Design Process for Human-Data Interaction: Combining Guidelines with

Semio-participatory Techniques

Eliane Zambon Victorelli

, Julio Cesar dos Reis

, Antonio Alberto Souza Santos

and Denis Jos

e Schiozer

Institute of Computing, University of Campinas (UNICAMP), Campinas, Brazil

Center for Petroleum Studies, University of Campinas (UNICAMP), Campinas, Brazil

Keywords:

Human-Data Interaction, Design Approaches, Visual Analytics.

Abstract:

The complexity of analytically reasoning to extract and identify useful knowledge from large masses of data

requires that the design of visual analytics tools addresses challenges of facilitating human-data interaction

(HDI). Designing data visualisation based on guidelines is fast and low-cost, but does not favour the en-

gagement of people in the process. In this paper, we propose a design process to integrate design based on

guidelines with participatory design practices. We investigate, and when necessary, adapt existing practices

for each step of our design process. The process was evaluated on a design problem involving a visual an-

alytics tool supporting decisions related to the production strategy in oil reservoirs with the participation of

key stakeholders. The generated prototype was tested with adapted participatory evaluation practices. The

obtained results indicate participants’ satisfaction with the design practices used and detected the fulﬁlment of

users’ needs. The design process and the associated practices may serve as a basis for improving the HDI in

other contexts.

1 INTRODUCTION

The analysis of large amounts of data has become es-

sential for the success of organisations. This is a com-

plex task that often requires the judgement of domain

experts to make the best possible assessment of in-

complete and inconsistent information. The construc-

tion of tools for this purpose that facilitate human in-

teraction with the data is an important step in achiev-

ing the desired results.

Visual analysis (VA), the science of analytical

reasoning facilitated by interactive visual interfaces,

consists in interactive and iterative dialogue between

the human and the computer. The interactive analy-

sis process is a sequence of actions by the user and

responses by the computer motivated by analytical

questions (Thomas and Cook, 2005) (Turkay et al.,

2017). While several VA tools have been developed to

support exploration of large amounts of data, they do

not yet sufﬁciently support some complex exploratory

analysis scenarios. There is a shortage of support

environments where domain specialist and machine

work in harmonious interaction for data exploration

(Behrisch et al., 2018).

Several studies argue that designing visualisations

and VA tools based on guidelines is an important ap-

proach to help materializing the knowledge and expe-

rience gained by various experts in the ﬁeld (Shnei-

derman, 1996); (Scapin and Bastien, 1997); (Freitas

et al., 2002); (Amar and Stasko, 2004); (Zuk and

Carpendale, 2006); (de Oliveira, 2017). However, in

our view, this approach alone does not allow people

to participate in the process of building the system.

Recently, the Human-Data Interaction (HDI) area

has investigated how people interact with data in a

manner analogous to the Human-Computer Interac-

tion (HCI) area in relation to people and comput-

ers (Holzinger, 2014); (Knight and Anderson, 2016).

HDI studies human manipulation, analysis and mean-

ing creation from voluminous, unstructured and com-

plex datasets (Elmqvist, 2011).

HDI literature have often addressed the broad con-

text of ensuring the privacy and management of data

ecosystems (Bach, 2018); the means for people to be-

come aware of the data and the implications of their

use and manipulation (Mortier et al., 2014); the cap-

ture of personal preferences (Dimara et al., 2018); and

the design of mechanisms to support presentation, in-

terpretation and editing by laypersons (Crabtree and

Mortier, 2015). Our analysis of the HDI literature

410

Victorelli, E., Reis, J., Santos, A. and Schiozer, D.

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques.

DOI: 10.5220/0007744504100421

In Proceedings of the 21st International Conference on Enterprise Information Systems (ICEIS 2019), pages 410-421

ISBN: 978-989-758-372-8

has indicated that further studies are required to ob-

tain approaches to capture at project time the user’s

needs in relation to the interactions and the context

in which the data will be produced and used (Green

et al., 2015); Also, we need to achieve user engage-

ment with data interaction solutions through partici-

pation in the design process (Locoro, 2015).

In this context, questions are raised about design

approaches. Can the adequate involvement of stake-

holders in the VA design process be valuable to im-

prove HDI? Is it feasible to combine the advantages

of designing visualisations and VA tools combining

guidelines and participatory approaches? In our view,

the non-engagement of stakeholders in design can re-

sult in a system produced from a single view, which

although may be correct, does not describe the reality

in a comprehensive way with the vision of all parties

potentially affected by its construction.

We claim that the option for the VA interaction

design approach based on guidelines should not nec-

essarily exclude the possibility of taking advantage of

the participation of people with different proﬁles. Our

approach to HDI investigates ways to enable in-depth

interactions with data analysis tools, highlighting the

importance of taking into account stakeholders’ con-

tributions, the entire data life cycle, and the ability to

proactively engage with data in a variety of ways.

In this paper, we propose a design process for HDI

in decisions supported by VA that combines the ad-

vantages of using participatory practices and guide-

lines. We consider that the novelty of our study is the

combination of the two approaches. The successful

realisation of such processes involves a number of el-

ements that need to work in coordination: (i) identify-

ing key stakeholders and ensuring effective participa-

tion during design and choice of solutions; ii) search

of a set of guidelines to consider as a starting point;

iii) selection of relevant guidelines adequate for the

context; iv) ﬁnding the best way to unravel, explain

and facilitate understanding of the chosen guidelines

to the various participants; v) deﬁnition of practices

and adequate ﬂow for the conduction of design activi-

ties to ensure they harmoniously ﬂow and produce the

desired results.

While developing and reﬁning our approach for

design process, we were signiﬁcantly informed by a

series of practices that we have carried out in UNISIM

at Center for Petroleum Studies (CEPETRO) that de-

velops methodologies and tools to support an inte-

grated decision analysis in the development and man-

agement of petroleum ﬁelds (Schiozer et al., 2019).

In this study, the design situation involving a VA tool

that supports analytical needs for optimisation of oil

production strategy provided the opportunity to con-

duct the necessary design workshops for evaluating

our proposal.

The methodology used includes the elicitation and

problem clariﬁcation meetings based on organisa-

tional semiotics artefacts (Liu, 2008). Storyboard-

ing and braindrawing techniques supported the design

stage. The results of the design were materialised in

a functional prototype and evaluated using conceived

practices: participatory HDI design guidelines evalu-

ation and adapted Thinking-aloud. The practices of

design and evaluation were performed in an iterative

way and a questionnaire was answered by involved

participants.

The key contributions of this investigation in-

clude: 1) the deﬁnition of a design process tailored for

HDI design describing the conceived steps to allow

future reuse and replication of the dynamics; and 2)

the conduction of a case study in a data intensive en-

vironment related to oil reservoir management where

the process was applied.

2 BACKGROUND AND RELATED

WORK

Subsection 2.1 presents the approach to semio-

participatory design; Subsection 2.2 introduces de-

sign guidelines and evaluation techniques; and Sub-

section 2.3 reports on a summary of related work.

2.1 Semio-participatory Design

Organisational semiotics helps to get the understand-

ing of the context in which the technical system is in-

serted and the main forces that direct or indirectly act

on the situations (Liu, 2008). It studies the effective

use of information in business context and assumes

that organised behaviour is effected through the com-

munication and interpretation of signs by people, in-

dividually and in groups. Organisational semiotics in-

vestigates the organisation at different levels of for-

malisation - informal, formal, and technical.

The ﬁeld of participatory design spans a rich di-

versity of theories, practices, analyses and actions,

with the goal of working directly with users and other

stakeholders in the design of social systems that are

part of human work (Kuhn and Muller, 1993). This

approach considers that everyone involved in a design

situation is capable of contributing for it.

The semio-participatory approach to interactive

system design combines the concepts of organisa-

tional semiotics (Liu, 2008) and participatory design

(Kuhn and Muller, 1993). It includes shared knowl-

edge and mutual commitment to establish communi-

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques

411

cation during the design process (Baranauskas et al.,

2013). Various artefacts are proposed to mediate

this communication and facilitate creative and collab-

orative design engagement into semio-participatory

workshops. Artefacts and dynamics are proposed

to facilitate the interaction and communication of

a group with diversiﬁed proﬁle (Baranauskas et al.,

2013). The term semio-participatory techniques is

used to refer to the application of principles of par-

ticipatory design with the support of artefacts of or-

ganisational semiotics for conducting the workshops.

Semio-participatory workshops are the engine of

the model. It moves the design process through the

different levels of formalisation throughout the design

life cycle. In the beginning of the design process, the

conduction of speciﬁc participatory practices helps to

generate the artefacts called the stakeholder identiﬁ-

cation diagram (Liu, 2008) and evaluation framework.

Stakeholder identiﬁcation diagram is a layered

structure that facilitates the identiﬁcation of the in-

volved parties (stakeholders) in a process of new

technology conception and introduction. Evaluation

framework supports the articulation of problems and

the initial search for solutions. It informs about

speciﬁc issues from stakeholders and ideas or solu-

tions envisaged that have potential impact in the de-

sign. The evaluation framework extends the stake-

holder identiﬁcation diagram by considering for each

stakeholder issues and solutions to the problems. In

our proposal, semio-participatory artefacts is used to

guide workshops taking advantage of people’s partic-

ipation since conception and elicitation activities (cf.

Subsection 3.1).

2.2 Design Conception and Evaluation

Design specialists can compile recommendations ac-

quired in their experience in various projects, and pro-

vide designers with the ability to determine the con-

sequences of their design decisions. Design guide-

lines are recommendations a designer can follow to

enhance the interactive properties of the system (Dix

et al., 2004). One example of guideline is the in-

formation density guideline that suggests “to provide

only necessary and immediately usable data; do not

overload your views with irrelevant data” (Scapin and

Bastien, 1997). Design guidelines vary in their level

of abstraction, generality and authority.

In this article, we use the term guideline to talk

about design recommendation made by experts and

that can be used in the design of other systems in a

comprehensive way, without distinguishing the level

of generality, abstraction or authority. Guidelines are

used as an approach to bring specialists’ knowledge

to help the identiﬁcation of points for redesign that

favour HDI. Our work combines some speciﬁc rec-

ommendations collected from HDI literature to a set

of guidelines brought from related areas. Our set of

heuristics includes, for example, those of Nielsen that

refer to general guidelines applicable to all user inter-

faces (Nielsen, 1994).

The evaluation of systems that make use of visual

representations is an extremely complex task. Differ-

ent from a common user interface, a VA system must

be evaluated not only in terms of the interface, but

also for the information that it manages. There are

several approaches for this kind of evaluation. Some

of them are based on expert evaluation whereas oth-

ers involve ﬁnal users assessment. The VA evaluation

methodology can be subdivided: analytic evaluations

and empirical evaluations (Mazza, 2009).

The analytic type of evaluation is carried out by

experts who verify whether a certain system is com-

pliant with a series of heuristics or guidelines. Em-

pirical evaluation methods make use of functioning

prototypes of systems and involve the ﬁnal users. In

our participatory approach, it is important to count

on stakeholders contribution in the evaluation stage.

Therefore, in addition to analytical assessments based

on guidelines, we conduct empirical assessments.

One of the usability test techniques known as

Thinking-aloud (Lewis, 1982) consists in asking users

to think aloud about what they are doing while using

the system. The expectation is that the thoughts show

how user interprets each interface item.

Another set of evaluation techniques relies on ask-

ing the user about the interface. Query techniques can

be useful in eliciting detail about user’s view of a sys-

tem. They can be explored in evaluation with the ad-

vantage of getting users’ viewpoint directly and may

reveal issues that have not been considered by design-

ers. Interviews and questionnaires are the main types

of this technique (Dix et al., 2004).

The techniques mentioned here are not enough to

support our study. Therefore, we propose adaptations

to the evaluation practices to engage stakeholders at

the same time that we aggregate the knowledge and

experience acquired by experts through the use of

guidelines.

2.3 Related Work

The incontestable alternatives to our proposal are

purely participatory processes or purely based on

guidelines, as discussed in Section 1. This Section

improves the literature analysis by discussing similar

study alternatives and their limitations for the appli-

cation in our context.

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

412

Ceneda et al. (Ceneda et al., 2017) proposed a

process to favour interaction in the VA which seeks

solutions to support users during data analysis. Their

work concerns the user guidance aspects of a generic

visualisation environment to progress in their data

analysis. Their study does not focus in the interaction

design of VA tools.

The proposal for design data practices presented

by Churchill (Churchill, 2016) attempts to demystify

the “genius designer” whose instincts and intuition

lead to great design decisions. The work states the

need of taking a proactive and critical stance to de-

sign, develop, or evaluate products that incorporate

capture, storage and data analysis. It lists some prac-

tical things that can be done, but does not propose a

design process for tools that deal with data.

Leman et al. (Leman et al., 2013) studied typical

data visualisations that results from linear pipelines

that start by characterising data and end by displaying

the data. The proposal goal was to provide users with

natural means to adjust the displays to support good

HDI. This method supports a dynamic process for

deﬁning visualisations in which users learn from visu-

alisations and the visualisations adjust to the expert’s

judgement. This proposal differs from ours mainly

because it is a method for the execution time and not

a process for design.

Buchdid et al. (Buchdid et al., 2014) described an

approach for design combining participatory practices

with Interactive Digital Television patterns using the

principles of organisational semiotics. Their context

dealt with a few patterns. It was possible to explain

all the patterns in advance and train the participants

on Interactive Digital Television patterns before be-

ginning the design workshops. Our proposal distin-

guishes itself from this mostly because participatory

guidelines evaluation was carried after the ﬁrst pro-

totype generation as a way to incentive creative solu-

tions.

A review of Nielsen’s heuristic evaluation method

based on participatory concerns including users

(work-domain experts) as inspectors was proposed by

Muller et al. (Muller et al., 1998). They extended the

original Nielsen’s heuristic set with several process-

oriented heuristics. Their evaluation method can help

to guide iterative designs process. This technique is

similar to one of the practices proposed in our work,

but it is focused in a speciﬁc Nielsen’s heuristics set

extension and does not address the issue of HDI.

The online community was target of a study that

combined participatory methods for design and devel-

opment with heuristic evaluation applied iteratively

(Preece et al., 2004). First, a speciﬁc set of guidelines

was developed, extending the Nielsen’s heuristics and

adding a speciﬁc set of sociability heuristic. To re-

ﬁne the set of sociability guidelines, it was turned

into a questionnaire and open questions that was it-

eratively tested with online communities. The study

of the feedback allowed the elaboration of new items

for the test and the reﬁnement of the heuristics.

The studies conducted by Muller et al. (Muller

et al., 1998) and Preece et al. (Preece et al., 2004)

conceived practices of evaluation by guidelines com-

bined with participatory methods. However, these

studies did not involve VA and did not focus on HDI

nor propose a process. These facts highlight the inno-

vation of our proposal for design process which com-

bines participatory design and HDI guidelines.

3 INTEGRATING HDI

GUIDELINES AND

SEMIO-PARTICIPATORY

PRACTICES

This section presents our proposal for a new design

process for HDI that combines guidelines with semio-

participatory practices. Our proposed process in-

cludes several activities. Figure 1 presents the pro-

posed ﬂow that drives the activities.

3.1 Problem Clariﬁcation Activities

Initially, it is necessary to identify the stakeholders,

understand their concepts, terms and values in the

design problem (Liu, 2008). Stakeholder identiﬁca-

tion diagram is the supporting artefact for this stage

in semio-participatory workshops. It helps think-

ing beyond traditional participants and involving the

ones who may direct or indirectly inﬂuence or be in-

ﬂuenced by the solution under design (Baranauskas

et al., 2013).

From the overview obtained with stakeholder

identiﬁcation diagram, it is important to know the

problems and issues as well as the ideas and solutions

related to each stakeholder. They can have different

perspectives about the subject. The evaluation frame-

work is the artefact used to support this part of the

process (Baranauskas et al., 2013).

Several types of activities can be held to allow the

delimitation of scope for the design project. We pro-

pose to use the semio-participatory artefacts to under-

stand the subject during the elicitation activities. The

techniques can be presentations, group dynamics, in-

terviews, document analysis or others (cf. item A of

Figure 1).

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques

413

Figure 1: Process for design HDI combining guidelines with participatory design approaches.

In this context, Laddering interview, a special type

of interview approach that can be carried out with

users to understand the participants’ views and per-

sonal values is specially useful (Bourne and Jenkins,

2005). Discussion begins with concrete aspects and

evolves to more abstract level and the topic of conver-

sation evolves naturally.

3.2 Design Activities

In some situations the design involves a game of

forces, where some stakeholders may exert a dispro-

portionate inﬂuence on the project. In the participa-

tory design philosophy, this situation must be avoided

through mechanisms that seek to balance the forces.

Participants of all hierarchical levels should give their

contributions during participatory design workshops.

Techniques such as storyboarding and braindrawing

(cf. item B of Figure 1) facilitate the engagement.

Storyboarding. An interaction scenario must be

properly deﬁned and tasks performed in the scope

of the activity should be well delineated. This helps

in the understanding of the scenario by all those in-

volved in the design. The functionality under design

and the various steps involved in it need to be well

described. Sometimes a verbal or written description

is enough. However, if the problem involves partic-

ipants with different proﬁles, it may be necessary to

adopt visual techniques to promote a complete under-

standing. In our approach, stakeholders might con-

tribute in the storyboarding creation.

A description based on storyboarding can help in

the pre-visualisation of a digital interface that might

support some task. It uses a sequence of stages where

each stage shows a moment in time, e.g., an interface

state. It visually tells a story. It is useful for partici-

pants to interpret how each interface proposal ﬁts into

the design problem. After reaching a consensus about

the ﬂow, the solution supporting each interface state

can be designed.

Braindrawing. It is a technique used to materialise

ideas and proposals into interface low-ﬁdelity proto-

types for digital applications. In our process, each

interface state identiﬁed in the storyboard is the target

of a braindrawing section.

In this step, all participants are arranged in a cir-

cle. Each participant draws its initial idea for the in-

terface that would support the deﬁned task using a

sheet of paper and pen. After a predetermined short

period of time, the drawing is passed to the next par-

ticipant who completes it. This step is repeated un-

til each participant received his/her original drawing

back. Then, each participant presents his proposal to

the others. The group must agree on a candidate solu-

tion to be adopted. The solution may involve common

elements of several proposals.

3.3 Pre-evaluation Activities

In our proposal, the group creates during design ac-

tivities a low ﬁdelity prototype without design guide-

lines orientation. The guidelines are introduced in the

evaluation phase. The evaluation of the prototype re-

quires some preparation activities (cf. item C of Fig-

ure 1):

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

414

Construction of a Non Functional Navigable Pro-

totype. It is important that the prototype be navigable

to facilitate the evaluation of HDI. Based on the low

ﬁdelity prototypes and the accorded storyboarding, a

designer can propose a navigable prototype.

Selection of HDI Guidelines and Elaboration of

Explanation for HDI Guidelines. Designers should

deﬁne which sets of guidelines are relevant to be

used. Given the nature and complexity of the prob-

lem of facilitating the HDI for VA application, we

consider it is important to aggregate guidelines of

the VA, HCI and HDI area. We draw upon previ-

ous studies (de Oliveira, 2017) with a compiled set of

VA and HCI guidelines and heuristics (Shneiderman,

1996), (Scapin and Bastien, 1997), (Freitas et al.,

2002), (Amar and Stasko, 2004), (Zuk and Carpen-

dale, 2006). In this work, we call the set that re-

sulted from this junction as the initial set of HDI de-

sign guidelines. If there are any set of guidelines for

the application domain or standards used by the target

organisation, they should also be included.

Afterwards, the designers should select which

guidelines from the initial set are related to the scope

of the prototype and that matters in the context. One

way to facilitate the selection is grouping the guide-

lines by subject in clusters. One set could aggregate,

e.g.,, all guidelines related to the amount of informa-

tion showed to the user. It is then necessary to ﬁnd

ways to unravel, explain and facilitate understanding

of those chosen guidelines to allow participants com-

prehending them and to decide on their use. Using

examples, even if they are from different contexts, are

a good way to explain the guidelines.

3.4 Evaluation Activities

The navigable prototype and the selected guidelines

support the evaluation activities and help decisions

to improve and reﬁne the prototype. We propose a

participatory approach for the activities of evaluation

phase. All identiﬁed points in the evaluation phase

needs to be organised in a priority order to be treated

in a redesign cycle.

We need to verify to which extent users think the

prototype might help them accomplishing their tasks;

and how the prototype can be improved. For support

this stage, we propose two new participatory evalua-

tion practices: Participatory Thinking-aloud Evalua-

tion and Participatory HDI Design Guidelines Evalu-

ation. These new practices should be associated to an

User Evaluation (cf. item D of Figure 1).

Participatory Thinking-aloud Evaluation. Our

proposed dynamics was adapted from the original

Thinking-Aloud method deﬁned in (Lewis, 1982) and

the adaptation presented in (Buchdid et al., 2014). A

participant is invited to interact with the navigable

prototype to complete an use case, conducting a pre-

deﬁned task, previously explained to the group. In

our proposal, during the interaction, the whole group

is stimulated to report their thoughts and impressions

about the prototype. All participants in the workshop

speak aloud while one participant interacts with the

prototype. This activity allows knowing the impres-

sion that a prototype caused.

Participatory HDI Design Guidelines Evaluation.

We claim that the participants should, in the begin-

ning of the design process, propose alternatives freely

before worrying about guidelines. In this sense, they

do not stop thinking in creative ways to solve the

problem and are not skewed in attending a guideline.

Thus, in each round of the process the guidelines us-

age starts after the elaboration and consensus of the

design activities.

This activity should be leaded by a designer. For

each guideline, participants are introduced to the rec-

ommendation of design with an explanation and ex-

amples of applications, previously prepared.

After the explanation, participants should discuss

the application, the impacts as well the advantages

and disadvantages of adoption. They need to decide if

the guideline are going to be adopted. Finally, a sub-

set of the discussed guidelines potentially useful and

the associated ideas for redesign can be generated by

this activity.

We recommend to ﬁrst carry out the Thinking-

aloud and then the HDI design guidelines evaluation

so participants have in mind the status of the proto-

type to relate it with the guidelines. Based on the is-

sues found during the participatory evaluation activi-

ties, participants elaborate a list of problems and sug-

gestions for improving the application in a redesign

phase.

At this point, the team has the opportunity to de-

cide if they are going to make a redesign activity to

adjust the prototype to issues identiﬁed. If the group

decision is to change the prototype, it may be neces-

sary to return to the storyboarding and braindrawing

activities. This cycle can be repeated more than once

until participants feel that the prototype design is ap-

propriate for their needs.

User Evaluation. When the design of the prototype is

mature, an user evaluation based in the Query Tech-

nique (Dix et al., 2004) with all participants is con-

ducted asking the participants about the results di-

rectly. It is applied by interview or questionnaire.

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques

415

4 CASE STUDY

One challenge being addressed by UNISIM consists

in the investigation of technologies for determinis-

tic optimal production strategy selection in oil ﬁelds

(Schiozer et al., 2019). This process involves a lot of

efforts in analysis of voluminous data. SEPIA is a VA

software tool developed to facilitate this process.

One step for optimisation of the production strat-

egy requires the performance of many simulations,

with some variations among them. After some simu-

lations, it is necessary to make comparisons to verify

the differences that resulted from the changes made

from one simulation to another. SEPIA does not have

speciﬁc functionalities to support this scenario. In this

case study, we address how to allow the evolution of

SEPIA with HDI design for comparisons among dif-

ferent oil production strategies. Our proposed design

process was applied to this scenario.

The activities of this study were conducted from

June to December, 2018, and involved 2 Computer

Science researchers and 6-8 participants playing dif-

ferent roles at UNISIM One of the Computer Sci-

ence researcher played the role of designer through-

out the process. The application required 7 meetings

and workshops of 3 hours each on average. Thus,

the whole process was conducted in approximately 21

hours of meeting with 6 participants on average. In

addition, a similar effort was made to prepare the pre-

sentations and practices for each meeting. We present

the results for each step in the following.

4.1 Results of Design Problem

Clariﬁcation Activities

The process began with stakeholders identiﬁcation

followed by issues and requirements elicitation.

Stakeholders, Values and Interest. There are many

stakeholders potentially involved with SEPIA tool

and they were detected by the stakeholder identiﬁca-

tion diagram elaborated in the participatory practices.

The initial group of stakeholders was distributed in

the stakeholder identiﬁcation diagram inner layers.

For example, “Developers” was inserted in the op-

eration layer of stakeholder identiﬁcation diagram;

“Designers” was inserted between the operation and

contribution layers. UNISIM and the development

project manager ﬁgured in the contribution layer be-

cause they are responsible for the production of the

tool. In the source layer are the engineers and re-

searchers which are very important stakeholders be-

cause they are real users of SEPIA tool.

The presentation of the initial stakeholder diagram

helped the detection of others interested indirectly

connected with SEPIA. For example, in market layer

there is CMG

, a reservoir simulation software devel-

oper, and other tools vendors. In the community layer

we inserted the Brazilian regulation agency for the ac-

tivities that integrate the oil and natural gas and bio-

fuel industries in Brazil, the ANP

After this activity, the end users with the engineer-

ing and research proﬁle began to participate in the

workshops. They participate very actively during the

clariﬁcation and design workshops.

Issues and Requirements. In the elicitation phase,

there were presentations related to underlying domain

concepts, so the designers and developers could begin

understanding the complex domain of strategies for

petroleum exploration. To deepen this understanding,

we conducted 4 individual interviews with an aver-

age duration of one and a half hour based in laddering

techniques with the SEPIA users.

The users were encouraged to speak about their

daily activities, their issues and problems and solu-

tions they see to the problems. The results of the

interviews revealed that one issue was the execution

of several attempts to optimise the strategies and the

comparison of the results. However, time, volume of

data and number of ﬁles involved in these attempts

turn difﬁcult the comparisons among obtained results

from simulations. One user, e.g., made several sim-

ulations by slightly varying the position of a speciﬁc

well that was part of a particular production strategy.

She used bar chart views to compare strategy results

with the various well positions in relation to return of

investment and oil volume produced. The support for

the comparison between strategies was chosen as the

central requirement to be addressed in the ﬁrst ver-

sion of the prototype and to validate the execution of

the process proposed in this article.

4.2 Results of Design Activities

The issues, problems, ideas and solutions identiﬁed

during design problem clariﬁcation activities were the

source for the design.

The Storyboarding. The ﬁrst participatory design

activity was conducted in a meeting with the goal of

consolidating the needs reported by users in the inter-

views. They needed to agree on the scope for the next

activities.

A Storyboarding was conceived as a state transi-

tion diagram and illustrated with visual interface pro-

totypes to support the discussion of the execution ﬂow

https://www.cmgl.ca/software

http://www.anp.gov.br

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

416

of two actual use cases of comparisons. The ﬂow pre-

sented by the storyboard had to be changed as the de-

sign activities were progressing. The technique was

very useful during all the process to build an agree-

ment about the execution ﬂow under discussion. Fig-

ure 2 shows the ﬂow in the storyboard.

Figure 2: Storyboard showing the execution ﬂow agreed in

participatory design activity.

Braindrawing. After getting consensus for the ﬁrst

version of the storyboard, it was possible to iden-

tify the goal of each interface involved in the pro-

cess. Braindrawing sections were guided by the ﬂow

and states deﬁned in the Storyboarding practice. We

conducted one Braindrawing section for each relevant

state identiﬁed in the storyboard.

In the Braindrawing sections, each participant had

one minute to draw the screen for achieve the desired

goal following the steps explained in Subsection 3.2.

On average, the Braindrawings were elaborated by six

participants and the drawing phase took no more than

30 minutes. This procedure generated several alter-

natives ideas for each interface. Each idea was pre-

sented by who initially proposed the design solution.

After the team discussed the ideas, a consolidated in-

terface screen was deﬁned. It took another 30 to 60

minutes. In general, the consolidated drawing joined

ideas from several participants.

4.3 Results of Pre-evaluation Activities

After the initial design and before the evaluation

phase, preparation activities were necessary to facil-

itate the work of the entire group. These activities

were performed by the designer.

Prototype Construction. The Braindrawing work-

shops results, low-ﬁdelity paper prototypes, were

transformed into navigable medium-ﬁdelity proto-

types. A navigable prototype was very useful during

the Thinking-aloud activities and helped to raise a lot

of issues, questions and suggestions.

The Guidelines Selection and Explanation. The

designer was responsible for selecting the guidelines

that were most related with the prototype scope and

for explaining the guidelines to the group. The previ-

ously compiled set on which we based was useful be-

cause it ranked the recomendations found in inﬂuen-

tial contributions in the VA and HCI areas by creating

clusters of guidelines (de Oliveira, 2017). The clus-

ters grouped the guidelines by subject and facilitated

the selection of the subset that matter to our scope.

We organised an additional cluster with HDI

guidelines found in literature. Examples of guidelines

found in HDI and used in this study are: i) “Consider

all stages of the data life-cycle and the relevant stake-

holders” (Hornung et al., 2015); ii) “Consider values

in the design and implementation of analytics con-

sidering how technologies can materialise values, and

their interpretive ﬂexibility.” (Knight and Anderson,

2016); and iii) “Provide visual or textual indicators as-

pects of search like relevance, usability and quality on

the interface, backed up by automatically computed

metrics or user-generated reviews and annotations.”

(Koesten et al., 2017).

There are not any known set of guidelines for visu-

alisations in the oil production domain, to our knowl-

edge. In UNISIM there is a colour standard to dif-

ferentiate the representation of activities according to

the stage of the methodology in which it is executed.

One colour is used for activities related to reservoir

modelling and construction, another for reduction of

scenarios (data assimilation) a third colour represents

long-term production optimization (prediction) and a

four colour represents sort-term production optimiza-

tion activities for the future. Then, this speciﬁc guide-

lines was included in the set.

4.4 Results Evaluation Activities

The Participatory HDI Design Guidelines Evalu-

ation. This practice, facilitated by one designer, in-

volved all design workshop participants. The selected

guidelines were explained to the group so they can

decide whether changes suggested by the guidelines

could beneﬁt the prototype. In general, each selected

guideline was explained and discussed at a time.

During the consolidation of the proposal for the

comparison visualisation interface (cf. item 4 of Fig-

ure 2), the participants liked an idea different from the

approach currently used in SEPIA. The participants

liked the new approach, but they were uncomfortable

with the paradigm shift at the ﬁrst moment. It was

considered that the evaluation based on HDI guide-

lines could help to support the choice. In particular,

the following guidelines were discussed:

• “Shneiderman mantra”: overview ﬁrst, zoom and

ﬁlter, than details on demand.

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques

417

• “Information density” (Scapin and Bastien,

1997): provide only immediately usable data; do

not overload views with irrelevant data.

• “Filter the uninteresting” (Shneiderman, 1996):

allow users to control display content and quickly

focus on interests by eliminating unwanted items.

• Cycle of data (Hornung et al., 2015): “Consider

all stages of the data lifecycle”.

Currently in SEPIA, in situations similar to those

under design, all graphs of all the wells would be

generated and displayed at the same time giving an

overview of the data. However, it may overload the

user view with irrelevant data and make it difﬁcult the

selection of view with data of the desired well. Con-

sidering the new approach designed, only one well

graphic would be shown in each frame with a wid-

get control to quickly focus on the interesting well

and eliminating others wells graphics. This approach

seems to reduce the information density, but elimi-

nates the overview of data that was familiar to users.

With the presentation of the guidelines, it was con-

sidered that the overview guaranteed by the more tra-

ditional approach did not bring relevant information

to the analysis at this moment. Considering all the

data life-cycle, the overview was necessary only at the

beginning of the process. In addition, users would not

like to have to undertake complex activities when the

complexity is not required by the task at hand. Sup-

ported by the analyses of the guidelines in the context

of the interface under discussion, participants were

able to approve the innovative design option more

comfortably.

Participatory Thinking-aloud Evaluation Activi-

ties. The Thinking-aloud process was undertake in

groups of ﬁve to six people. In the ﬁrst section, only

the designer handled the prototype, because other par-

ticipants did not want to do it. In the second time, the

prototype was more familiar to them, and one of the

participants volunteered to navigate in the prototype.

All the participants deeply contributed, which

made the sections long. They lasted one and a half

hour on average. They were very productive to reﬁne

the ideas for the redesign of the prototype.

Sometimes the expressed thoughts led to modiﬁ-

cations with low impact like changing the location of

an interface component, e.g., the action buttons posi-

tions. But, it was also identiﬁed an opportunity of im-

provement concerning the interaction approach that

would demand a high impact change in the ﬂow of

execution.

User Evaluation Activities Results. In the ﬁnal stage

of the process, the participants were invited to answer

a questionnaire and one open question evaluating the

resulting prototype from previous activities.

The evaluation used a Likert scale from 0 to 3 in

which the respondents speciﬁed their level of agree-

ment with the adequacy of the generated prototype.

Notes 2 and 3 were considered positive. Five people

answered the questionnaire (all of them had partici-

pated in the design and evaluation activities).

The questions attempted to identify users’ assess-

ment of the prototype in general, and speciﬁcally re-

garding the adequacy of interfaces design, interac-

tion ﬂow and meeting general users’ needs. Table 1

presents results for each question, where Aver. Grade

is the average grade for all respondents, %Positive

represents the percentage of grades greater than or

equal to 2 and %Max. Grade is the percentage of

grades equals to 3. The overall evaluation was con-

sidered positive for 100.0% of the participants and

66.7% of them rated at the maximum grade. The ade-

quacy of the execution ﬂow was the only item that did

not get 100% of positive opinion. The adequacy of

consolidated design to meet the stakeholders’ needs

was rated with the highest grades.

4.5 Participants’ Assessment of the

Process

After the design and evaluation activities, all par-

ticipants were invited to evaluate the process used

through a questionnaire and open question. The scale

and the criteria to consider a positive opinion was the

same used for the prototype evaluation. Five partici-

pants answered the questionnaire.

Table 2 presents the obtained results. The process

overall evaluation was considered positive for 100.0%

of the participants and 33.3% rated them at maximum

grade. For the questions about speciﬁc issues regard-

ing the various activities of the process, the item with

the worst grade was about the time involved, having

only 40% of positive opinion and 0% of maximum

grade. The best aspect considered by the participants

was the adequacy of practices used for the objectives

with 100% of positive opinion and maximum grade.

5 DISCUSSION

We proposed to combine the advantages of using

guidelines and participatory practices in a design pro-

cess for HDI in decisions supported by the VA. The

participatory evaluation practices helped us to under-

stand if the generated prototype made sense for stake-

holders. We veriﬁed to what extent users think the

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

418

Table 1: Prototype Evaluation.

Question Aver.

Grade

%Positive % Max.

Grade

What is your overall evaluation of the results generated by the

workshops

2.5 100.0 66.7

Is the consolidated design suitable for the proposed screens? 2.5 100.0 66.7

Is the ﬂow of the proposed interaction suitable for the compari-

son functionality?

2.5 75.0 66.7

Does the comparison functionality as prototyped suit your needs

as a user?

2.63 100.0 66.7

Does the comparison functionality as prototyped suit meet the

needs of the majority of users?

2.63 100.0 66.7

Table 2: Process Evaluation.

Question Aver.

Grade

%Positive % Max.

Grade

What is your overall evaluation of workshops you attended? 2.25 100.0 33.3

What is your opinion about the time involved in the workshops? 1.4 40.0 0.0

Did you feel comfortable expressing your opinions? 2.8 100.0 75.0

Were the practices used in the workshops adequate for the ob-

jectives?

3.0 100.0 100.0

Did the activities allow the reconciliation of different points of

view?

2.4 100.0 25.0

Did the meetings allow the creation of shared understanding of

the problem addressed?

2.2 80.0 25.0

Did braindraw’s activity help bring about new design solutions? 2.6 100.0 75.0

Was the navigable prototype useful for understanding the solu-

tions being discussed?

2.8 100.0 75.0

Was Thinking-aloud useful for prototype evaluation? 2.8 100.0 75.0

Was the presentation of design guidelines sufﬁcient to under-

stand the recommendation?

2.0 80.0 25.0

prototype might help them and in what points the pro-

totype can be improved. The evaluation practices also

sought to understand relevant aspects about the suit-

ability of the process.

Our achieved results indicate that both the pro-

posed process and the product generated by it, the

prototype, had good acceptance. However, some

points of improvement were evidenced by users feed-

backs. Two main issues found were the duration of

the meetings and the guidelines understanding.

One of the challenges to ensuring the effective

participation of key stakeholders is the time and effort

required to participate during the process as a whole.

In our study case, many meetings were necessary for

understanding, delimiting the scope, design and ap-

prove the solution. The process proved to be some-

what onerous in relation to the number of meetings

and their duration. The time spent in meetings and

workshops, as detailed in Section 4, was perceived by

the participants as taking much time.

It was not feasible to ensure the participation of all

experts over a very long period of time. The strategy

used to minimise the effort required was to alternate

meetings with and without ﬁnal users, or at a high

level with focus on the domain understanding and low

level with focus on practical aspects of the tool. Af-

ter the meetings with speciﬁc focus, we gathered all

stakeholders to hold a hands-on workshop.

The time involved in the activities should be con-

sidered to enable further applicability of the process.

Some participants suggested ways to speed up the

meeting with stricter control of the meeting agenda.

However, care must be taken that stricter control does

not inhibit the participation and creativity of all.

Regarding the moment of the guidelines applica-

tion, both the use of guidelines at design time and

at the time of evaluation have their advantages and

challenges. In the scenario of this investigation, if the

guidelines were used in the beginning of the design,

it would be difﬁcult and time consuming to train the

participants in all the guidelines. On the other hand,

the guidelines-based evaluation conducted by design

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques

419

experts would require a lot of effort to train them in

the application domain due to the complexity of the

subject. Teaching the design guidelines to the par-

ticipants was challenging, but the training of design

specialists on the subjects of oil production strategies

domain would be even harder.

Therefore, it would not be possible to include

several design specialists in the evaluation process.

We involved only two design specialists and priori-

tised the participation of users in the evaluation phase.

More speciﬁcally, we found that conducting a partici-

patory evaluation based on guidelines after the design

phase allowed the reduction of the number of relevant

guidelines that needed to be addressed in the training.

It proved to be advantageous in terms of training ef-

fort and the process was viable with good results.

This investigation demonstrated that the adequate

involvement of stakeholders in the VA design pro-

cess is valuable to improve HDI. In addition, our ﬁnd-

ings revealed the feasibility of combining the advan-

tages of designing visualisations and VA tools based

on guidelines and participatory approaches.

As future steps, we plan to identify well-

delineated scopes that can be easily explained to de-

sign specialists and conducting a guideline evaluation

involving several design specialists. In addition, we

seek to investigate other ways to facilitate the under-

standing of the guidelines. We intend to combine ex-

amples from other contexts and examples adopted in

other visualisation tools for the same domain. After

discussing these examples, we can devise activities to

practice exercises with the use of guidelines in fake

problems.

The ease or difﬁculty to obtain consensus in the

decisions is an interesting point to consider in a par-

ticipatory process. In a context in which a single par-

ticipant suggested altering one proposal made individ-

ually by another, consensus probably would be not so

easy. But in this process, both the initial ideas and

the proposals for change in the prototype came from

several different participants and the ideas were dis-

cussed as soon as they arose. The acceptance of the

changes was facilitated and solutions were adopted in

consensus by the whole group.

Finally, during the participatory practices, the

constructive nature of the process allowed to observe

how shared understanding about the problem domain

was obtained, different viewpoints were conciliated,

different proposals were consolidated, and the appli-

cation was created. The discussions, , were very fruit-

ful and led to the materialisation of the proposed so-

lutions.

So we found that, although sometimes costly in

the design time, it is achievable to combine the advan-

tages of designing visualisations and VA tools based

on guidelines and of the participatory approaches. In

addition, stakeholder involvement in the VA design

process can help improve HDI. Advances in HDI de-

sign reduce the likelihood of rework after solution de-

velopment, thus offsetting the additional efforts made

with participatory practices.

6 CONCLUSION

The design of HDI-oriented software applications in-

volving the visualisation of huge volumes of data to

guide sensitive decisions is still a open research prob-

lem. This paper proposed a design process to improve

HDI in VA that combines the advantages of both data

visualisation guidelines and participatory practices.

We applied it in a case study for VA tool used in the

decision on the oil production strategy. Our obtained

results indicated users positive evaluation of the pro-

totype generated by the process. Participants demon-

strated satisfaction with the practices used and com-

fort to express their ideas. Our research ﬁndings re-

vealed that the combination of design approaches may

allow better decisions to be made through VA of volu-

minous data. Our proposal presents good potential for

applications in the design of VA solutions involving

HDI in domains of high complexity. In future work,

we plan to address open challenges involving the ex-

tension of the process to allow reﬁning the guidelines

set from users’ answers to open questions. We plan to

investigate how to measure the HDI improvement in

addition to test the process in other HDI domains.

ACKNOWLEDGMENTS

We thank the support from Petrobras and Energi Sim-

ulation inside of the R&D of ANP. We appreciate the

involvement of the Research Group of Simulations

and Management of Petroleum Reservoirs (UNISIM-

UNICAMP) at CEPETRO and Institute of Computing

at UNICAMP.

REFERENCES

Amar, R. and Stasko, J. (2004). A Knowledge Task-Based

Framework for Design and Evaluation of Information

Visualizations. IEEE Symposium on Information Vi-

sualization, pages 143–149.

Bach, B. (2018). Ceci n’est pas la data: Towards a Notion of

Interaction Literacy for Data Visualization. In VisBIA

2018 - Workshop on Visual Interfaces for Big Data

Environments, pages 10–12.

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

420

Baranauskas, M. C. C., Martins, M. C., and Valente, J. A.

(2013). Codesign Parte I. In Codesign de Redes Dig-

itais: Tecnologia e Educac¸

ao a Servic¸o da Inclus

Social, pages 17410–17416. Penso Editora.

Behrisch, M., Streeb, D., Stoffel, F., Seebacher, D., Mate-

jek, B., Weber, S. H., Mittelstaedt, S., Pﬁster, H.,

and Keim, D. (2018). Commercial Visual Analytics

Systems-Advances in the Big Data Analytics Field.

IEEE Transactions on Visualization and Computer

Graphics, pages 1–20.

Bourne, H. and Jenkins, M. (2005). Eliciting Managers ’

Personal Values : An Adaptation of the Laddering In-

terview Method. Organizational Research Methods,

8(4):410–428.

Buchdid, S. B., Pereira, R., and Baranauskas, M. C. C.

(2014). Playing Cards and Drawing with Patterns

- Situated and Participatory Practices for Designing

iDTV Applications. Proceedings of the 16th Interna-

tional Conference on Enterprise Information Systems.

Ceneda, D., Gschwandtner, T., May, T., Miksch, S., Schulz,

H. J., Streit, M., and Tominski, C. (2017). Charac-

terizing Guidance in Visual Analytics. IEEE Trans-

actions on Visualization and Computer Graphics,

23(1):111–120.

Churchill, E. F. (2016). Designing data practices. Interac-

tions, 23(5):20–21.

Crabtree, A. and Mortier, R. (2015). Human data in-

teraction: historical lessons from social studies and

cscw. In ECSCW 2015: Proceedings of the 14th Eu-

ropean Conference on Computer Supported Cooper-

ative Work, 19-23 September 2015, Oslo, Norway,

pages 3–21. Springer.

de Oliveira, M. R. (2017). Adaptac¸

ao da Avaliac¸

Heur

ıstica para Uso em Visualizac¸

ao de Informac¸

ao.

Master thesis, Universidade Estadual de Campinas.

Dimara, E., Bezerianos, A., and Dragicevic, P. (2018).

Conceptual and Methodological Issues in Evaluating

Multidimensional Visualizations for Decision Sup-

port. IEEE Transactions on Visualization and Com-

puter Graphics, 24(1):749–759.

Dix, A., Finlay, J., Abowd, G. D., and Beale, R. (2004).

Human-Computer Interaction. Pearson Prentice-Hall,

third edit edition.

Elmqvist, N. (2011). Embodied Human-Data Interaction.

ACM CHI 2011 Workshop “Embodied Interaction:

Theory and Practice in HCI, pages 104–107.

Freitas, C. M. D. S., Luzzardi, P. R. G., Cava, R. A.,

Winckler, M. A. A., Pimenta, M. S., and Nedel, L. P.

(2002). Evaluating Usability of Information Visual-

ization Techniques. Proceedings of 5th Symposium

on Human Factors in Computer Systems (IHC 2002),

pages 40–51.

Green, M. H., Davies, P., and Ng, I. C. L. (2015). Two

strands of servitization: A thematic analysis of tradi-

tional and customer co-created servitization and future

research directions. International Journal of Produc-

tion Economics.

Holzinger, A. (2014). Extravaganza tutorial on hot ideas

for interactive knowledge discovery and data mining

in biomedical informatics. Lecture Notes in Computer

Science, 8609 LNAI:502–515.

Hornung, H., Pereira, R., Baranauskas, M. C. C., and Liu,

K. (2015). Challenges for Human-Data Interaction –

A Semiotic Perspective. Springer International Pub-

lishing Switzerland, 9169:37–48.

Knight, S. and Anderson, T. D. (2016). Action-oriented,

accountable, and inter(active) learning analytics for

learners. CEUR Workshop Proceedings, 1596:47–51.

Koesten, L. M., Kacprzak, E., Tennison, J. F. A., and Sim-

perl, E. (2017). The Trials and Tribulations of Work-

ing with Structured Data. Proceedings of the 2017

CHI Conference on Human Factors in Computing Sys-

tems - CHI ’17, pages 1277–1289.

Kuhn, S. and Muller, M. (1993). Participatory Design.

Communications of the ACM.

Leman, S. C., House, L., Maiti, D., Endert, A., and North,

C. (2013). Visual to Parametric Interaction (V2PI).

PLoS ONE, 8(3).

Lewis, C. (1982). Using the ’thinking-aloud’ method in

cognitive interface design. Research Report RC9265,

IBM TJ Watson Research Center.

Liu, K. (2008). Semiotics In Information Systems Engineer-

ing. Cambridge University Press.

Locoro, A. (2015). A map is worth a thousand data:

Requirements in tertiary human-data interaction to

foster participation. CEUR Workshop Proceedings,

1641:39–44.

Mazza, R. (2009). Introduction to Information Visualiza-

tion. Springer Verlag.

Mortier, R., Haddadi, H., Henderson, T., McAuley, D., and

Crowcroft, J. (2014). Human-Data Interaction: The

Human Face of the Data-Driven Society. SSRN Elec-

tronic Journal.

Muller, M. J., Matheson, L., Page, C., and Gallup, R.

(1998). Participatory heuristic evaluation. Interac-

tions, 5(5):13–18.

Nielsen, J. (1994). Usability inspection methods. In Confer-

ence companion on Human factors in computing sys-

tems, pages 413–414. ACM.

Preece, J., Abras, C., and Krichmar, D. M. (2004). De-

signing and evaluating online communities: research

speaks to emerging practice. International Journal of

Web Based Communities, 1(1):2.

Scapin, D. L. and Bastien, J. M. (1997). Ergonomic cri-

teria for evaluating the ergonomic quality of interac-

tive systems. Behaviour and Information Technology,

16(4-5):220–231.

Schiozer, D. J., Santos, A. A. S., Santos, S., and Hohen-

dorff Filho, J. (2019). Model-Based Decision Anal-

ysis Applied to Petroleum Field Development and

Management. Oil & Gas Science and Technology –

Rev. IFP.

Shneiderman, B. (1996). “The Eyes Have It: A Task by. In

Proc., IEEE Symposium on Visual Languages.

Thomas, J. and Cook, K. (2005). Illuminating the path: The

research and development agenda for visual analytics.

IEEE Computer Society, page 184.

Turkay, C., Kaya, E., Balcisoy, S., and Hauser, H. (2017).

Designing Progressive and Interactive Analytics Pro-

cesses for High-Dimensional Data Analysis. IEEE

Transactions on Visualization and Computer Graph-

ics, 23(1):131–140.

Zuk, T. and Carpendale, S. (2006). Theoretical analysis of

uncertainty visualizations. In Proceedings of SPIE -

The International Society for Optical Engineering.

Design Process for Human-Data Interaction: Combining Guidelines with Semio-participatory Techniques

421