Social analysis in the requirements engineering process:

from ethnography to method

Stephen Viller & Ian Sommerville
Computing Department, Lancaster University, UK
{viller, is}@comp.lancs.ac.uk

Abstract study. Effectively communicating these findings to

Over a number of years, we have been involved in requirements engineers is not straightforward.
investigations into using workplace observation to inform ¥ Culture. There are significant differences in language
requirements for complex systems. This paper discusses and culture between sociologists and software
how our work has evolved from ethnography with developers. This can lead to problems of
prototyping, through presentation of ethnographic communication between the two groups.
fieldwork, to developing a method for social analysis that ¥ Abstraction. It is difficult to draw design principles
has been derived from our experience of applying and other abstract lessons from a technique that is
ethnographic techniques. We discuss the strengths and concerned with the detail of a particular situation.
weaknesses of each of these approaches with a particular ¥ Skill. The lack of a systematic approach to conducting
focus on our most recent work in developing the ethnography makes the technique dependent on the
Coherence method. This method is based on a fusion of individual ethnographerÕs skill.
viewpoint-oriented and ethnographic approaches to We have been addressing these issues in our research
requirements engineering and uses an industry-standard for several years. We are not alone in wanting to improve
notation (UML) to represent knowledge of work. We use a how RE deals with social and organisational factors. Our
common example of an air traffic control system to experience with ethnography is analogous to many others
illustrate each approach. investigating the boundaries between the social and the
technical. These include other ethnographically informed
1. Introduction approaches [2, 4, 15], Participatory Design [17], and
sociotechnical methods [5, 7].
This decade has seen ethnography become increasingly Rather than engaging in a broader theoretical debate,
popular in the requirements engineering (RE) community. this paper wishes to convey our practical experiences. The
Studies have been performed in a variety of domains, following sections describe our experience with three
including: underground control rooms [8], air traffic different approaches to ethnographically informed design.
control [1], and banking [3]. By spending time alongside They present not only a chronological view of how the
workers, observing what they do, ethnographers develop a work has moved on, but also a progression from the locus
deep understanding of the work. Ethnographers can of ethnography being in the ethnographersÕ heads,
therefore provide designers with detailed insights into the towards being embodied in the method itself [4].
work as it is actually performed, presented in the
language and terminology of the users. Ethnographic 2. Ethnographers working with designers
studies can uncover subtle features of the social nature of
work that are vital to successful operations, yet at the Early work at Lancaster on ethnographically informed
same time appear to be so trivial that other techniques can design took place on a project concerned with developing
miss them. This is in contrast to some other human- displays for air traffic controllers [1]. This was the first
centred approaches, which tend towards studying project at Lancaster to involve sociologists and software
simulations of work in artificial laboratory-based settings, engineers collaborating together in the design process.
and imposing their own vocabulary to describe the work. Reports from both sociologists [12], and software
Ethnography has a great deal to offer as a technique for engineers [18] reflected their respective concerns about
RE, but a number of issues limit its use in practice: the nature of this collaboration.
¥ Time. Ethnography can be a very lengthy process, The model of work involved periods of fieldwork by
lasting months or even years in the context of social the ethnographers, in parallel with prototype development
research. Much shorter time-scales are required for the by the software engineers, followed by de-briefing
RE process. meetings where their respective findings were reported
¥ Results. Ethnography tends to produce a great deal of back to the rest of the team. This model was subsequently
detailed, textual description as a result of performing a referred to as concurrent ethnography [9] (Figure 1).

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
 Later on in the study, the description in the notes
becomes much closer to a transcript of the talk alone, as
Ethnographic Systems in this excerpt.
Study Debriefing Meetings ...
Development
Controller: ‘Speedbird 799L … delays running at 13-15
minutes at the moment … I’ll try to keep you advised.’
Pilot: ‘Speedbird 799L … thank you.’
Controller: ‘Speedbird 799L … descend flight level 120’
Pilot: ‘Speedbird 799L … roger, descend flight level 120’
System Prototype Writes it on strip
(takes place over about a three minute period)

Figure 1: Concurrent ethnography In this second excerpt, there is more taking place, but
the ethnographerÕs familiarity with the domain has
During the periods of fieldwork, the ethnographers removed the need for detailed description.
observed controllers at work, making audio recordings of In the debriefing meetings, however, it was not the
the conversations, supplemented with notes. The tapes fieldnotes that communicated information to the software
were transcribed, and the transcriptions augmented with engineers. Rather, it was the ethnographer who acted as a
further details from the notes. These notes were used in ÔproxyÕ for the field site. Their familiarity with the setting
the debriefing meetings to illustrate points that the enabled them to respond to questions and suggestions on
ethnographer wished to make and in support of comments the controllersÕ behalf.
about the latest version of the prototype or design
suggestions proposed by the software engineers. Figure 2 2.1 Strengths
illustrates this pattern of working. Involving ethnographers in a prototyping cycle has a
Debriefing
meeting number of benefits. The ethnographer can bring insights
about the domain to the debriefing meetings, that other
approaches would struggle to obtain. The strength of
problem these insights is that they are what actually happens.
domain
Observation can also uncover workplace features that are
Ethnographer Requirements System
not obvious to the participants being observed, because
engineer specification they are so deeply embedded in what they do on a day to
day basis.
EthnographersÕ understanding of the workplace can be
used when evaluating prototypes, where they can act as a
ÔproxyÕ for the actors in the domain. Disruption can
Field notes
therefore be minimised for Ôreal usersÕ.
Figure 2: Ethnographers collaborating with
software engineers 2.2 Weaknesses
Fieldnotes take the form of a diary or journal, with This early work highlighted the issues identified in
short descriptions of the interaction taking place between section 1. A good ethnographer can provide useful, and
controllers and with pilots. These are supplemented with sometimes counterintuitive (for software engineers, at
comments on the nature of the work being observed, least) insights into the effects of introducing technology
reports from other activities (e.g., a report on a session into the workplace. However, the wealth of information
using an ATC simulator), transcriptions of interviews that is contained within an ethnographerÕs notes remains
with controllers, and so on. largely untapped, because of their detailed, unstructured,
Early on, the notes contain detailed descriptions of the and highly personal nature.
various activities engaged in by the controllers. As the 2.3 Outcomes
observations continue, however, the descriptions
gradually adopt the terminology and abbreviations used A number of specific findings from the ATC project
by the controllers, without further explanation. For changed our approach to the development of interactive
example, the following excerpt from the first visit systems. For example, Ôcommon senseÕ approaches to
describes the coordination of a flight, which takes place automation in software engineering could turn out to be
whenever a flight is not going to pass between sectors detrimental for what appear to be routine manual tasks,
according to a Ôstanding agreementÕ. which actually incorporate error checking procedures or
I watched as the controller began to write ‘↓260L’ in red on that allow system users to build and maintain a mental
a strip, whist at the same time instructing a plane “descend domain model. There is also a tension between allowing
flight level 260 to be level at the Isle of Man …” This strip
was towards the bottom of one of three vertical positions, users to tailor their interfaces while other users need to be
each separated by yellow plastic strips with beacon able to understand the displayed information Ôat a glanceÕ.
information printed on them eg IOM, along with some
heading and frequency information…

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
3. Modifying ethnography Presentation viewpoints. The first attempt to modify
how ethnographic analyses are presented used a number
Following on from the ATC project, a number of of viewpoints, each one addressing a particular aspect of
collaborative projects coordinated at Lancaster extended the social organisation of the work [10]. The viewpoints
our experience with using ethnography in design. Most of focus how ethnographers present their understanding of
the work was concerned with how ethnographic the domain, as recorded in their field notes. This model of
techniques could be modified to orient the results towards working differs from the previous situation (Figure 2) by
the needs of RE. The modifications proposed were both in the fact that communication between ethnographers and
terms of the process by which ethnography is integrated designers is mediated by the description that is structured
into RE, and also the form in which the results of and presented as viewpoints (Figure 3).
ethnographic studies are communicated to designers. Presentation Debriefing
viewpoints/ meeting
framework
3.1 Moving out from the control room
In a paper that recognised, and attempted to address, problem
domain
ethnographyÕs limited contribution to RE in general [9], a
number of different processes for integrating ethnography Ethnographer Requirements System
into design were examined. These included: engineer specification
¥ Quick and dirty ethnography. This directly
addressed the criticisms of the time scale involved
with ethnography. Several short, focused studies take
place, interspersed with debriefing meetings, similar to
those in concurrent ethnography. The shorter time Field notes

scale for this approach is justified because diminishing Figure 3: Modifying ethnographic records
returns set in quite early with ethnography in RE.
Quick and dirty ethnography allows the effort of the Three viewpoints were identified in this work:
ethnographer to be redirected towards new points of The ecology of work is concerned with representing the
interest as they arise in the ongoing fieldwork. spatial distribution of the workplace. The participants,
¥ Evaluative ethnography. This is a focused version of where they work, and the resources that they use are
quick and dirty ethnography. Where it differs is that it relevant to this viewpoint.
is intended for evaluating an existing design, or in In the ATC study, the spatial layout of each suite, and
other words performing a Ôsanity checkÕ on an already its situation relative to other suites bears a direct
formulated design. relationship to the physical organisation of the airspace.
The above are concerned with the process of ATC suites controlling neighbouring sectors of airspace
ethnographers and designers working together, which are placed next to each other. This enables controllers to
address some of the issues presented in section 1. For coordinate flights by talking with their neighbours, and
example, quick and dirty ethnography greatly reduces the allows them to get a picture of their future workload
lengthy duration of study. Evaluative ethnography opens thanks to the Ôat a glanceÕ availability of paper flight
up new possibilities for using ethnography in RE. strips in the racks of neighbouring suites.
However, these alternative processes still follow the basic Views of work is a collection of possible viewpoints
model of ethnographers understanding and describing the from which the detailed ethnographic data can be
problem domain, and designers making use of the presented. Conversations with controllers about the nature
information provided by them. of their work appear here, alongside the ethnographerÕs
observations and remarks.
3.2 Presenting ethnography in RE The flow of work can also be considered as a collection
of potential viewpoints focused on workflow, information
Of all the issues identified in section 1, the most
flows, etc. In the ATC setting, this viewpoint includes the
difficult to address are those concerning how the results
tracking of an aircraft through controlled airspace, and the
of ethnography are communicated to designers. Field
notes are detailed, discursive documents, with little related process of working with the paper flight strips.
structure, and are oriented towards understanding and The presentation viewpoints were supported in an
describing the social nature of a workplace. Further, the augmented version of a tool previously developed at
backgrounds of software engineers and sociologists make Lancaster [20]. This provided a means for cross-
communication between the two disciplines open to referencing the ethnographic material with abstract
misunderstanding. The approach taken at Lancaster was models of the system as they were developed.
to examine how the results of ethnography can be Presentation framework. Following on from the work
modified to be more appropriate for RE. In particular, this on presentation viewpoints, the Presentation Framework
took the form of presentation viewpoints [10] and, later, a [11] structures ethnographic data in terms of three
presentation framework [11].

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
dimensions of work; ÔDistributed CoordinationÕ, ÔPlans 3.4 Weaknesses
and ProceduresÕ, and ÔAwareness of WorkÕ.
Distributed coordination is concerned with how tasks Whilst cited as a strength, the tool used was non-
are performed within the broader context of the standard, and therefore difficult to integrate with other
organisation, as steps in continuing processes, and as part tool support. None of the modified approaches to
of a division of labour. Workers rapidly build up an ethnography addressed the problems of communication
understanding of what constitutes their work, and what is between sociologists and software engineers.
Òsomebody elseÕsÓ. Looking at work from this perspective 3.5 Outcomes
is useful for understanding the roles played by different
individuals as they collaborate together. The work that followed on from the ATC project has
Returning to the ATC study, this dimension is shown that a number of different approaches to using
concerned with how the paper flight strips, for example, ethnography in design are possible. The Ôquick and dirtyÕ
which represent the progress of a flight through a sector, approach has been used extensively, and these studies
are used to coordinate the work of various actors. Features have reinforced the appropriateness and utility of the
of the flight strip determine actions to be taken by the categories used in the presentation framework. The
various actors, and colour coded annotations on the strip presentation viewpoints served to demonstrate that
record who did what with it. viewpoints are a useful concept for structuring
Plans and procedures focuses on how the organisation ethnographic fieldnotes and making them available to the
supports distributed coordination through job design process.
descriptions, workflow diagrams, instruction manuals and
procedures, etc. Of particular interest here is the way that 4. Ethnographically informed method
work in practice can differ from documented procedures.
An example of this from the ATC study was the The most recent work at Lancaster in this area has
practice of controllers deliberately placing flights on been conducted in the Coherence project. This represents
conflicting paths to solve an immediate problem before a fusion of the research on the presentation framework
returning the flights to safe routes. with other work conducted at Lancaster on viewpoints for
Awareness of work refers to how individuals perform RE [19]. The difference between this and the work on
their tasks so that what they are doing is made ÔvisibleÕ or presentation viewpoints is that Coherence can fit into an
ÔavailableÕ to others. Two people working alongside each established framework for eliciting and reasoning about
other will have a good impression of what each other is requirements from a number of perspectives.
doing, without being explicitly informed by their In the Coherence project, we have been concerned
neighbour. with addressing the issues in section 1 from a different
An example of this that has emerged from a number of approach to previous work. Rather than modifying the
control room based studies, including ATC, is gaze process of ethnographic study or its outputs, we have
awareness. A controller will often know which flight developed an ethnographically informed approach to RE.
another is talking about before they explicitly identify it, Figure 4 presents how this approach differs from the
simply because one can see where the other is looking. previous models of work.
Cutting across these dimensions are a number of other
aspects of work. These are used to further structure the problem
presentation of features in the ethnographic record that are domain
pertinent to the design in question. Focus
Requirements System
3.3 Strengths engineer specification
Of the work focusing upon the process of ethnography,
the Ôquick and dirtyÕ approach directly addressed the time
taken to perform ethnographic studies. Useful results can
be obtained within a shorter time-scale, and debriefing
meetings can be used to redirect the ethnographerÕs effort Ethnographically
to other aspects of interest in the domain as they emerge. informed method
The main benefit of the work on the presentation Figure 4: The Coherence method
viewpoints and framework was the structuring of
ethnographic data, and orienting it towards RE. Coupled Rather than requirements engineers relying on an
with Ôquick and dirtyÕ ethnography, this makes the ethnographer for analysis of the social nature of the
approach accessible to designers and procurers [3]. The workplace, they now perform the analysis themselves,
tool support made the fieldwork notes more accessible to supported by ethnographically-informed guidance. The
designers and enabled cross-referencing between design Coherence method delivers this guidance in a systematic
models and the fieldwork. manner by using viewpoints to structure the requirements
elicitation and analysis.

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
4.1 Viewpoint-oriented requirements Table 1: Social viewpoint questions

Coherence brings together two well-established Distributed Coordination

threads of research at Lancaster University: the ¥ How is the division of labour manifest through the work of
presentation framework for design described in section individuals and its coordination with others?
3.2, and viewpoint oriented RE. ¥ How clear are the boundaries between one personÕs
Viewpoints are encapsulations of information about a responsibilities and anotherÕs?
system or process, each one being a partial analysis of the ¥ What appreciation do people have of the work/tasks/roles of
others?
workplace as seen from a particular perspective. For
¥ How is the work of individuals oriented towards others?
example, a pilot or controller would both be viewpoints in
Plans and Procedures
an analysis of an ATC system, as would be the Manual of
¥ How do plans and procedures function in the workplace?
Air Traffic Control which defines standard ATC
¥ Do they always work?
procedures. Reconciling and integrating the separate
¥ How do they fail?
viewpoints leads to a complete analysis of the system.
¥ What happens when they fail?
Viewpoints have been conceptualised and
¥ How, and in what situations, are they circumvented?
implemented in a variety of ways in RE [6, 14]. The
Awareness of Work
particular viewpoint oriented approach we use is called
¥ How does the spatial organisation of the workplace facilitate
PREview [19]. What distinguishes PREview is the use of interaction between workers and with the objects they use?
concerns to drive the analysis. Concerns shift the ¥ How do workers organise the space around them? Which
perspective of PREview from what the system should do, artefacts that are kept Ôto handÕ are likely to be important to
to how it can best serve the organisation. They explicitly the achievement of everyday work?
link organisational goals and objectives with system ¥ What are the notes and lists that the workers regularly refer
requirements. Concerns such as safety, compatibility, etc. to?
are elaborated into questions, which must be asked of ¥ What are the location(s) of objects, who uses them, how
every viewpoint source to collect information about the often?
system, and/or external requirements, which apply across
all viewpoints to ensure compliance with the Table 2: Social concern questions
organisational concerns. Paperwork and computer work
The structure of viewpoints, each of which focuses on ¥ How do forms and other artefacts on paper or screen act as
a particular aspect of the workplace, and concerns, which embodiments of the process?
cut across all of the viewpoints, is analogous to the ¥ To what extent do the paper and computer work make it
structure of the dimensions and aspects of work adopted clear to others what stage people are at in their work?
in the presentation framework. This is exploited in ¥ How flexible is the support for the work processÑis a
Coherence, where the framework is represented as social particular process enforced, or are alternatives permitted?
viewpoints and concerns. Skill and the use of local knowledge
¥ What are the everyday skills employed by individuals and
4.2 Social viewpoints and concerns teams in order to get the work done?
Coherence structures the three dimensions of work in ¥ How is local knowledge used and made available, e.g.
the presentation framework as viewpoints. To assist their through the use of personalised checklists, asking experts,
elaboration in a given context, Coherence provides a etc.?
number of questions to consider when elaborating each ¥ To what extent have standard procedures been adapted to
viewpoint (Table 1). These questions are not checklist take local factors into account?
items that must be responded to but are proposed as Spatial and temporal organisation
guidance for analysts, to sensitise them to the social ¥ How does the spatial organisation of the workplace reflect
how the work is performed?
features of the workplace. Elaboration of the social
¥ Which aspects of the work to be supported are time-
concerns is similarly supported by a number of questions dependent?
for each concern, and these are presented in Table 2. ¥ Does any data have a Ôuse-by-dateÕ?
In determining the relevance of the concerns in a given ¥ How do workers make sure that they make use of the most
context, each of the questions in Table 2 are re-cast to up-to-date information?
make questions of the form ÒDoes X exist?Ó rather than
ÒHow is X manifest in the workplace?Ó For example, ÒDo Organisational memory
forms and other artefacts on paper or screen act as ¥ How do people learn and remember how to perform their
embodiments of the process?Ó or ÒAre there aspects of the work?
work to be supported that are time dependent?Ó If the ¥ How well do formal records match the reality of how work
is done?
answer is mainly no, then the concern can be eliminated
from the analysis, thus reducing the amount of work Once irrelevant concerns have been eliminated, those
required to reconcile each concern with every viewpoint. remaining must be elaborated, with the help of the

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
questions in Table 2. Each concern record consists of the well as through attention to the workplace from the
concern name, a description, and any external viewpointÕs perspective. They are noted in the viewpoint
requirements and/or questions arising from the concern. record, and elaborated elsewhere. In this case, the three
Table 3 presents an example from the ATC study to requirements listed relate to:
illustrate a social concern that has been elaborated. In the ¥ providing support for controllers to make their work
table, the first cell contains the title of the concern, and a available for scrutiny by others (AW1);
description of workplace features relevant to the concern, ¥ providing information on controllers work so that it
inspired by the pertinent questions in Table 2. The second may be scrutinised (AW2); and
cell includes a number of external requirements (ERs) ¥ attending to how the physical layout of the control
arising from the concern, which must be considered by suites maps onto the layout of the airspace (AW3).
each viewpoint in the analysis. Questions arising from the Table 4: Example social viewpoint for ATC
concern, which would be asked of each viewpoint source,
appear in the third cell (none in this instance). Name: Awareness of work
Table 3: Example social concern for ATC Focus: How the physical organisation of the control suites
affects how controllers can make sense of each
Paperwork and computer work other’s activities. How controllers monitor the work
of other controllers, and how controllers orient their
The main feature of a control suite that this concern is work to facilitate others monitoring it.
interested in is the flight strip itself. As a consequence, the
following concern questions responses all focus on how the Concerns: Paperwork and computer work
flight strip is used by controllers in the course of their work. Skill & the use of local knowledge
Flight strips embody the process of an aircraft’s progress Spatial and temporal organisation
through the sector of airspace controlled by a suite. As an Organisational memory
aircraft approaches the sector, its strip is moved progressively Safety
to the bottom of the rack until it becomes the current strip for Capacity
the controller to deal with. The work of the controller can
therefore be viewed in terms of dealing with the flow of strips Sources: Controllers, and observation of controllers at work
as aircraft enter, traverse, and leave the controller’s sector.
Requirements
The collection of strips in various racks in a suite provide AW1 (Making work available)
an ‘at a glance’ means of determining the current and future AW2 (Availability of awareness information)
workload of a particular controller. The practise of ‘cocking AW3 (Relationship of suite layout to controlled
out’ strips—raising them slightly in the racks—informs the airspace)
controller that there is something non-standard about the flight
concerned. Glancing at the strips provides a controller with an
indication of their current and future workload, in the same 4.3 Linking with system models
way as it allows other controllers to see the relative loading on
other sectors. One of the major concerns we had when developing
Flight strips provide incredibly flexible support for the work Coherence was that it should be of use to requirements
of controllers. Different practices exist regarding whether
strips are placed into the racks in a top to bottom sequence or engineers in practice. Therefore, we decided to look for
vice versa. All instructions given by controllers to pilots, and an existing notation which could be used to document the
the pilots’ acknowledgements, are recorded onto the relevant social analysis. The notation we decided on is the Unified
flight strip. These annotations are made using a standard set
of symbols, and different coloured pens according to the Modelling Language (UML) [16]. Our first task was to
annotator’s role within the controlling team. In this way, flight establish that UML is capable of expressing information
strips record a flight’s progress through a sector. about the social nature of workplaces [21]. We have
ER1. The system shall support controllers ‘getting the subsequently been concerned with the process of
picture’ by providing the ability to determine current and future
load for a sector ‘at a glance’ undertaking social analysis in a systematic way, and
ER2. The system shall provide a facility to mark providing links from our work to standard approaches to
exceptional or non-standard flights requiring special attention requirements analysis and systems development.
ER3. Annotations to flight records shall be recorded and Coherence is particularly suited to use case driven
presented in such a way that they identify the person who requirements analysis [13], as there are a number of
made them.
correspondences between the two approaches. For
No questions for this concern.
example, transcripts that result from observation of work
Returning to the ATC example, Table 4 illustrates the such as that presented in section 2 can be modelled as
Awareness of work viewpoint. Here, it can be seen that sequence diagrams, which describe the scenarios which
the Organisational memory concern has been eliminated make up use cases (Figure 5)
from the analysis as not relevant. There are also two When Coherence is used alongside PREview analysis,
organisational concerns, namely Safety and Capacity. a number of additional domain viewpoints are identified
These reflect two goals that the organisation has for the and elaborated. An important class of viewpoint is known
introduction of any system in that safety should not be as interactor viewpoints, which correspond directly to
compromised, and the capacity (i.e. the number of aircraft actors in use case models. Use cases themselves are
that can be handled in a given time scale) should be generated from the plans and procedures viewpoint,
maximised. Requirements are generated by considering which focuses analysis on workflow. Being informed by
the pertinent questions for the viewpoint in Table 1, as ethnography, Coherence is concerned with how work is

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
actually performed, and as a consequence how it may
differ from documented procedures. It can therefore be
used to generate initial use case models, based upon
observed interaction, such as that presented in Figure 6. ActiveController Standard Flight

<<Extends>>
<<Uses>>

Flightstrip
Controller Speedbird 799L
ChiefController Non-standard flight
<<Uses>>

Controller: Speedbird 799L

there is a delay at Lambourn
slow down if you wish and on
reaching Lambourn I will require Create flight strip
you at 110 or less take the hold <<Extends>>
at Lambourn
Pilot: Speedbird 799L, roger
hold at Lambourn. <<Extends>>

Create pending strip

Controller: Speedbird 799L
delays running at 13-15 minutes at
the moment I ll try to keep you
advised.

Pilot: Speedbird 799L thank Create live strip

Assistant
you.
Controller

Controller: Speedbird 799L Figure 6: an initial use case model for the ATC
descend flight level 120
study
Pilot: Speedbird 799L roger, 4.5 Weaknesses
descend flight level 120

Whilst it could be claimed that Coherence has removed

Writes it on strip
the communication problem between sociologists and
software engineers, the effect has been to shift the
problem from one of understanding people in
conversation, to understanding people via the methodÕs
Figure 5: Observed interaction modelled in a documentation. This places an onus on Coherence to
sequence diagram communicate the intricacies of understanding social
Thanks to the lessons learned from experience with interaction to an audience with little or no background in
Ôquick and dirtyÕ ethnography, the time scales involved in the social sciences.
Coherence are similarly reduced in comparison with the
initial work on the ATC project. Using a systematic 4.6 Outcomes
approach to the analysis also means that requirements In the Coherence project, we have tried to demonstrate
engineers who do not necessarily have any training in that ethnographically informed RE can make the
sociology can apply the lessons embodied in the transition into a systematic approach for social analysis.
presentation framework. Conversely, this work also We have demonstrated that an industry standard notation
enables ethnographers to use viewpoints and a standard can be used to represent the type of information that
notation to present the results of their studies to designers. ethnographic studies reveal about the social nature of
4.4 Strengths work [21]. We have also shown how our approach can be
used in concert with other established approaches to RE
The approach to ethnographically informed design in to make an ethnographically informed understanding of a
Coherence has a number of strengths. First and foremost, domain available to the rest of the RE process.
we are using an industry standard notation to describe the
social features of a domain alongside the rest of the 5. Conclusions and further work
requirements. Our approach is flexible in that it can
complement a viewpoint-oriented approach to RE, or be This paper has presented a review of the work at
used to provide social analysis as a Ôfront endÕ to any Lancaster University on integrating ethnography into the
other preferred approach. We have examined use case requirements process, which has culminated in the
driven design as one approach in particular, and found ethnographically informed Coherence method of social
that Coherence analysis can assist in the identification and analysis. Previous work in this field has made progress by
description of use cases, scenarios, and actors and objects modifying the process of ethnographic study, and how its
in the domain. A further benefit of adopting and results are presented. Our approach in Coherence has
integrating with standard approaches to analysis and been from the other direction, aiming to produce a method
design is that it will be easier to draw out general and that is informed by ethnography, rather than modify
abstract lessons about the social nature of work, and how ethnography to suit the needs of design. The resulting
to reuse them in future designs.

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.
method has a number of attributes which facilitate the now?: Ethnography in the commercial world. Interactions 4,
incorporation of social issues into system requirements. 3 (1997) 38-47.
¥ The method itself is informed by ethnography, 4 Button, G. and Dourish, P., Technomethodology: paradoxes
providing an encapsulation of our experience of using and possibilities. In Proc. CHIÕ96 (Vancouver, 1996) ACM
Press, 19-26.
ethnography in RE. 5 Eason, K., Information Technology and Organisational
¥ We have used an existing viewpoint-oriented approach Change. Taylor & Francis, London, 1988.
to structure analysis in Coherence, and to enable 6 Finkelstein, A., Kramer, J., Nuseibeh, B., Finkelstein, L. and
requirements arising from the social analysis to be Goedicke, M., Viewpoints: A framework for integrating
reconciled against those from other sources. multiple perspectives in system development. International
¥ In contrast with ethnography, Coherence provides a Journal of Software Engineering and Knowledge
systematic approach to social analysis which does not Engineering 2, 1 (1992) 31-58.
require a background in sociology to apply. 7 Harker, S.D.P., Eason, K.D. and Dobson, J.E., The change
¥ Coherence uses a standard notation (UML) to and evolution of requirements as a challenge to the practice
communicate its results to the rest of the RE process, of software engineering. In Proc. REÕ93 (San Diego, CA,
1993) IEEE Computer Society Press, 266-272.
rather than burden requirements engineers with yet 8 Heath, C. and Luff, P., Collaboration and control: crisis
another notation to learn. management and multimedia technology in London
¥ Finally, we have provided links to use case and object Underground control rooms. Computer Supported
models to enable Coherence to act as a Ôfront endÕ to Cooperative Work 1, 1 (1992) 69-94.
existing approaches which do not currently address 9 Hughes, J., King, V., Rodden, T. and Andersen, H., Moving
social issues in the workplace. out from the control room: ethnography in system design. In
We believe that Coherence is a significant step Proc. CSCWÕ94 (Chapel Hill, 1994) ACM Press, 429-439.
forwards in terms of making ethnographically informed 10 Hughes, J., OÕBrien, J., Rodden, T., Rouncefield, M. and
analysis acceptable in mainstream RE. Yet there are still Sommerville, I., Presenting ethnography in the requirements
areas requiring further attention. Users other than the process. In Proc. REÕ95 (York, 1995) IEEE Computer Soc.
Press, 27-34.
authors have not yet used Coherence in a ÔrealÕ 11 Hughes, J.A., O'Brien, J., Rodden, T. and Rouncefield, M.,
development context. Feedback from usage such as this Designing with Ethnography: A Presentation Framework for
will be invaluable for making the method more usable, Design. In Proc. DISÕ97 (Amsterdam, 1997) ACM Press,
and understanding where extra guidance is required for 147-159.
the social analysis. Tool support does not exist for 12 Hughes, J.A., Randall, D. and Shapiro, D., Faltering from
Coherence, but is in development for the PREview Ethnography to Design. In Proc. CSCWÕ92 (Toronto, 1992)
approach which underlies it. ACM Press, 115-122.
A working document is available from the Coherence 13 Jacobson, I., Christerson, M., Jonsson, P. and …vergaard, G.,
projectÕs world wide web pages at Object-Oriented Software Engineering: A Use Case Driven
http://www.comp.lancs.ac.uk/computing/research/cseg/ Approach. Addison-Wesley, Reading, MA, 1992.
14 Leite, J.C.S.P., Viewpoint analysis: a case study. ACM
projects/coherence/ describing the method in more detail. Software Engineering Notes 14, 3 (1989) 111-119.
15 Potts, C. and Hsi, I., Abstraction and context in requirements
6. Acknowledgements engineering: toward a synthesis. Annals of Software
Engineering 9, (1997) 1-39.
We wish to acknowledge the contribution to the 16 Rumbaugh, J., Jacobson, I. and Booch, G., The Unified
ongoing work on ethnographically informed RE by Modelling Language Reference Manual. Addison-Wesley,
various members of the sociology department, in Reading, MA, 1999.
particular John Hughes, Jon OÕBrien, Dave Randall, and 17 Schuler, D. and Namioka, A., Eds., Participatory Design:
Mark Rouncefield. Thanks also to Tom Rodden and the Principles and Practices. Lawrence Erlbaum Associates,
Hillsdale, NJ, 1993.
anonymous reviewers for comments on this paper. The 18 Sommerville, I., Rodden, T., Sawyer, P. and Bentley, R.,
Coherence project is funded by the UKÕs Engineering and Sociologists can be surprisingly useful in interactive systems
Physical Science Research Council. design. In Proc. HCIÕ92 (York, 1992) Cambridge University
Press, 341-353.
7. References 19 Sommerville, I., Sawyer, P. and Viller, S., Viewpoints for
requirements elicitation: a practical approach. In Proc.
1 Bentley, R., Hughes, J.A., Randall, D., Rodden, T., Sawyer, ICREÕ98 (Colorado Springs, 1998) IEEE Computer Soc.
P., Shapiro, D. and Sommerville, I., Ethnographically- Press, 74-81.
informed systems design for air traffic control. In Proc. 20 Twidale, M., Rodden, T. and Sommerville, I., The Designers'
CSCWÕ92 (Toronto, 1992) ACM Press, 123-129. Notepad: supporting and understanding cooperative design.
2 Beyer, H. and Holtzblatt, K., Contextual Design: Defining In Proc. ECSCWÕ93 (Milan, 1993) Kluwer, 93-108.
Customer-Centered Systems. Morgan Kaufmann, San 21 Viller, S. and Sommerville, I., Coherence: an approach to
Francisco, CA, 1998. representing ethnographic analyses in systems design.
3 Blythin, S., Rouncefield, M. and Hughes, J.A., Never mind HumanÐComputer Interaction 14, Special issue on
the ethno stuffÑwhat does all this mean and what do we do representations in interactive systems development (1999)

Authorized licensed use limited to: Bahria University. Downloaded on June 09,2021 at 15:41:11 UTC from IEEE Xplore. Restrictions apply.