple clients or projects at the same period of time. In the last few years, companies are also actively adopting the concept of hot-desking, a setting where an employee does not have an assigned desk but can occupy any vacant place within an office (Millward, Haslam, & Postmes, 2007). Hot-desking results in a significant cost saving for employers due to the reduction of office space and facilities. For the same reasons, employees are often encouraged to work from home part of their working time, which adds a home office to the list of multiple work places. Finally, the concept of nomadic work implies working “any-where, anytime” (Perry, O’Hara, Sellen, Brown, & Harper, 2001) thus extending the workplace to any public place: hotel rooms, airports and even transportation vehicles such as trains. One of the characteristic problems of mobile workers identified by Perry et al. (2001), is that it is impossible to predict whether a mobile worker will
be able to access the place where information is stored when it is required. Mobile workers therefore often prefer to print their documents and take paper along, despite all mobile technology being available. This signals that digital systems do not yet provide sufficient support for mobile information workers.
The absence of a single assigned working place, whether due to mobile work or hot-desking, has significant consequences for the workflow of information workers. A look at any traditional office, where each employee is assigned to a desk, reveals that information workers adjust their environments to a large extent in order to serve their needs (Norman, 1988). Piles of documents, post-it notes, announcements on the walls and folders in cabinets constitute a space with multiple access points to information within the context of one’s work (Kirsh, 2001). Detachment of this context requires additional compensating support from the digital systems.
Concluding, increasing mobility demand requires dedicated support of informa-tion workers with the use of digital tools that are available independently of one’s physical location.
1.2. The aim and position of the current research
We identified three major challenges for supporting information workers: infor-mation overload and fragmentation, multitasking within an unstructured, frequent-ly interrupted workflow, and increasing mobility demand. With the development of digital technology, it could be expected that dedicated support of information work-ers with appropriate digital tools will reduce information and cognitive overload and facilitate performance.
Full reliance on digital technologies could presumably allow reducing and po-tentially eliminating the use of paper documents. The fact that all documents can in principle be digitized has led to the idea of the paperless office, which has been pro-claimed to be “around the corner” for a long while (BusinessWeek, 1975). Up to the current moment, however, the paperless office has never actually become true. Print volumes have even grown with the development of digital technologies (although passing the growth peak around the turn to the XXI century (Hesseldahl, 2008)), also due to the increased availability of personal printers and the decrease of printing costs. Nowadays, it has been proposed that paper and digital documents comple-ment each other (Liu & Stork, 2000) due to the fundacomple-mental differences between the physical and digital domain.
The “digital office” is fundamentally different from its physical counterpart. On the one hand, digital document systems do not represent many of the essential prop-erties of physical documents, such as volume, orientation, visibility and location in three-dimensional space. On the other hand, digital document systems represent other properties of documents, such as date and time of creation, history of
interac-tion, data type. They also allow for various types of interacinterac-tion, such as automatic sorting according to a particular feature (e.g., alphabetic order of file name), auto-matic extraction of metadata (e.g., number of symbols in a document) or search, which are impossible to perform in the physical domain. In part due to these fun-damental differences, attempts to re-create properties of physical space in the design of document management systems, such as digital “piles”, “cabinets” or “mountains” (e.g., Lawrie & Rus, 1999; Mander, Salomon, & Wong, 1992; Robertson, et al., 1998), have not resulted in successful document systems design. It has been sug-gested that a direct mimicking of physical properties will not have an added value for digital document management (Kirsh, 2001; Sellen & Harper, 2002) due to the aforementioned significant differences between the two. Instead, by analyzing the added value of the unique properties and interaction mechanisms that paper docu-ment allow for, it could be possible to transfer the advantages of paper into the digital domain.
Understanding the role of paper documents in the support of information work-ers and transferring advantages of paper to the digital domain is of crucial impor-tance for the support of information workers. Firstly, the use of paper documents contributes to information fragmentation, as information workers have to operate in both physical and digital domains simultaneously. Secondly, increased availability of information sources and communication channels requires active use of documents in their digital form as when trying to print everything one’s office quickly gets bur-ied under growing paper piles. The resulting paper flood revokes advantages of paper, such as reminding, as important documents get lost in between paper piles (Lans-dale, 1991). Finally, increased mobility demand forces the use of digital documents as it is impossible to transfer large amounts of paper in between workplaces. Reduc-ing the reliance on paper could therefore significantly contribute to the support of information workers. However, the continuing active use of paper in modern offices suggests its unique assets that are not sufficiently reproduced in digital tools.
Investigating the advantages of paper is of great importance for research in the domain of Personal Information Management that looks into the support of docu-ment and information managedocu-ment. For most people many of their daily tasks are loaded with information (Jones, 2007), and the amount of this information in a variety of its forms increases. PIM has therefore received growing attention from re-searchers in various disciplines over the recent years. PIM is outlined as “the practice and the study of the activities people perform to acquire, organize, maintain and retrieve information for everyday use” (Jones & Teevan, 2007). Information work-ers, however, use information as input and produce information as output of their work. The term “information worker” is by itself giving an idea of how significant information is for those whose job is to deal with it. This thesis is therefore devoted for the support of information workers in the area of PIM.
The last decade has been very fruitful for the development of PIM. Many research studies, among which those discussed in the next section, provided invaluable in-sights into understanding the needs of information workers, and proposed and test-ed solutions that can move designers forward in developing systems for supporting PIM. The current research, building on previous studies in PIM domain and relevant theories from related disciplines, such as cognitive psychology, aimed to systemati-cally investigate the current needs and strategies of information workers, focusing on the role of paper in information management, and propose a set of requirements for the design of digital systems for personal information management. Provided the significant role of paper and of the physical environment, on one side, and increasing demand to reduce the reliance on paper, on the other side, we focused on identifying and translating the fundamental advantages of paper into the digital domain.
Research questions
Based on the aforementioned discussion and according to the three challenges in the support of information workers, the following main research questions were identified:
1. Which document management strategies are currently applied by information workers in order to reduce information fragmentation and which needs (in par-ticular related to information fragmentation) are supported better by paper than by digital tools?
2. What is the role of information management (in particular, the organization of paper documents) in task switching and interruption handling within a mul-titasking environment?
2a. Which document management strategies are currently applied by informa-tion workers in order to overcome the challenges of multitasking and frequent task switching and to support the smooth transition between tasks?
2b. Are there differences in document management strategies in the paper and digital domains that are directed to overcome the challenges of multitasking and frequent task switching?
3. How can properties of paper documents and collections of paper documents be transferred to the digital domain in order to support information workers who have limited access to a fixed, assigned personal office space?
4. Which requirements for the design of digital document management systems can be derived from the answers on the three previous research questions? The above research questions outline the structure of this thesis. They have served as a starting point for the studies within this project and have been translated into specific research questions in each chapter.
Approach and methodology
Personal information management is a significant part of the activities of informa-tion workers. It is firmly grounded in cognitive processes such as recall, recogniinforma-tion and categorization (Lansdale, 1988). These processes must be therefore investigated and taken into account for the design of any personal information management sys-tem. Research in cognitive psychology, such as investigation of processes and effects of interruptions, discussed above, has significantly contributed to the development of PIM field and understanding of underlying processes in information manage-ment.
The contribution of traditional studies in psychology to the design of informa-tion systems, however, is to some extent limited. Such studies have to be performed experimentally in a laboratory setting to ensure full control over the conditions and influencing factors. The results of the laboratory studies, although having a doubt-less scientific value, are not sufficient for deriving requirements for the design of a digital system. Firstly, the office environment, as any real world environment, consist of an indefinite number of interdependent factors, many of which, such as social or contextual dependencies, are impossible to re-create in a laboratory. The results of laboratory experiments, therefore, are not representative when applied to the situ-ation in a real office environment. For instance, it has been suggested as a result of experimental studies that spatial location of documents is not particularly well remembered (Jones & Dumais, 1986), and therefore should not play a significant role for the design of information management systems, whereas empirical evidence demonstrated the opposite (Lansdale, 1991). Secondly, the results of experimental studies are often difficult to translate into the system requirements, as they provide conclusions on cognitive processes, frequently on a very fine level, that have to be further abstracted in order to be translated into the user needs.
A large body of studies in PIM, therefore, has exploited a variety of ethnographic research methods. Such methods allow taking into account the diversity of inter-weaving contextual factors essential for understanding the needs of the information workers. The variety of ethnographic techniques that has been adopted by HCI re-searchers from social sciences, such as anthropology, include field observations, in-situ interviewing and surveying. Combining these methods helps in creating a full picture of subjects behavior and environment (Naumer & Fisher, 2007). The current research, therefore, will rely on ethnographic methodology to ensure the holistic
view on the needs of information workers with regard to personal information man-agement in the office environment; however, it will take full account of cognitive aspects of information management to reassure the applicability of the results.
The next section will discuss relevant theories from related disciplines such as cognitive psychology and empirical investigations in the domain of PIM.
1.3. Related work
The domain of PIM has received significant attention from researchers, in par-ticular, in the field of HCI, during the last years. The number of research studies and applied solutions for PIM keeps growing. The current section aims at providing an overview of selected works that have influenced the research performed within this project, and relevant theories from related disciplines. A detailed review of relevant research is presented in the introductory parts of each following chapter.
Information and document types
There has been a major discussion about defining the concept of information (i.e., what exactly “information” is, see for a overview (Floridi, 2005)), that remains out of the scope of this thesis. Information, though, is a rather abstract term, whereas we are usually dealing with specific objects containing information, such as documents. Jones (2007) introduced the term “information item” for a tangible representation of a “package” (container) of information, such as a paper document or an email message. Although Jones distinguished between documents and other “information items”, such as web pages, depending onto the degree one can manipulate them, these differences among different types of “information items” are quickly blurring (Jones & Teevan, 2007). For example, one can print a webpage thus converting it to a document that can be freely manipulated. Hence, in this thesis, the conventional term “document” will be used for any information item that information workers may possibly use.
Information workers use documents in a great variety of forms and formats for their work. Information they receive as input for their work is most often shaped in some kind of document, whether it is a book from a library, an article from an on-line journal, a note from a colleague or an email message. They produce, in turn, the same great variety of documents as an output of their work, with this thesis being only one among many examples. Investigating information management, one most frequently ends up with looking into document management as a practice of manag-ing information.
Due to the large variety of types and formats of documents in the modern office, researchers have made a distinction among documents based on their usage. Kwasnik
(1991) has shown that documents’ usage has a major impact on the way they are classified by their owners. Sellen and Harper in their work “The myth of the paper-less office” (Sellen & Harper, 2002) have proposed three major types of documents dependent on how recently they have or will be used. “Hot” documents are currently in use; “warm” documents were used recently or are expected to be used in the near-est future, whereas “cold” documents are not in use for a longer period. Somewhat earlier, Barreau and Nardi in their investigation of digital document organization (Barreau & Nardi, 1995) described three types of information: “ephemeral” (having a short life or a short-term purpose, such as a to-do list), “working” (frequently used information relevant to one’s work) and “archived” (long-term information indi-rectly relevant to one’s work). Both studies, fundamental for PIM, have observed the significant differences in organization and usage of currently used documents and documents that were stored for potential later use.
One of the major differences also relevant for this work is in the activities sur-rounding the documents’ usage. Archived, or “cold” documents, have to be stored and retrieved when needed again; they are therefore stored according to a pre-defined structure that facilitates retrieval. Once retrieved, a “cold” document joins the army of documents in use, until stored again. “Hot”, “warm” and “ephemeral” paper docu-ments usually reside at one’s office desk, often unstructured or structured in an ad-hoc manner according to the needs of an active task, until their relevance expires or the amount of documents exceeds available physical space (Kirsh, 2001). In the digital domain, the most widely spread way of file organization enforces a hierarchi-cal structure, resembling rather a paper archive than an office desk. Digital workplace does not have an “analogue” of the physical desk where currently used digital docu-ments could be kept in an unstructured and visible space.
Differences between the physical and digital environments and their influence on document management of information workers
Ad-hoc organization (i.e., the possibility to place a document “somewhere”) which does not require the user to make specific classification decisions for each document is especially important for information workers, with their unstructured workflow consisting of a variety of unpredictable tasks (Kirsh, 2001; Malone, 1983). Flexibility in ad-hoc organization is one of the major reasons for paper to remain so popular. These reasons have been the subject of many research studies within the PIM domain. For example, Sellen and Harper identified ease of reading from paper compared to a computer screen as one of the major causes for the continuing active use of paper (Sellen & Harper, 2002). Another important reason is the ease of us-ing multiple documents simultaneously, which simplifies activities such as editus-ing, reviewing and collaborative work (Sellen & Harper, 2007). Due to their unique visual appearance and spatial layout, paper documents being placed in easily
vis-ible places within an office serve the important function of reminding about related activities (Malone, 1983). Tangibility and portability of paper make it easy to create a collection of relevant documents and take it along (Liu & Stork, 2000). Finally, a paper document remains the most authentic and trustworthy medium for convey-ing formal appointments (Bouwhuis, 2006; Rimmer, Warwick, Blandford, Gow, & Buchanan, 2008) and therefore paper is actively used for legal purposes.
Paper documents are part of the physical environment, the importance of which has been accepted and investigated by researchers in daily activities of information workers. Most of the tasks of information workers make use of information and are related to information items, such as documents. These items serve as external rep-resentations of tasks, allowing cognitive resources to be freed that would otherwise be loaded with details of all tasks (Jones, 2007, p.71). To-do lists, notes, bills, books, reports and articles placed at the visible locations, are all examples of information about tasks offloaded onto the physical environment. One’s knowledge about the task is partly stored in the world, and uploaded as needed (Norman, 1988). These processes are well grounded in modern theories of cognitive psychology, such as the distributed cognition framework, which propose that human behavior is generated by the interaction between external and internal representations (Hutchins, 1995; Norman, 1988). The framework of distributed cognition, as opposed to traditional approaches to cognition that are not accommodating external representations as part of it, proposed to view internal and external representations as two indispensable components of a cognitive task (Zhang & Norman, 1994). Distributed cognition has been suggested as a framework for HCI that could account for both human pro-cesses as well as interaction with systems (Hollan, Hutchins, & Kirsh, 2000).
Theories within the distributed cognition framework suggested that human activ-ities are to a large extent driven by information from the external world. The theory of Situated Action, introduced by Suchman (1987) considered a human as an actor whose actions are mainly defined by interaction with the environment, thus op-posed to the “purposeful action” approach, which assumed the actions are executed as planned in advance. According to Suchman, plans serve as resources before the action is performed or as reflections afterwards whereas the action itself is situated and highly depended on the environment that is changing the planned sequence. Consequently, by their actions people change the environment, creating new context that dynamically reflects these changes, which, in turn, guide people further in their actions. Dourish (2004) proposed a model of context “in which context and activity are mutually constituted” as a new approach for the design of digital systems (p. 32). “Embodied interaction”, according to Dourish, would ensure that environmental context evolves in the course of user interaction, thus providing meaningful informa-tion based on how it is used.
Further investigating the relationships between environment and action in the context of office setting, Kirsch proposed that physical office workspaces are deeply structured to support task-specific needs of their users (Kirsh, 2001). The notion of an activity landscape, firmly grounded in the cognitive theories of distributed cognition mentioned above, identified the environment as a mentally and physically constructed space, which is actively built up by users from task-related resources while working to accomplish their tasks. Kirsch suggested the important role of task-related artifacts, such as documents, that trigger an action or coordinate relationships among different resources within a task. Dix et al. also suggested the importance of task-related artifacts for effective work in the office (Dix, Wilkinson, & Ramduny, 1998).
We can therefore conclude that the physical environment, of which paper docu-ments are part, plays an important role in the activities of information workers. The office environment, however, is rapidly transformed by new tools and technologies (Jones, 2007, p.52). It is impossible to imagine the modern office without its digital counterpart.
Digital documents offer a wide range of unique advantages. Firstly, the ease of creating and editing a digital document is incomparable even to the most advanced techniques of creating a paper document, such as using a typewriter (BusinessWeek, 1975). Digital storage is getting cheaper over the years, thus, along with miniaturiza-tion of hardware, it became possible to fit the content of a whole library in a pocket. In addition, digital information can be extracted automatically and can easily be made accessible; huge amounts of digital documents can be searched, analyzed and summarized within seconds (Rimmer, et al., 2008).
The digital environment is significantly different from the physical space. Firstly, physical objects possess a large variety of external properties, such as visual features, weight and size. Further, they are located somewhere in a 3D space, which implies re-lational properties such as distance, orientation and location, and are surrounded by other objects (Bouwhuis, 2006). Secondly, human interaction with physical objects is grounded in their knowledge about these properties as well as in their expectations for the physical environment to be consistent and not change without obvious rea-sons. Predictability and consistency of the physical environment is not its feature per se, but it arises from the fact that each human is a product of evolution within this environment and learns to operate it from the very first moment of existence.
Objects in the digital environment possess less variety of external properties, and these properties are of a different nature (Bouwhuis, 2006). Firstly, most popular modern interfaces of digital operating systems, such as Microsoft Windows or Apple Macintosh OS, are icon-based, with the same icon usually representing a class of objects (for example, a document or a picture) rather than a unique object. Although the latest versions of these systems allow displaying a miniature view (a thumbnail)
of the content of certain file types, such as pictures, limitations of screen space prac-tically revoke the advantage of this feature, as visual details of objects are not well distinguishable when miniaturized; further on, thumbnail view is not available for all file types. Although it might be possible to identify each unique object using a combination of properties, such as file name, date of modification, size, location within file hierarchy etc., these properties are essentially different from the properties of physical objects. A file name, for example, must be assigned by the person who creates the file, thus depending on the choice of the creator rather than reflecting the content. In addition, the file name can be changed, thus making the file unidentifi-able. File names and other properties are text-based and therefore require cognitive effort for reading and interpreting, which limits ease of perception of digital objects (Bouwhuis, 2006).
Secondly, there are essential differences in interaction with digital objects, such as multiplication of objects (copying), multiple representations for accessing the same object (“shortcuts”), variations in the object’s behavior, often unpredictable for a non-expert user. For example, clicking on an icon may evoke different processes: opening a file, but also starting an application, playing a video record or even con-necting to a remotely located object. Finally, the digital environment constitutes an essentially dissimilar space to physical: spatial structuring is partly absent and replaced with a hierarchical structure of file systems; visually accessible screen space has no natural correlate in physical space (Hollan, et al., 2000).
Applied PIM research and proposed solutions
The active use of paper in modern offices suggests that digital systems do not suf-ficiently support document management of information workers. This fact, and the increasing need of transferring document and information management to the digi-tal domain has resulted in a large body of research in the PIM domain. Researchers investigated the needs of information workers and proposed solutions for the design of digital systems.
To replicate the advantages of paper, multiple attempts have been made to mimic physical properties in the design of digital systems: displaying collections of digi-tal documents as piles (Mander, et al., 1992), file cabinets (Lawrie & Rus, 1999), representing a 3D view on documents collections (Robertson, et al., 1998). The fundamental differences among the physical and digital domains, discussed above, prevented further development of such systems, as essential object properties and interactions with the physical world could not have been transferred into the digital domain. Modern technology, though, allows for high-quality simulation of physical interaction in the digital domain (Agarawala & Balakrishnan, 2006) and it has been successfully used for purposes other than document management (Hollemans, Wi-jdeven, Bergman, & Loenen, 2006 ).
Research has also been done in combining physical and digital objects into one system, using various technologies such as printing on paper with digital patterns (Guimbretiere, 2003), storing digital information on physical objects (Barrett & Maglio, 1998), projecting information on paper (Holman, Vertegaal, Altosaar & Johns, 2005) or on a desk (Newman & Wellner, 1992). Such systems would allow using the advantages of both domains and would have been helpful for activities where the affordances of paper are crucial for performance, such as collaborative work (Luff, Heath, Kuzuok, Yamazaki, & Yamashita, 2006). Technological limita-tions, however, such as a need of a expensive permanent projector-based set up to ensure the system working properly, prevent such systems from being implemented or make them insufficient for supporting the full range of the needs of information workers in personal information management.
Other studies, discussed below, have tried to exploit the advantages of digital domain to improve the support for PIM based on current digital tools and operating systems in use.
Since digital information can easily be searched on a large variety of attributes, from size and date of creation of a file to a document’s keywords, a number of search-based solutions have been proposed (Cutrell, Robbins, Dumais, & Sarin, 2006; Du-mais, et al., 2003). Although modern search systems have indeed significantly pro-gressed in technical terms, the provided search attributes often do not fully match user needs (Blanc-Brude & Scapin, 2007). In addition, browsing remains the pre-ferred way of navigation through familiar document collections (Bergman, Beyth-Marom, Nachmias, Gradovitch, & Whittaker, 2008; Teevan, Alvarado, Ackerman, & Karger, 2004) that often simply do not require searching as the location of ac-tual information is well-remembered (Barreau & Nardi, 1995). Therefore, providing storage space and search functionality for stored information is not sufficient for a document management system which should also include a flexible organization of currently used documents.
Research showed that the design of current digital systems for document manage-ment implies restrictions on its usage (Marsden & Cairns, 2004) and such systems do not allow for sufficient flexibility in the organization of information (Ravasio, Schär, & Krueger, 2004). Paper documents can be arranged in an ad-hoc way according to what is needed for information worker’s tasks, personality and environment. In contrast, digital systems impose strict requirements for categorization of documents that should be put in a system. Rigid hierarchical structures, where each document should be named and placed in a dedicated folder are often unsuitable for informa-tion workers, especially if their personal preferences (such as to pile up unstructured documents) or specific features of their tasks (such as those of research work) require using ad-hoc structures (Kidd, 1994; Kirsh, 2001; Malone, 1983).
Investigating the organization of digital documents, researchers and designers directed their efforts in developing unified solutions that would combine documents from various sources across tools (Bergman, Beyth-Marom, & Nachmias, 2006; Boardman, 2004; Karger & Quan, 2004), also accounting for flexibility by letting the users to define user-specific properties and categorization (Dourish, et al., 2000). Systems based on combining document types and sources have been proposed to sup-port task and project management (Jones, Klasnja, Civan, & Adcock, 2008; Morteo, González, Favela, & Mark, 2004; Smith, et al., 2003). The principles on which these systems are based have a vast standpoint for the development of user-oriented ment management systems. However, it has been suggested that unification of docu-ment sources and formats can be harmful as the format of a docudocu-ment along with the system where it is stored by itself provide useful information (Jones, 2007).
Finally, the significant role of email in personal information management has been investigated. Email has become the central point of the digital workplace for many information workers (Ducheneaut & Bellotti, 2001). It is not only used for its main purpose, namely sending and receiving electronic messages, but also for docu-ment and task managedocu-ment. The initial design of email clients, however, has not been accounting for such use which resulted in email being heavily overloaded with extra functionality (Whittaker & Sidner, 1996). Researchers explored personal strat-egies of task management in email (Gwizdka, 2004), investigated various approaches to dealing with email messages (Dabbish, Kraut, Fussell, & Kiesler, 2005) and pro-posed email-based solutions for personal document management (Bälter & Sidner, 2002; Kerr & Wilcox, 2004) and task management in email (Bellotti, Ducheneaut, Howard, & Smith, 2003). These solutions also have a good potential for develop-ing a unified personal document management solution and modern email clients have accommodated many task and document management related features. Many documents however, mainly those that are created by the owner of the mailbox, remain outside the email client and therefore such solutions do not allow managing all documents.
1.4. Overview of this thesis
Building on previous research and theories discussed in the above sections, the current research has further investigated the needs of information workers, focusing on the use of paper documents in the office.
Chapter 2 presents the results of an interview-based field study of paper and digi-tal document management in the office to identify advantages of paper. The signifi-cant role of the organization of paper documents for task management is confirmed. Answering research questions 1 and 2a, it is suggested that unique affordances of paper support successful suspension and retrieval of tasks essential for performance
in multitasking environments with frequent task switching. Chapter 3 presents a systematic investigation and classification of task switching patterns and influencing factors. Answering research question 2b, it is suggested that self-switching requires different support than task switching caused by external interruptions. It is also ob-served that self-switching often involves active manipulations of task-related physical artifacts. Based on these insights, we hypothesized that information about a task’s se-mantics is encoded in visually distinctive properties of task-related physical artifacts and that manipulation of these properties is performed in order to represent changes in the task state at the moments of task suspension. Based on this hypothesis, a study was performed, described in Chapter 4, that resulted in a model of the relationships between recognized visual cues and semantic properties of tasks. Answering research question 3, we conclude that properties related to semantics of relevant tasks are currently encoded in the visual properties of paper documents and this mechanism should be transferred to the digital domain. Chapter 5 summarizes the results of the research studies and answers research question 4 by proposing a holistic set of requirements for the design of personal information management systems. Chapter 6 presents the conclusions.
Documents at hand
1
In this chapter the results of a two-year ethnographic study of the personal document management of 28 information workers is described. Both the paper and digital domains were taken into account during the study. The results reaffirmed that document management is strongly related to task management. Digital tools do not adequately support two important user needs related to task management, namely that documents should be embedded within meaningful (task-related) context in-formation, and that they should be easily accessible for regrouping as the task goes on. In contrast, paper supports these needs very well. Following a discussion of per-sonal document management using paper, email, and digital file folder structures, six implications are outlined for the design of digital document management systems that combine the advantages of both domains.
2.1. Introduction
At the beginning of this research project we performed an extensive in-situ interview-based study of the personal document management of 28 information workers. Both the paper and digital domains were taken into account during the study. The main objective of the study was to investigate paper and digital document management of information workers and learn from the use of paper to improve digital tools.
1 This chapter is based on the paper: Bondarenko, O., & Janssen, R. (2005). Documents at Hand: Learn-ing from Paper to Improve Digital Technologies. In ProceedLearn-ings of the SIGCHI Conference on Human Factors in Computing Systems (CHI’2005), New York: ACM Press, 121 - 130.
The extensive use of paper at the times when all documents, in principle, can be made digital, suggested that paper documents provide significant advantages to the document management of information workers that are lacking in the digital domain. Some of the advantages of paper, such as ease of reading or support of collaborating activities, have been discussed before. This was however not sufficient to explain such widespread use of paper documents as can be observed in practice. We therefore hypothesized that information workers share fundamental needs with regard to document management support that are best satisfied by paper document management. Often, a digital document is only a reference of its paper analogue with various degrees of similarity.
Our goal was therefore to identify joint needs in document management across personal strategies and job types, investigate how well these needs are supported by paper and by digital tools and learn from paper to improve digital technology.
2.2. Related work
For decades already, the paperless office seems to be just around the corner. Yet, looking at any modern office there is paper all over it. It is on our desks, in our draw-ers and cabinets. We read from it, write on it, keep it on our desks and take it with us. It is becoming clear that paper is here to stay (Sellen & Harper, 2002).
The fact that information workers continue to rely on paper and digital docu-ments for their everyday work is a clear sign that they need tools where the advan-tages of both domains are taken into account. Whereas, in theory, all documents can nowadays be handled digitally and are from a digital origin, people still use paper because it supports some important user needs that digital tools do not (Bouwhuis, 2006).
The fact that people nowadays still use a lot of paper has inspired investiga-tions into the reasons why they continue to rely on it. Sellen and Harper (2002) described some unique affordances of paper that allow for easy authoring, reading, reviewing and annotating. They also emphasized the convenience of paper for col-laborative working and sharing (Sellen & Harper, 1995). Whittaker and Hirschberg (2001) looked at paper archiving and discussed its implications for digital informa-tion management.
Investigators also focused on the question of how people use paper. For ex-ample, there is an extensive discussion in the latest research about personal strategies in paper (e.g., Whittaker & Hirschberg, 2001). Malone (1983) made a distinction be-tween “filers” (people who clean up and file documents frequently, leaving as little as possible on their desks) and “pilers” (people who do not tend to classify documents, but keep them around in piles).
On the other hand, the importance of digital document management has tre-mendously increased over the last years, and there is a growing body of research in this domain. Studies into Personal Information Management (PIM) investigated the organization of documents with regard to digital tools such as email (Ducheneaut & Bellotti, 2001; Whittaker & Sidner, 1996) and file management (Barreau & Nardi, 1995). Boardman and Sasse gave a good overview of PIM research in their cross-tool study (Boardman & Sasse, 2004).
Email management attracted attention of researches and designers as well. Al-ready in 1996, Whittaker and Sidner (1996) described the problem of email overload. Designed as a tool for asynchronous communication, email is now used for many tasks, such as planning, task management, document and information storage and retrieval. However, email applications were originally not designed to support such a broad range of functionality and they have become overloaded (Bellotti, Duch-eneaut, Howard, & Smith, 2003; Bellotti, DuchDuch-eneaut, Howard, Smith, & Neuwirth, 2002; Ducheneaut & Bellotti, 2001; Gwizdka, 2002). Designers tried to reduce the problem of email overload, in particular by integrating task management into email (Bellotti, et al., 2004; Bellotti, et al., 2003).
More and more research in the HCI domain concentrates on supporting docu-ment managedocu-ment either by improving existing tools (e.g., Bälter & Sidner, 2002) or by providing new solutions (Gwizdka, 2002; Kerr & Wilcox, 2004). In particular, it has been noticed that the contemporary hierarchical file folder structure does not support natural ways of organizing. As a result, alternatives such as a timeline metaphor (Fertig, Freeman, & Gelernter, 1996) or attribute-based organization of documents (Dourish, et al., 2000) were proposed.
2.3. Approach
Despite the growing interest in document management, most of the existing research concentrates either on the paper or on the digital domain. However, in modern offices people use both paper and digital documents, often converting them from one domain to the other.
This study was triggered by the fact that, even in a modern and well-equipped office, desks are still covered with documents whereas, in theory, everything could be handled digitally.
This means that paper has certain advantages that digital systems miss. The goal of the study was to find these advantages of paper and to see if they can be trans-lated into design guidelines for digital document management systems.
Research questions
Based on the existing research and on preliminary investigations within the study, the following research questions were formulated:
1. Are there commonalities in document management strategies (across personal preferences and job types) in particular related to reducing information fragmen-tation?
2. To which extent are these needs in particular, related to information fragmen-tation, currently supported by paper and by digital tools?
3. What can be learned from paper to improve digital tools supporting these needs?
By answering these questions we aimed to identify generic user needs that should be supported by any document management system (DMS, for short).
Research focus
We chose to carry out an exploratory study focusing on the way people currently manage their documents. Our research team did not have any design solution in mind when starting the study. This allowed an unbiased view on the user needs that were to be translated into design guidelines. The focus of the study was narrowed down along the following lines:
Paper and digital domains
The study described in this chapter investigated how people handle their docu-ments in the paper domain (on their desks) and in the digital domain (email, file fold-ers, dedicated tools). The aim was to see how people use different tools in different domains, what the benefits and shortcomings of each of them are, and whether this can provide design guidelines for digital DMS’s.
Non-clerical information workers
Kidd (1994) defined three types of office workers: knowledge workers, commu-nication workers and clerical workers. These types differ with respect to how they deal with document management. Clerical workers, such as secretaries and personal assistants manage somebody else’s documents and document management is often a significant part of their duties. In contrary, non-clerical workers (thus both commu-nication and knowledge workers) manage (part of) their documents themselves, but for them this is not a primary activity – they “have other, more important things to
do”. The target group was therefore restricted to non-clerical information workers, which we refer to as “information workers”.
“Hot” and “warm” documents
The way information workers are using documents within their current activities is different from the way they deal with those that are not of actual need anymore but still have to be kept. Sellen and Harper (2002) defined three types of documents: those that one uses actively at the current moment, those that were just in use or will be used in the nearest future and documents that are not used at the moment. They referred to these types as “hot”, “warm” and “cold” documents, respectively.
“Cold” documents that are not in active use anymore usually end up in a struc-tured collection or archive. Much research has been done in this area (Whittaker & Hirschberg, 2001) and many digital systems were designed for (“DAX Archiving Solutions,” 2008; EMC Archiving solutions,” 2008; IXOS File System Archiving,” 2008) to mention only a few. However, in using “hot” and “warm” documents, people have specific needs that are not adequately met by digital tools. For example, Barreu and Nardi (1995) have reported computer users having troubles with man-aging “ephemeral information” (similar to “hot” documents) already in 1995. This study therefore focuses on the management of documents that are actively in use. As a result, archiving is beyond the scope of this study.
“Documents at hand”
Barreau and Nardi (1995) reported that “... users have no difficulty finding their working information as they use it repeatedly and thus can easily remember where it is.” If people need to search for a document this usually means that the document has been “cold” for a while and its location has been forgotten. Once found, the document is “at hand” (i.e., “hot” or “warm”) and therefore does not need search anymore.
Furthermore, information retrieval tools – such as search engines – cannot serve some important functions that a “document at hand” typically does, such as remind-ing one of an activity that needs to be performed (Barreau & Nardi, 1995). Informa-tion retrieval systems are therefore not within the focus of this study.
2.4. Method and participants
The study was based on contextual inquiry techniques, designed to gather data from the users in the field, where people are working or living. (Beyer & Holtzblatt, 1998). As a subset of contextual techniques, semi-structured interviews were used to explore people’s document collections, paying special attention to documents’ flow:
how do documents appear, move and disappear from desks or computers. Each interviewer followed a list of predefined key issues: what type of job and activities do people perform, what types of documents are in use, what the main information sources are, and which tools and strategies are used. However, the structure of the conversation depended on the specific situation. Every participant was interviewed once. Both paper and digital document management were discussed during the same interview.
We used the artifact walkthrough technique (Hackos & Redish, 1998) to recon-struct artifact flows in the paper and digital domains. It narrowed down the stories from the participants (these were often too general) to the particular example of the life story of a document. The critical incident collection technique (Hackos & Redish, 1998) was used to identify problems that people experience while managing their documents.
An interview took approximately 1.5 hours. For each interview, a separate report was made for further analysis. The report included visual data such as a detailed drawing of the documents in the office (see Fig.1), pictures of the participant’s desk and of digital document collections shown on the PC screen.
Qualitative analysis of the data was performed using an affinity diagram tech-nique (Hackos & Redish, 1998). An affinity diagram session consists of two parts: a brainstorm part in which team members put all the issues they observed on sticky
notes and attach them to a wall, and a structuring part where the notes are catego-rized and the categories are named (see Fig.2). Several affinity diagram sessions were organized during the study.
In total, 28 information workers were interviewed at their workplaces (see Table 1). All of them were experienced computer users, whereas two were experts in com-puter science. Taking such a wide range of people allowed finding commonalities in document management that were job and task independent.
Table 1. Participants of the study: demographics and occupation # Age M/F Position Domain
1-4 29-38 M Financial controller Health, IT
5-8 28-51 3M,F HR manager IT, Edu
9,10 28, 38 F,M HR development manager IT 11-14 26-30 3M,F PhD student Edu 15 32 F Planner Health 16 38 M Engineer IT 17 38 M Software developer IT 18 30 M Researcher IT
19 37 M Assistant professor Edu
20 35 F Office manager Edu
21 50 M R&D manager IT
22,23* 30,40 M,F Project manager Consultancy
24-28* Av 40 4M,F IT consultant Insurance
2.5. Results: differences
Although common patterns in document management were notable for all par-ticipants, there were differences as well. From the results it appeared that the differ-ences were of two types: personal-related and activity-related.
Personal differences and document management
A large part of research into document management has focused on differences in how people manage documents – for example, personal strategies (Malone, 1983; Whittaker & Sidner, 1996) , job types (Kidd, 1994; Whittaker & Hirschberg, 2001) and tools (Boardman & Sasse, 2004; Ducheneaut & Bellotti, 2001). There is an ex-tensive discussion in the latest research about personal strategies in paper (e.g., Whit-taker & Hirschberg, 2001) as well as in digital (Boardman & Sasse, 2004) document management.
Returning to Malone’s distinction between “filers” and “pilers” (Malone, 1983), we have encountered that almost all participants interviewed during the study ap-plied a mixed approach – filing some documents, piling others. However, digital systems usually support filing better than piling. To store a document, one has to categorize it, and, by being placed in the system, a document already belongs to a certain category. Kidd (1994) already mentioned that especially knowledge workers experience difficulties categorizing documents. Yet, a hierarchical file folder struc-ture does not allow one to leave a document “somewhere around” or to put a report aside “to work on it later on”.
