A Conceptual Method for Eliciting Trust-related Software Features
for Computer-mediated Introduction
Angela Borchert, Nicolás Emilio Diáz Ferreyra and Maritta Heisel
Department of Software Engineering, University of Duisburg-Essen, Duisburg, Germany
Keywords: Trustworthiness, Computer-Mediated Introduction, Requirements Elicitation.
Abstract: Computer-Mediated Introduction (CMI) describes the process in which individuals with compatible intentions
get to know each other through social media platforms to eventually meet afterwards in the physical world
(i.e. sharing economy and online dating). This process involves risks such as data misuse, self-esteem damage,
fraud or violence. Therefore, it is important to assess the trustworthiness of other users before interacting with
or meeting them. In order to support users in that process and, thereby, reducing risks associated with CMI
use, previous work has come up with the approach to develop CMI platforms, which consider users’ trust
concerns regarding other users by software features addressing those. In line with that approach, we have
developed a conceptual method for requirements engineers to systematically elicit trust-related software
features for a safer, user-centred CMI. The method not only considers trust concerns, but also workarounds,
trustworthiness facets and trustworthiness goals to derive requirements as a basis for appropriate trust-related
software features. In this way, the method facilitates the development of application-specific software, which
we illustratively show in an example for the online dating app Plenty of Fish.
1 INTRODUCTION
Social media offers many possibilities as a media
channel due to its large number of users who want to
be connected with other people. Services like the ones
using Computer-Mediated Introduction (CMI)
support this wish for connectivity. CMI offers users a
realm in which they can get to know and connect with
unfamiliar individuals with compatible interests to
potentially have offline encounters (Obada-Obieh
and Somayaji, 2017). Compatible interests may
involve human qualities or resources users possess to
satisfy mutual needs. Examples for CMI are sharing
economy and online dating. While sharing economy
is based on monetary exchange to enable services like
private lodging, car drives or dog sitting between
users, online dating focuses on social exchange.
Compared to other kinds of social media, CMI can
further be characterized into the property that it has
different stages concerning the interaction with users
of interest. These stages are before, during and after
CMI users are connected (Obada-Obieh and
Somayaji, 2017). The before stage includes the search
for an appropriate other user who fits a user’s needs.
The during stage denotes the establishment of
contact, the online interaction as well as the offline
encounter so that the during stage can be further
subclassified in these steps. The after stage describes
the disconnection of both users on the online
platform.
Though the merit of CMI is that users get to know
new people, this also bears risks such as fraud,
damaged self-esteem or violence (Obada-Obieh et al.,
2017). Cues that are usually available in face-to-face
interactions and are important to get an impression of
an other individual are partly missing, different than
in the offline context or easy to manipulate for giving
altered impressions (Walther et al., 2005). This
complicates the trustworthiness assessment of users
of interest in CMI. However, it can be assumed that
the trustworthiness assessment is a decisive factor for
the decision-making process whether to interact with
or meet another person (Rotter, 1980). Especially in
the context of offline encounters based on online
introductions, users have stated concerns about safety
(Couch et al., 2012).
In a previous work, we thus identified the need
that CMI applications should better assist users in
assessing the trustworthiness of other end-users
(Borchert et al., 2020). Since a CMI system i)
modulates the perception users have about each other,
ii) mediates their interaction and iii) may trigger
Borchert, A., Ferreyra, N. and Heisel, M.
A Conceptual Method for Eliciting Trust-related Software Features for Computer-mediated Introduction.
DOI: 10.5220/0009328102690280
In Proceedings of the 15th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE 2020), pages 269-280
ISBN: 978-989-758-421-3
Copyright
c
2020 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
269
offline encounters, it can impact peoples’ well-being
to a large extent. Seeing this as a responsibility the
system should take, this work introduces a
requirements engineering method for eliciting
software features that shall support users in their
trustworthiness assessment. The objective of the
method is to build CMI systems whose previously
described risks are reduced so that CMI use is safer.
This shall be accomplished by offering a user-centred
software solution that respects users’ trust concerns.
Trust concerns can be regarded as the expression of
doubts in the trustworthiness of other CMI users or
specific interaction situations that differ in each
context (cf. Kipnis, 1996). The proposed method is,
thus, issue- and application-specific.
2 RELATED WORK
To the best of our knowledge, little effort has been put
into methods for incorporating trustworthiness in the
development process of information systems to elicit
software features. Concerning the context of CMI,
Obada-Obieh and Somayaji (2017) have identified
trust mechanisms for the three stages of online dating
applications, which are helpful for users to better
assess the trustworthiness of other users and evaluate
their own safety. Such trust mechanisms are, for
example, safety guidelines shared by the CMI service
provider that give users safety advice on how to
behave in social interactions with other online daters.
Currently not all CMI services offer these guidelines
or those are not well presented so that online dating
users may not notice and benefit from them. Another
proposed trust mechanism for the during stage is to
provide evolving communication steps for a better
verification of CMI users. Those may involve first
text-based and then voice-based communication,
which might end up in time-limited video
conversations. However, these proposals are not
based on a structured development method or
connected to further details valuable for software
engineering. Therefore, they may give first creative
impulses, but not support requirements engineers in
developing relevant trust mechanisms in a structured
way on their own.
Regarding such development methods, main
work is done by Mohammadi et al. (2015), who
introduced the term “trustworthiness-by-design” and
proposed general mechanisms for social-technical
systems. These mechanisms serve as an extension of
existing software development methods by including
procedures for systematically achieving trustworthy
software. Striving for the same objective, Di Cerbo et
al., (2015) suggest considering so-called
trustworthiness certificates in order to measure and
document trustworthiness-related properties of
software during its development. By these
trustworthiness certificates, the relation of
trustworthiness to the information system can be
controlled in every phase of the software life-cycle
process.
However, both works aim to build trustworthy
software in the sense that the system really performs
as it promises. For that reason, they relate their
methodological proposals to the concept of trust. In
contrast, our proposed software development method
considers the concept of trust in order to build
software that supports end-users in evaluating i)
whether users will act as expected and ii) whether
offline encounters are safe. Our method focuses on
interpersonal trust that is mediated by the system.
This kind of trust differs in its nature and
accompanying issues compared to trust in a system.
Our method is especially developed for CMI services
and focuses on users’ mutual trustworthiness
assessment.
3 BACKGROUND
Our method presented in this work aims to give
requirements engineers a step-wise guideline how to
build CMI applications that respect trust in their
design. Therefore, we first give an overview of trust
and trustworthiness in Section 3.1. In Section 3.2, the
trustworthiness framework for CMI (Borchert et al.,
2020) places trust in the context of CMI services. Its
elements are incorporated into our development
method. Furthermore, our method extends the method
for analysing and modelling trustworthiness
requirements by Mohammadi and Heisel (2016a,
2016b), which is presented in Section 3.3. By
referring to the trustworthiness framework for CMI in
our method and building on the method of
Mohammadi and Heisel, we provide a development
approach that is tailored for CMI services. Moreover,
we briefly present the online dating service Plenty of
Fish (POF) in Section 3.4, since we refer to the
application in our illustrative example.
3.1 Trust and Trustworthiness
Research has identified various characteristics as key
elements of trust. On the one hand, trust can be
described as a trustor’s (subject that trusts)
acceptance of and exposure to vulnerability due to
certain risks and uncertainties linked to an interaction
ENASE 2020 - 15th International Conference on Evaluation of Novel Approaches to Software Engineering
270
process with a trustee (subject/object to be trusted)
(Mayer et al., 1995). On the other hand, trust can be
defined as positive expectations a trustor has in the
trustee’s intentions or behaviour (Möllering, 2005).
Those expectations are related to the belief in the
trustee being good and honest towards the trustor
though having the ability to betray (Barber, 1983).
Trust comes into existence based on certain
trustworthiness cues that the trustor perceives from
the trustee. These cues vary depending on the context.
Trustworthiness cues are then assessed by the trustor
so that she can decide whether the trustee is
trustworthy and whether the outcome of an
interaction is fruitful (Beldad et al., 2010).
3.2 The Trustworthiness Framework
for CMI
The trustworthiness framework for CMI (Borchert et
al., 2020) places trust in the context of CMI. It
represents the relation of trust, cues for assessing
trustworthiness and the CMI information system.
Therefore, the framework is considered within the
method presented in this work as it is valuable for
supporting CMI users in their trustworthiness
assessment concerning other users.
The trustworthiness framework considers three
types of trust that are involved with CMI use: i)
system trust, ii) brand trust and iii) computer-
mediated interpersonal trust. They originate from
different disciplines, namely computer science,
sociology, social psychology and business
psychology. Taking the user-perspective in the
context of the framework, the trustor is an individual
CMI end-user, while the trustee differs regarding the
type of trust. In the case of system trust, the trustee is
an impersonal structure (Luhmann, 2018) as for
example an information system (Keymolen, 2016),
which is the CMI system here. In the case of brand
trust, the service provider – meaning the organization
that makes the information system available - can be
regarded as trustee (cf. Ha and Perks, 2005; Thaichon
et al., 2013). Finally, computer-mediated
interpersonal trust describes interpersonal trust
(Rotter, 1980) established via information systems. It
denotes trust in the person of interest with whom the
user interacts via the CMI system. Since the user
assesses whether to trust or not to trust the other user
based on certain cues presented by the information
system, their trust relationship is mediated by the
system itself. Borchert et al. (2020) assume that
especially computer-mediated interpersonal trust
develops during the stages of CMI. This is because
user interactions on CMI portray the development of
interpersonal relationships from the beginning to
oftentimes the end. During interpersonal
relationships, trust in each other is a dynamic variable
that can strongly vary (Lewicki and Wiethoff, 2000).
In comparison, system trust and brand trust are
assumed to be relatively stable during the stages.
They are more relevant for starting with and further
using the CMI application (Borchert et al., 2020).
Another construct of the trustworthiness
framework for CMI are the so-called trustworthiness
facets. Trustworthiness facets represent cues for
assessing trustworthiness originating from the
disciplines mentioned before. Within these
disciplines, trustworthiness cues are related to the
different kinds of trustees and differ in their
terminology.
In the field of computer science, these cues are
called trustworthiness attributes and relate to system
trust (Mohammadi et al., 2013). Examples for
trustworthiness attributes are privacy, security,
usability or data-related quality. Originating from
social psychology, factors of trustworthiness are
linked to interpersonal trust and, thus, considered for
computer-mediated interpersonal trust. Factors of
trustworthiness are benevolence, integrity, ability and
predictability (Mayer et al., 1995). In the context of
business psychology and sociology, cues like
reputation, performance, benevolence or
intentionality have been associated with brand trust
(Sztompka, 1999; Büttner and Göritz, 2009).
Originating from different works, they are not
represented by a specific term like the other cues
before. For distinction reasons, Borchert et al. (2020)
have called them trustworthiness characteristics.
Some facets appear in different disciplines (e.g.
benevolence), but still have a similar meaning. Others
have a different terminology but are highly related
with each other regarding their definition (e.g. ability
and performance). Therefore, it is conceivable that
facets may relate to types of trust that they have not
been considered for originally.
The trustworthiness framework for CMI proposes
to address trustworthiness facets by CMI software
features. As a conclusion, the user is supported in
assessing the trustees. This is assumed to reduce risks
associated with CMI use.
Figure 1 gives an overview of the trustworthiness
framework for CMI by using the UML notation
(OMG, 2003). Here, the relation of trust,
trustworthiness facets and the CMI system become
visually apparent. The framework says that a software
feature, which is part of a CMI system, shall address
trustworthiness facets which in turn affect trust.
System trust, brand trust and computer-mediated
A Conceptual Method for Eliciting Trust-related Software Features for Computer-mediated Introduction
271
interpersonal trust are child classes of trust and, thus,
specified as kinds of trust. The same is applicable for
factors of trustworthiness, trustworthiness
characteristics and trustworthiness attributes
concerning trustworthiness facets.
Figure 1: The trustworthiness framework for CMI
(Borchert et al., 2020).
3.3 Method for Systematic Analysis of
Trustworthiness Requirements by
Mohammadi and Heisel
The method for systematic analysis of trustworthiness
requirements by Mohammadi and Heisel (2016a, b)
serves as basis for the method presented in this paper.
It describes a top-down approach for requirements
engineers whose objective is to achieve
trustworthiness in information systems. An overview
of the method is given by the grey boxes in Figure 2.
The first step of the method of Mohammadi and
Heisel (2016a, b) is to obtain trust concerns of
stakeholders that are involved with the software-to-
be. Trust concerns describe their uncertainness of
whether an outcome of a specific issue is as expected.
Identifying trust concerns is valuable for gaining
further understanding about the stakeholders
themselves, their intentions and the context. Based on
the identified trust concerns, the second step is to
derive trustworthiness goals for the software.
Trustworthiness goals describe the objectives
stakeholders have for the given context and that are
trust-related (Mohammadi et al., 2015). Those are
then addressed by trustworthiness requirements
which determine what capabilities or conditions need
to be considered within the system (IEEE Standard
Glossary of Software Engineering Terminology,
1990). The last step is to relate the requirements to
trustworthiness properties, which realize the
requirements in the business process for software
development. Trustworthiness properties describe
capabilities or qualities the system must meet to
influence trust in a positive way (Mohammadi and
Heisel, 2016a, b). All four steps mutually depend
each other and can be seen as an iterative process.
The method of Mohammadi and Heisel (2016a, b)
strongly relates to the i* goal modelling notation (Yu,
1997) and Business Process Model and Notation
(BPMN) (Stroppi et al., 2011) on a fine granular
model-based level. Goal models are used to map
trustworthiness goals of organizations and other
stakeholders to then relate them to trustworthiness
requirements. They are tailored to the application
context and valuable for obtaining rationales for the
software development. Business process models
visualize activities of business processes as well as
their in- and output in a temporal order. In this
context, they are useful for embedding
trustworthiness requirements within the business
process for developing software. For that reason,
trustworthiness properties are included as elements
within BPMN for directly addressing trustworthiness
in the software development process. In addition to
goal and business process modelling, Mohammadi
and Heisel (2016b) propose pattern-based approaches
for realizing the steps of their method.
3.4 The Online Dating Application
“Plenty of Fish”
Plenty of Fish (POF) is an online dating application
that has users in various countries like the US,
Sweden or Germany. POF reported that it had over 4
million active daily users (datingsitesreviews.com,
2017).
In the stage before users are connected and
interacting with each other, they can edit their profile
by adding pictures or disclosing information like
demographics, appearance, race, religion, interests or
consumption behaviour regarding alcohol or drugs. In
addition, users are able to view partner suggestions
based on the matching of a POF personality
questionnaire, generally browse through pictures and
profiles of POF users or look out for other users based
on the search criteria age, distance or online activity.
In order to begin with the stage during connection and
interaction, users can show their interest in a profile
by signalizing that they would like to interact with the
other user. Another option is to directly start
exchanging messages. After some messages are
exchanged, POF unlocks new communication
features like exchanging pictures, voice messages or
calls. The last stage, which is after a connection or
interaction, can be reached by blocking and, thereby,
ending the connection to another user so that an
ENASE 2020 - 15th International Conference on Evaluation of Novel Approaches to Software Engineering
272
interaction is not possible anymore or simply end the
interaction by not exchanging messages anymore. In
addition to the basic activities of POF, users can
purchase an updated version, which allows them to
get more insights about user profiles, whether a sent
message has been read, who and when someone has
viewed the own profile and to be more often proposed
to others.
Examining trust mechanisms of online dating
applications, Obada-Obieh and Somayaji (2017)
classified POF as one of those services that do not
check the authenticity of user identities. Therefore,
previous traits or records that could jeopardize users’
safety cannot be identified.
4 METHOD FOR ELICITING
TRUST-RELATED SOFTWARE
FEATURES FOR CMI
In this section, the method for eliciting trust-related
software features for CMI is introduced. It follows a
top-down approach that extends the method of
Mohammadi and Heisel (2016a, b) by elements of the
trustworthiness framework (see Figure 2). Unlike the
method of Mohammadi and Heisel (2016a, b), its
objective is not to build trustworthy software, but
CMI software that supports trustworthiness
assessments of its users regarding parties to interact
with. Moreover, the method not only provides an
approach which leverages existing CMI
functionalities, but also provides guidelines for
developing new software features that address users’
safety. For that reason, the method mainly refers to
computer-mediated interpersonal trust even though it
is not precluded that system trust and brand trust
might also be affected by resulting software features.
Figure 2 gives an overview of the method for eliciting
trust-related software features for CMI. It depicts the
original method of Mohammadi and Heisel (2016a,
b) by the grey boxes and shows the extensions via the
green ones. The further one advances in the method,
the more concrete the constructs of each step get for
software development, which correlates with the
dependency on implementation (x- and y-axis). Our
method consists of the following steps that succeed
each other, but may also affect former steps so that
the approach is an iterative process:
1) Identifying trust concerns and workarounds.
2) Deriving trustworthiness goals and
trustworthiness facets.
3) Determining trustworthiness requirements.
4) Inferring trust-related software features.
5) Establishing a collection of trust-related software
features.
In the next subsections, each step will be further
defined and explained. Examples for each step can be
found in the subsections of Section 4.5. There, the
conceptual method is exemplarily applied to the
online dating application POF.
Figure 2: Overview of the method for eliciting trust-related
software features in CMI. The grey boxes represent the
method of Mohammadi and Heisel (2016a, b) (Section 3.3).
The green boxes show the extension of the model of
Mohammadi and Heisel.
4.1 Step 1: Identifying Trust Concerns
and Workarounds
In order to design user-centred software, Marcelino-
Jesus et al. (2014) recommend software engineers to
consider knowledge, concerns and behaviour of the
system’s end-users. Since the method aims to reduce
risks associated with CMI use by considering users’
trustworthiness assessment of other users, we are
especially interested in the trust concerns they have
regarding those. Trust concerns are issues in a
specific area of application that involve uncertainness
whether the outcome of the issue is as expected
(Kipnis, 1996). In addition, trust concerns convey a
lack of trust in the situation or the trustee. Trust
concerns are assumed to be crucial for the interaction
decision with involved parties of CMI services (cf.
Rotter, 1980). However, CMI platforms do not
always meet the users’ trust concerns and lead them
to the application of alternative behavioural strategies
(Obada-Obieh et al., 2017). For that reason, we
additionally consider workarounds in our approach,
because they are relevant for deriving software
features that support users in addressing their trust
concerns or performing their behavioural strategies
directly within the application.
There are different possibilities how trust
concerns and workarounds can be conducted.
A Conceptual Method for Eliciting Trust-related Software Features for Computer-mediated Introduction
273
Mohammadi and Heisel (2016b) introduced the
pattern for identification of trust concerns - though it
does not take workarounds into account. Other
possibilities are to ask experts of the application area
or the user target group. Regarding the user target
group, three kinds of people can be asked:
1) Individuals, who are active in the application field
offline, but are not using any related online
service.
2) Individuals, who are active in the application field
by using related online services from other service
providers and not the one to be
improved/developed.
3) Individuals, who already use an existing version
of the system.
Engineers should choose the respondents depending
on the status of the system to be developed and the
objective they pursue. If only a concept of the service
application exists, then individuals of type one or
experts of the adequate offline activity might be
relevant. Their offline experiences might give
impulses to design the information system. If a
software version already exists, then individuals of
type two or three might be valuable to receive specific
feedback.
4.2 Step 2: Deriving Trustworthiness
Goals and Trustworthiness Facets
For the method presented in this work, we include
both – the trustworthiness goals from the method of
Mohammadi and Heisel (2016a, b) and the
trustworthiness facets from the trustworthiness
framework for CMI (Borchert et al., 2020).
Trustworthiness goals correspond to trust-related
objectives that the various stakeholders intend to
achieve in the given context (Mohammadi et al,
2015). Like in the method of Mohammadi and Heisel
(2016a, b), we intend to derive them from trust
concerns. Trustworthiness goals should be pursued
by the software to be developed to satisfy end-user’s
trust-related objectives, which in turn have an impact
on the overall satisfaction of the application
(Mohammadi and Heisel, 2016a, b).
In contrast to trustworthiness goals,
trustworthiness facets describe cues that have been
identified by literature as important for end-users to
assess the trustee’s trustworthiness (Borchert et al.,
2020). Since facets are a basis for the emergence of
trust, we see a relation to trust concerns and
workarounds. Trust concerns and workarounds refer
to a lack of trust that would not exist if the user had
perceived facets of trustworthiness. Obtaining
knowledge about the facets is important to later
respect and include them in the system design. It is
likely that trustworthiness facets differ regarding the
diverse stages of CMI, since the requirements for
each stage differ, too (Obada-Obieh and Somayaji,
2017). We assume that addressing as many
trustworthiness facets of those that have been
identified as important for end-user over the different
stages of CMI increases the quality of users’
trustworthiness assessment.
Both trustworthiness goals and facets can be
derived from trust concerns and workarounds and
provide an objective and a benchmark for how to
overcome the concerns. Therefore, we assume a
relation between trustworthiness goals and facets. It
is conceivable that a goal may relate to several facets.
Since goals and facets are still on an abstract level
(see Figure 2), we conclude that the relation of
trustworthiness goals and facets can be regarded
detached from specific trust concerns and
workarounds but rather as a general relation valid for
the application area. Therefore, we propose to
establish a collection of trustworthiness goals and
facets for a specific application field, as for example
online dating. Based on the collection, requirements
engineers can infer what trustworthiness goals and
facets a software feature needs to target.
4.3 Step 3: Determining
Trustworthiness Requirements
Trustworthiness requirements are a subtype of
software requirements. Like software requirements,
they can be defined as a condition or capability that i)
is “needed by a user to solve a problem or achieve an
objective” or that ii) “must be met or possessed by a
system or system component to satisfy a contract,
standard, specification, or other formally imposed
documents” (IEEE Standard Glossary of Software
Engineering Terminology, 1990). However,
trustworthiness requirements target the trust issue and
are specifically characterized by addressing end-
users’ trust concerns (Mohammadi and Heisel,
2016c). Therefore, they are valuable for purposefully
developing trust-related software features by
determining concrete configurations for service-
based systems.
In the method of Mohammadi and Heisel (2016b),
trustworthiness requirements are a further refinement
of the previously identified goals. Similar to their
approach, we aim to determine trustworthiness
requirements from the collection of trustworthiness
goals and facets. By using the collection,
trustworthiness requirements address a goal and also
ENASE 2020 - 15th International Conference on Evaluation of Novel Approaches to Software Engineering
274
consider the manner how the goal is achieved –
namely by respecting the facets. This means that a
trustworthiness goal is met by at least one
trustworthiness requirement that in turn addresses at
least one of the trustworthiness facets associated with
the goal.
4.4 Step 4: Inferring Trust-related
Software Features
In the last step of this method, trust-related software
features are inferred from trustworthiness
requirements from step three. Software features are a
very abstract concept, for which a multitude of
definitions exist (Berger et al., 2015). Common
definitions describe features as “a logical unit of
behaviour specified by a set of functional and [non-
functional] requirements” (Bosch, 2000, p.194) or “a
feature is also a distinguishable characteristic of a
concept (e.g. system, component, etc.) that is relevant
to some stakeholder of the concept” (Robak et al.,
2002, p.288). They can be seen as reusable solutions
within a software for a specific problem
corresponding to, for example, user-interface
requirements, certain application logics or tasks on an
infrastructural level (Berger et al., 2015). In the case
of our method, we speak of trust-related software
features, because we set them in the context of trust
concerns end-users have. Trust-related software
features are particularly valuable to CMI services
because they relate to trustworthiness facets that are
important to help users assess the trustworthiness of
other end users and the safety of interaction.
For deriving and developing trust-related software
features, identified workarounds from step one can
serve as creative support for the practitioner of this
method. By keeping workarounds in mind, one can
make sure to include software-features in the system
that are not yet available but required by the users. It
is up to the practitioner’s expertise or creativity how
trustworthiness requirements can be realized by trust-
related software features. Another option is to consult
experts or take a look at existing solutions to adapt
them to CMI services.
With the trust-related software features, step four
deviates from the method of Mohammadi and Heisel
(2016b), because they replace trustworthiness
properties (Figure 2, greyed out). Both trust-related
software features and trustworthiness properties are
concrete and implementable (see Figure 2, axis).
However, trustworthiness properties have a very
close connection to BPMN, which is included in this
method. Moreover, they do not address
trustworthiness facets, which is crucial for the deve-
development of CMI services.
4.5 Step 5: Establishing a Collection of
Trust-related Software Features
In order to support reusable solutions for specific
CMI applications, we propose to establish a
collection of trust-related software features that
contains a solution portfolio of implementable trust-
related software features. It builds upon the collection
of trustworthiness goals and facets mentioned in
Section 4.2. The collection serves as an overview of
identified constructs of the whole method, namely
trust concerns, workarounds, trustworthiness goals,
facets, requirements, software features and the CMI
stage the features are relevant for. A feature is linked
to a trustworthiness requirement and the associated
trustworthiness goal and facets. It does not need to
address all of the associated facets, but at least one.
During the development process and by building the
collection, additional facets that do not yet refer to the
discussed trust concern, can by identified as relevant
for the features. This shows the iterative process of
the method and support enhancing system design.
The collection serves as a documentation of the
method and facilitates a structured detection of i)
trustworthiness facets that are not yet included in the
system, ii) the identification of requirements and iii)
the appropriate software features. The objective is to
collect a multitude of software features over time so
that in the end, every trust concern and facet is
covered. We assume that this maximises the support
that can be provided for the user’s trustworthiness
assessment.
4.6 Example: Applying the Method for
Eliciting Trust-related Software
Features to the Online Dating
Application Plenty of Fish
POF is an online dating app, where end-users are
mainly responsible for their own security and safety
(Quiroz, 2013). By applying the here presented
method, POF users could be more supported by
offering trust-related software features. This example
illustrates the method step-by-step concerning a
specific use case. In the case of another instance, the
explicit constructs of the method can be completely
different. In the end, Section 4.6.6 shows an
exemplary collection of trust-related software
features for POF that documents the results of the
method.
A Conceptual Method for Eliciting Trust-related Software Features for Computer-mediated Introduction
275
4.6.1 Example: Trust Concerns and
Workarounds
The first step of the method is to identify trust
concerns and workarounds of POF. As far as we
know, no explicit research has been done in this
direction for POF. For this example, we therefore rely
on general trust concerns concerning online dating.
Online dating users have stated that they are
worried whether profiles are fake or not. In order to
check the authenticity of profiles, they employ the
workaround of looking for the person concerned on
other social network sites (Obada-Obieh et al., 2017).
We assume that this especially occurs in the stages
before and during a connection/match of two end-
users, when end-user decides to start or continue an
interaction.
4.6.2 Example: Trustworthiness Goals
Based on trust concerns and workarounds,
trustworthiness goals can be determined. The goal of
end-users in this context is to check the authenticity
of other users. Authenticity means that a presented
profile corresponds to a true identity. A true identity
is not conform with the misrepresentation of
identifying personal information like name, age,
ethnicity, gender, marital status or job, for example
(Leppänen et al., 2015). Authenticity precludes
identity theft or social bots (Douceur, 2002; Jin et al.,
2011). Currently, POF does not have any mechanisms
for verifying user authenticity (Obada-Obieh and
Somayaji, 2017).
4.6.3 Example: Trustworthiness Facets
In order to check the authenticity of another user, end-
users need cues like trustworthiness facets for
assessment. Before interacting with someone, users
tend to examine online dating profiles for further
information (Obada-Obieh et al., 2017). The more
detailed information is provided in a profile, the better
the trustworthiness assessment. Therefore, users may
look out for data-related quality, which is a facet that
describes the way information is provided
(Mohammadi et al., 2013). During the interaction,
facets like honesty and performance could be relevant
for checking whether a profile is fake or not. While
honesty means that users say the truth (Xia, 2013),
performance displays the actual behaviour presented
by the interaction partner (Sztompka, 1999).
The trustworthiness facets data-related quality,
honesty and performance can be linked to the
trustworthiness goal of checking authenticity. These
relations are detached from the POF example. This
illustrates to the description of the collection of
trustworthiness goals and facets (Section 4.2) as a
general overview of the relationship of goals and
facets, which is valid for the application area online
dating applications.
4.6.4 Example: Trustworthiness
Requirements
Trustworthiness requirements (TR) describe what
condition or capability POF needs to include, which
must correspond to the trustworthiness goal and relate
to at least one facet.
In order to satisfy data-related quality, POF needs
to provide information about users that are deemed
interesting or useful. Therefore, a requirement is to
obtain such information, which can be done by asking
users for self-disclosure (TR1).
Moreover, it is valuable for users to know whether
self-disclosed information of a profile corresponds to
the truth and represents a user’s identity. For that,
POF requires to prove and notify users about this
circumstance (TR2, TR3). In doing so, POF addresses
the facet honesty. In the case of honesty and
performance, both facets can be addressed, if POF
proves and notifies, or enables users to prove,
whether disclosed information of a user matches the
behaviour she shows (TR4, TR5, TR6).
4.6.5 Example: Trust-related Software
Features
Having a look at the trustworthiness requirements,
trust-related software features need to formulate how
these can be realized in a concrete way. They shall
address related trustworthiness facets and be assigned
to a CMI stage. An overview of the identified trust-
related software features is given in Table 1.
Requirement TR1 can be put into practice by
offering users empty text input fields for information
that they can include in their profile to motivate self-
disclosure (SF1). This feature is linked to data-related
quality and is relevant for the before stage, when
users create their profile.
Another possibility to realize TR1 is to trigger
self-disclosure behaviour by unlocking online dating
functionalities (e.g. providing access to more
information of other users or allowing to exchange
messages with other users), if the profile is mostly
completed (SF2). This software feature is provided by
the German online dating website Parship, for
example. Again, this feature is relevant for the before
stage and refers to data-related quality.
ENASE 2020 - 15th International Conference on Evaluation of Novel Approaches to Software Engineering
276
Table 1: Example for the collection of trust-related software features concerning the trust concern whether online dating
profiles are fake or not. Trust concern, workaround and trustworthiness goals are omitted in this table due to space constraints,
but briefly summarized in Section 4.6.6.
Trustworthiness
Requirement (TR)
Trust-Related Software Feature
Trustworthiness
facets
CMI
stage
TR 1: Asking users to
disclose more
information.
SF1: Including empty text input fields to motivate self-
disclosure
-Data-related
quality
before
SF2: Unlock online dating functionalities (e.g. accessing more
profile information of others) for completed profiles.
-Data-related
quality
before
Requirement 2:
Proving whether
profiles represent a
true identity.
SF3: Asking users to upload a photograph of their ID, which is
then manually checked by the service.
-Honesty before
Requirement 3:
Notifying users about
verification of profiles.
SF4: Graphical icon that classifies a profile as being verified.
(relates to SF3)
-Data-related
quality,
Honesty
Requirement 4:
Proving whether
disclosed information
matches shown
behaviour.
SF5: Algorithm that compares disclosed information of profile
and within communication.
-Honesty,
Performance
during
SF6: Algorithm to prove whether user does not comply to
“terms of use agreement” (e.g. identifying strong language as an
indicator for bullying)
-Honesty,
Performance
during
Requirement 5:
Notifying users about
mismatch of disclosed
information and shown
behaviour.
SF7: Warning message when there is a mismatch about
information disclosed during communication and within profile
(relates to SF5).
-Data-related
quality,
Honesty,
Performance
during
SF8: Warning message that informs users about own
misbehaviour and possible consequences. (relates to SF6)
-Performance during
SF9: Message to inform users affected by another user’s
misbehaviour. Comforting him/her and showing coping
strategies (e.g. blocking user, contact for finding help) (relates
to SF6)
-Benevolence,
Performance
during
Requirement 6:
Enabling users to
check users’
authenticity.
SF10: Option to link online dating profile with other social
media accounts (e.g. Instagram, Spotify) so that other users
have access to it
-Honesty,
Performance
before,
during
Requirement TR2 can be realized by proving the
user’s ID card. POF could ask users to photograph
and upload it (SF3). After the ID is manually checked
by POF employees, profiles could receive a graphical
icon notifying users that the profile is verified (SF4,
referring to TR3). This feature is used by sharing
economy platforms like Airbnb. Online dating users
also have stated interest in this feature (Obada-Obieh
et al., 2017). It is relevant before the interaction starts.
By agreeing on this feature, users can prove their
honesty. In addition, the graphical icon for
verification is also linked to data-related quality,
because it provides users with the additional
information about verification.
For realizing TR4, POF could check whether
information disclosed in messages during
communication correspond to those that have been
disclosed in the profile using an algorithm (SF5). If
the algorithm finds a mismatch, POF could display a
warning message directly after the behaviour has
been shown (SF7 referring to TR5). This relates to the
facets honesty and performance. Moreover, the
notification is again an indication for data-related
quality by the system. This feature is relevant for the
stage during interaction.
A similar comparison of honesty and performance
can be done by pointing out user behaviour that is
incongruent to the “terms of use agreement”, which
every POF user has to accept before usage (SF6
A Conceptual Method for Eliciting Trust-related Software Features for Computer-mediated Introduction
277
referring to TR4). There, for example, users have
agreed to not “harass, bully, stalk, intimidate, assault,
defame, harm or otherwise mistreat any person”. If
POF detects strong language within messages that
refers to such a behaviour, the person showing this
behaviour can be admonished by referring to possible
consequences if the behaviour is shown again (e.g.
banned from community) (SF8 referring to TR5).
This feature would refer to the users’ performance
and is relevant for the stage during interaction.
Moreover, the affected user could receive a message
of POF, which comforts him/her and offers
possibilities how to cope with it (e.g. blocking or
reporting) (SF9 referring to TR5). Such a feature
could help user’s in their well-being and safety. This
feature gives feedback to the performance of other
users. In addition, it relates to the facet benevolence
shown by the service provider or system towards its
users. At this point, requirements engineers should
think of including benevolence as a facet for this use
case. This triggers the development process in
looking for more features how this facet can be
satisfied, which improves system design. Features
SF6, SF8 and SF9 are relevant for the stage during
interaction.
Requirement TR6 is about users checking the
match of disclosed information and shown behaviour
on their own. This can be accomplished by realizing
the identified workaround within the POF
application. Therefore, POF could enable users to link
their profile with other social media accounts, such as
Instagram or Spotify, so that connected users have
access to it. By this feature, users can represent their
honesty and performance. Currently, POF is one of
the few online dating applications that does not offer
this feature (Obada-Obieh and Somayaji, 2017). This
feature is relevant for the before and during stages.
4.6.6 Example: Collection of Trust-related
Software Features
The collection of trust-related software features
summarizes the results of the whole method. Table 1
shows how the collection can be built. In order to
avoid a too large table, trust concerns, workarounds
and trustworthiness goals are omitted here. The trust
concern for this exemplary collection describes that
online daters fear fake profile that do not represent a
true identity. Workarounds imply the check of other
users on additional social network sites. The
trustworthiness goal for POF is to check the
authenticity of user profiles.
5 DISCUSSION AND FUTURE
WORK
This work introduces a conceptual method for the
elicitation of trust-related software features in CMI
services. Our method aims at supporting requirements
engineers in both, the development of new features
and the improvement of existing CMI services. The
objective is to reduce risks associated with CMI use
by supporting end-users in assessing the
trustworthiness of other users and their safety during
an interaction. Therefore, the method considers end-
users’ trust concerns, their workarounds,
trustworthiness goals and trustworthiness facets to
establish trustworthiness requirements, which are the
basis for developing trust-related software features.
Properties of CMI services are the introduction
and interaction with unfamiliar users online that
might lead to offline encounters. Based on that, CMI
usage can be divided into the stages before, during
and after a connection/match of two end-users.
Obada-Obieh and Somayaji (2017) have detected
different requirements concerning the stages of CMI
services (in particular online dating) and a need for
trust mechanisms in online dating applications.
Therefore, they propose ideas of how trust
mechanisms can look like for each stage. Our method
complies to their findings and provides a structural
approach for developing such trust mechanisms,
which we call trust-related software features. By
applying the method, requirements engineers are
encouraged to be diligent while formulating
requirements and linked facets for the CMI service.
This leads to a concrete description of software
features and an enhanced software design.
Having a detailed look at trust and
trustworthiness, trustworthiness facets are key
elements in our method for reducing risks in CMI.
They are the basis for deciding whether to trust
someone and whether an interaction is safe. Since
trust concerns for online dating are assumed to differ
concerning the various stages of CMI (cf. Gibbs et al.,
2011), this also applies for trustworthiness facets.
With the help of our method, software features can be
tailored to the relevant facets for the individual stages
leading to improved CMI use.
Overall, resulting trustworthiness requirements
and software features follow one specific goal and
meets one specific trust concern. However, they may
get into conflict with other trustworthiness goals
across the system. For instance, TR1 of our example
could jeopardize a trustworthiness goal such as
privacy because it demands the disclosure of personal
information from the users. Therefore, future research
ENASE 2020 - 15th International Conference on Evaluation of Novel Approaches to Software Engineering
278
should investigate ways to resolve conflicts between
requirements/features while maximizing the users’
benefit.
Moreover, this work describes a conceptual
method whose stepwise realization is not further
determined, yet. Mohammadi and Heisel (2016a)
have specified patterns for the identification of trust
concerns and for the specification of trustworthiness
requirements, in order to provide requirements
engineers with clear guidelines within their method.
Other possibilities are qualitative approaches like
interviews of user target groups or experts (Hubbard
et al., 2000). In future work, we will further define
how the method can exactly be applied step by step.
A limitation of this work is that it is based on
former research and theoretical conclusions.
However, it is important to evaluate the method by
using it for a concrete development of a CMI
application. In addition, relationships proposed by the
trustworthiness framework for CMI (Borchert et al.,
2020) can also be tested. It is important to prove,
whether the proposed software features really have an
effect on system trust, brand trust or computer-
mediated interpersonal trust.
Moreover, it is a challenge to evaluate whether
software developed with this method really reduces
risks and supports safe offline encounters. Future
research should survey end-users and developers
about their perception in this point or conduct long-
term studies to observe whether the rate of unwanted
incidents is reduced.
6 CONCLUSION
This work proposes an approach for requirements
engineers to build CMI services that support end-
users in their mutual trustworthiness assessment in
the CMI stages before, during and after they are
connected with each other. The objective is to reduce
risks associated with CMI use to increase the rate of
safe offline encounters. In order to accomplish this,
the method considers end-users’ trust concerns and
counter strategies to infer trustworthiness goals and
trustworthiness facets. By considering those,
requirements for CMI services can be obtained that
most likely have an impact on computer-mediated
interpersonal trust-relationships. Based on that, trust-
related software features can be derived that support
a safer use in each CMI stage and increases user
satisfaction with the CMI service. Since this work
presents a conceptual method for eliciting trust-
related software features for CMI, future work tackles
the refinement of the method by detailed procedures
for each step.
ACKNOWLEDGEMENTS
This work was supported by the Deutsche
Forschungsgemeinschaft (DFG) under grant No.
GRK 2167, Research Training Group “User-Centred
Social Media”.
REFERENCES
Barber, B. 1983. The logic and limits of trust (Vol. 96). New
Brunswick, NJ: Rutgers University Press.
Beldad, A., De Jong, M., Steehouder, M. 2010. How shall I
trust the faceless and the intangible? A literature review
on the antecedents of online trust. Computers in human
behavior, 26(5), 857-869.
Berger, T., Lettner, D., Rubin, J., Grünbacher, P., Silva, A.,
Becker, M., Czarnecki, K. 2015. What is a feature?: a
qualitative study of features in industrial software
product lines. In Proceedings of the 19th International
Conference on Software Product Line, 16-25. ACM.
Borchert, A., Diáz Ferreyra, N. E., Heisel, M. 2020.
Submitted for publication.
Bosch, J. 2000. Design and use of software architectures:
adopting and evolving a product-line approach.
Pearson Education.
Büttner, O. B., Göritz, A. S. 2008. Perceived
trustworthiness of online shops. Journal of Consumer
Behaviour: An International Research Review, 7(1),
35-50.
Couch, D., Liamputtong, P., Pitts, M. 2012. What are the
real and perceived risks and dangers of online dating?
Perspectives from online daters: Health risks in the
media. Health, Risk & Society, 14(7-8), 697-714.
Datingsitesreviews.com, 2017. Plenty of Fish debuts new
conversation feature and redesigned app. Retrieved
December 2019, from https://www.datingsitesreviews.
com/users.php?mode=profile&uid=4448.
Di Cerbo, F., Mohammadi, N. G., Paulus, S. 2015.
Evidence-based trustworthiness of internet-based
services through controlled software development. In
Cyber Security and Privacy Forum, 91-102, Springer,
Cham.
Douceur, J. R. 2002. The sybil attack. In International
workshop on peer-to-peer systems, 251-260. Springer,
Berlin, Heidelberg.
Gibbs, J. L., Ellison, N. B., Lai, C. H. 2011. First comes
love, then comes Google: An investigation of
uncertainty reduction strategies and self-disclosure in
online dating. Communication Research, 38(1), 70-100.
Ha, H. Y., Perks, H. 2005. Effects of consumer perceptions
of brand experience on the web: Brand familiarity,
satisfaction and brand trust. Journal of Consumer
Behaviour: An International Research Review, 4(6),
438-452.
A Conceptual Method for Eliciting Trust-related Software Features for Computer-mediated Introduction
279
Hubbard, R., Schroeder, C. N., Mead, N. R. 2000. An
assessment of the relative efficiency of a facilitator-
driven requirements collection process with respect to
the conventional interview method. In Proceedings
Fourth International Conference on Requirements
Engineering. ICRE 2000. 178-186. IEEE.
IEEE Standards Coordinating Committee. 1990. IEEE
Standard Glossary of Software Engineering
Terminology (IEEE Std 610.12-1990). Los Alamitos.
CA: IEEE Computer Society, 169.
Jin, L., Takabi, H., Joshi, J. B. 2011. Towards active
detection of identity clone attacks on online social
networks. In Proceedings of the first ACM conference
on Data and application security and privacy, 27-38.
ACM.
Keymolen, E. 2016. Trust on the line: a philosophycal
exploration of trust in the networked era.
Kipnis, D. 1996. Trust and technology. Trust in
organizations: Frontiers of theory and research, 39, 50.
Leppänen, S., Møller, J. S., Nørreby, T. R., Stæhrc, A.,
Kytölä, S. 2015. Authenticity, normativity and social
media. Discourse, Context and Media, 8.
Lewicki, R. J., Wiethoff, C. 2000. Trust, trust development,
and trust repair. The handbook of conflict resolution:
Theory and practice, 1(1), 86-107.
Luhmann, N. 2018. Trust and power. John Wiley & Sons.
Marcelino-Jesus, E., Sarraipa, J., Agostinho, C., Jardim-
Goncalves, R. 2014. A Requirements Engineering
Methodology for Technological Innovations
Assessment. In ISPE CE, 577-586.
Mayer, R. C., Davis, J. H., Schoorman, F. D. 1995. An
integrative model of organizational trust. Academy of
management review, 20(3), 709-734.
Mohammadi, N. G., Paulus, S., Bishr, M., Metzger, A.,
Koennecke, H., Hartenstein, S., Pohl, K. 2013. An
Analysis of Software Quality Attributes and Their
Contribution to Trustworthiness. In CLOSER, 542-552.
Mohammadi, N. G., Bandyszak, T., Paulus, S., Meland, P.
H., Weyer, T., Pohl, K. 2015. Extending Software
Development Methodologies to Support
Trustworthiness-by-Design. In CAiSE Forum, 213-220.
Mohammadi, N. G., Heisel, M. 2016a. Patterns for
identification of trust concerns and specification of
trustworthiness requirements. In Proceedings of the
21st European Conference on Pattern Languages of
Programs, 31. ACM.
Mohammadi, N. G., Heisel, M. 2016b. A framework for
systematic analysis and modeling of trustworthiness
requirements using i* and BPMN. In International
Conference on Trust and Privacy in Digital Business,
3-18. Springer, Cham.
Mohammadi, N. G., Heisel, M. 2016c. Enhancing business
process models with trustworthiness requirements. In
IFIP International Conference on Trust Management,
33-51. Springer, Cham.
Möllering, G. 2005. The trust/control duality: An
integrative perspective on positive expectations of
others. International sociology, 20(3), 283-305.
Obada-Obieh, B., Somayaji, A., 2017. Can I believe you?:
Establishing Trust in Computer Mediated
Introductions. In Proceedings of the 2017 New Security
Paradigms Workshop, 94-106. ACM.
Obada-Obieh, B., Chiasson, S., Somayaji, A. 2017. “Don’t
Break My Heart!”: User Security Strategies for Online
Dating. In Workshop on Usable Security (USEC).
Object Management Group, 2003. UML 2.0 Infrastructure
– Final adopted specification. http://www.omg.org/
docs/ad/03-09-15.pdf
Quiroz, P. A. (2013). From Finding the Perfect Love Online
to Satellite Dating and ‘Loving-the-One-You’re Near’
A Look at Grindr, Skout, Plenty of Fish, Meet Moi,
Zoosk and Assisted Serendipity. Humanity & Society,
37(2), 181-185.
Robak, S., Franczyk, B., Politowicz, K. 2002. Extending
the UML for modelling variability for system families.
International Journal of Applied Mathematics and
Computer Science, 12, 285-298.
Rotter, J. B. 1980. Interpersonal trust, trustworthiness, and
gullibility. American psychologist, 35(1), 1.
Stroppi, L. J. R., Chiotti, O., Villarreal, P. D. 2011.
Extending BPMN 2.0: method and tool support. In
International Workshop on Business Process Modeling
Notation, 59-73. Springer, Berlin, Heidelberg.
Sztompka, P. 1999. Trust: A sociological theory.
Cambridge University Press.
Thaichon, P., Quach, T. N., Lobo, A. 2013. Marketing
communications: Factors influencing brand loyalty of
internet service provider. In meeting of Australian and
New Zealand Marketing Academy Conference.
Auckland, New Zealand.
Walther, J. B., Loh, T., Granka, L. 2005. Let me count the
ways: The interchange of verbal and nonverbal cues in
computer-mediated and face-to-face affinity. Journal of
language and social psychology, 24(1), 36-65.
Xia, L. 2013. Effects of companies' responses to consumer
criticism in social media. International Journal of
Electronic Commerce, 17(4), 73-100.
Yu, E. S. K. 1997. Towards modelling and reasoning
support for early-phase requirements engineering. In
Proceedings of ISRE'97: 3rd IEEE International
Symposium on Requirements Engineering, 226-235.
IEEE.
ENASE 2020 - 15th International Conference on Evaluation of Novel Approaches to Software Engineering
280