Page 93 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 5 – Internet of Everything
P. 93
ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 5
L YSIS CHATBOT: A VIRTUAL ASSISTANT FOR IOT PLATFORMS
1
2
Raimondo Cossu , Roberto Girau , Luigi Atzori 3
1,3
Dept. of Electrical and Electronic Engineering, University of Cagliari and National Telecommunication Inter‑University
2
Consortium (CNIT), Research Unit of Cagliari, Italy., Dept. of Computer Science and Engineering,
University of Bologna, Italy
NOTE: Corresponding author: Roberto Girau, roberto.girau@unibo.it
Abstract – The con iguration and management of devices and applications in Internet of Things (IoT) platforms may be
very complicated for a user, which may limit the usage of relevant functionalities and which does not allow its full potential
to be exploited. To address this issue, in this paper we present a new chatbot which is intended to assist the user in interacting
with an IoT platform and allow them to use and exploit its full potential. The requirements for a user‑centric design of the
chatbot are irst analyzed, then a proper solution is designed which exploits a serverless approach and makes extensive use of
Arti icial Intelligence (AI) tools. The developed chatbot is integrated with Telegram to message between the user and the Lysis
IoT platform. The performance of the developed chatbot is analyzed to assess its effectiveness when accessing the platform,
set the main devices’ parameters and request data of interest.
Keywords – Chatbot, IoT platform, Lysis IoT, user experience
1. INTRODUCTION A chatbot can be used for a variety of purposes, and the In‑
ternet of Things (IoT) can easily be added to this list. One
In recent years, the development and deployment of chat‑ of the reasons why the IoT is struggling to take off is the
bots to be used in several scenarios have risen signi i‑ dif iculty of less experienced users installing or con igur‑
cantly, and many businesses have opted to use these in ing their devices, as well as solving small, common prob‑
their services. In many sectors, such as e‑commerce, in‑ lems. This forces users to rely on quali ied staff to ix sim‑
surance, banking, healthcare, inance, legal, and others, ple problems on a regular basis, which makes the user ex‑
chatbots are currently used to support the execution of perience frustrating. In this scenario, a chatbot can have a
a variety of business activities. Gartner Summits [1] pre‑ vital role in improving the user experience, as when prop‑
dicts that over 70 % of customer interactions will involve erly programmed and inserted within the reference IoT
emerging technologies such as Machine Learning (ML) platform, it would give the user the necessary support
applications, chatbots and mobile messaging by 2022. when dealing with complicated actions thus ful illing the
The objective of a chatbot is to emulate the conversa‑ lack of skills. Obviously, it would not only provide some
tional capabilities of humans so that when a person in‑ help in performing speci ic actions but it would also pro‑
teracts with a chatbot they behave as if they were inter‑ vide information that will be speci ically requested by the
acting with a peer. This is possible because the chatbot user, such as, for example, about the status of their car,
goes through a series of steps to process human data and home, work and so on, signi icantly reducing the barriers
then determine an appropriate response or action based between the user and connected objects. All in natural
on the user’s query. There are already various examples language (human language) rather than relying on navi‑
of Arti icial Intelligence (AI)‑based chatbots, for example: gation through a graphical interface in a mobile applica‑
Cleverbot, Cortana or Tay. First of all, Tay [2], Microsoft’s tion or website. Unfortunately, progress is currently be‑
irst public experiment involving the test of a bot on Twit‑ ing made at a snail’s pace. The truly conversational chat‑
ter, was so successful that it began to behave like its fol‑ bot, which will be able to autonomously interpret user in‑
lowers. Over time, however, after just 16 hours of activity puts, is still a long way off, but several research efforts
it was necessary to turn it off because they she had be‑ are moving towards the uni ication of an ecosystem that
gun to exhibit xenophobic, feminist and racist behavior. is currently very fragmented. The IoT space is at an in‑
It was a similar ending for the conversation between two lection point, with conversational user interfaces at the
AI entities developed in the Facebook labs, trying to make forefront. This process is becoming more achievable day‑
them talk to each other, after some time they began to to-day with services that help companies to easily inte‑
speak a language that was known only to them. While the grate natural language understanding into their products.
epilogue was not what was expected, these experiments The scope for conversational user interfaces is enormous,
showed how much the technology around smart chatbots and it continues to expand. With a variety of technology
had evolved. Chatbots are a hot topic among tech be‑ available for the implementation of such systems, the
hemoths like Facebook and Microsoft, as well as smaller next step is to figure out where machine learning
messaging platforms like Telegram and Slack, which have strategies make more sense than other technologies and
made their frameworks available to developers to ensure whether they can potentially save us time and enable
smooth development. people to focus on more valuable tasks.
© International Telecommunication Union, 2021 81