Towards task-sensitive assistance in public spaces
| Pages | 344-367 |
| Date | 20 May 2019 |
| Published date | 20 May 2019 |
| DOI | https://doi.org/10.1108/AJIM-07-2018-0179 |
| Author | Melanie A. Kilian,Markus Kattenbeck,Matthias Ferstl,Bernd Ludwig,Florian Alt |
| Subject Matter | Library & information science |
Towards task-sensitive
assistance in public spaces
Melanie A. Kilian, Markus Kattenbeck, Matthias Ferstl and
Bernd Ludwig
Department of Information Science,
University of Regensburg, Regensburg, Germany, and
Florian Alt
Research Center for Cyber Defence (CODE),
Bundeswehr University Munich, Munich, Germany
Abstract
Purpose –Performing tasks in public spaces can be demanding due to task complexity. Systems that can
keep track of the current task state may help their users to successfully fulfill a task. These systems, however,
require major implementation effort. The purpose of this paper is to investigate if and how a mobile
information assistant which has only basic task-tracking capabilities can support users by employing a least
effort approach. This means, we are interested in whether such a system is able to have an impact on the way
a workflow in public space is perceived.
Design/methodology/approach –The authors implement and test AIRBOT, a mobile chatbot application
that can assist air passengers in successfully boarding a plane. The authors apply a three-tier approach
and, first, conduct expert and passenger interviews to understand the workflow and the information
needs occurring therein; second, the authors implement a mobile chatbot application providing minimum
task-tracking capabilities to support travelers by providing boarding-relevant information in a proactive
manner. Finally, the authors evaluate this application by means of an in situ study (n¼101 passengers) at a
major European airport.
Findings –The authors provide evidence that basic task-tracking capabilities are sufficient to affect the
users’task perception. AIRBOT is able to decrease the perceived workload airport services impose on users.
It has a negative impact on satisfaction with non-personalized information offered by the airport, though.
Originality/value –The study shows that the number of features is not the most important means to
successfully provide assistance in public space workflows. The study can, moreover, serve as a blueprint to
design task-based assistants for other contexts.
Keywords Human-computer interaction, Assistance system, Cooperative problem solving, In situ study,
Mobile information behaviour, Mobile information needs
Paper type Research paper
1. Introduction
Chatbots have seen increasing interest in recent years (e.g. Chandel et al., 2018; Ma and Ho,
2018). One particular focus has been on acquiring knowledge about the cohesiveness of
discourse across multiple turns based on dialogue corpora (e.g. De Gasperis et al., 2013;
AbuShawar and Atwell, 2013; Callejas-Rodrıguez et al., 2016). State-tracking capabilities of
chatbots have seen less interest, though. This is a major research gap, in particular with
respect to mobile chat-based information companions which may be used in scenarios where
providing information in a timely manner is highly important. These activities, which are
often performed in public spaces, are frequently standardized, resulting in two challenges
for people: first, the correct order of sequences is less obvious for inexperienced persons
(e.g. first time air passengers). Second, disruptions of the sequence of activities by
unforeseen events may require users to find alternative solutions (e.g. switching modes of
public transportation in case of breakdowns of a bus). In both cases, the perceived
complexity of the activities performed in public spaces increases heavily, even for
experienced persons. Personal information systems capable of tracking the current state of
the activities can provide a means to reduce this perceived complexity. Building systems
Aslib Journal of Information
Management
Vol. 71 No. 3, 2019
pp. 344-367
© Emerald PublishingLimited
2050-3806
DOI 10.1108/AJIM-07-2018-0179
Received 27 July 2018
Revised 15 November 2018
20 March 2019
Accepted 21 March 2019
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/2050-3806.htm
344
AJIM
71,3
requires major implementation effort, though. With these costs in mind, it is useful to, first,
understand the benefits a chatbot with basic capabilities to track a user’s state within a
particular workflow might even have.
This paper reports on an empirical investigation of this least effort question by focusing
on support provided to departing passengers at an airport prior to boarding. To this end, we
take a multi-perspective approach which is reflected in the structure of this paper: first, we
report on a two-part interview study (see Section 3.2) during which we interviewed both
airport staff members (i.e. information experts) and departing air passengers. This study
was a suitable means to learn about the ideal workflow from the staff members’perspective
and the degree as to which passengers share their point of view. Taken together, both parts
of the interview study reveal a rather simple workflow model and a set of requirements an
information companion should fulfill. Subsequently, we take a least effort approach to
understand whether a personal information companion with a minimal set of state-tracking
abilities can positively influence the perception of passenger services[1]. We, therefore,
implement a chatbot application called AIRBOT, which is able to inform its users
proactively due to its basic state-tracking capabilities (see Section 3.3). Using this prototype
application we conduct an in situ user study using live data streams (see Section 4). We find
that even a system with minimal capabilities in natural language understanding and
generation has a positive effect on the perceived workload, resulting in an increased
satisfaction with passenger services (SPAS).
2. Related work
Three different strands of prior evidence are important for the work presented in this paper.
The first of these provides a justification for studying personalized assistance by reviewing
literature on delivering informationin public spaces. The second field of interest is concerned
with state-of-the-art, goal-oriented chatbots which focuses on the importance of task
awarenessto provide meaningful personalized assistance.Finally, the third partof this section
deals with existing mobile information systems providing assistance to airport passengers.
2.1 Delivering information in public spaces
Digital displays to inform and guide persons in public spaces have seen increasing interest
within the HCI community up to today (see, e.g. Iwai et al., 2006; Sato et al., 2010; Sharifi et al.,
2006; Väänänen-Vainio-Mattila et al., 2013; Tomitsch et al., 2014; Shi and Alt, 2016). All of
these share the conclusion that these displays are a useful means of presenting information,
if and only if they are visible and noticed. This conclusion implies that it is a major challenge
to draw the users’attention to these displays in public spaces (see, e.g. Muller et al., 2010,
2012 for studies on capturing a user’s attention with this display type).
Personalized information companions, such as smartphones, however, can push
information to their users and, thus, can draw the users’attention more easily.
Personalization is much easier to achieve using these systems compared to public displays,
which first need to identify users (see, e.g. Davies et al., 2014; Alt and Vehns, 2016). Location-
based tourism services for mobile devices have been among the first application areas with
respect to personalization. Studies such as Schmidt-Belz et al. (2003) provide evidence that
participants perceive this solution as more beneficial compared to other, non-personalized
information sources like free web content, books or maps. In addition to personalization,
context awareness of information presented on smartphones has been revealed to be
particularly important (see, e.g. Sohn et al., 2008; Nugent et al., 2015). Overall, prior evidence
suggests that a personal device can be a useful means of providing information to users in a
personalized and context-aware manner. Proactively acting chatbots are one option, among
others, to fulfill both goals.
345
Task-sensitive
assistance in
public spaces
To continue reading
Request your trial