Abstract
Smart speakers equipped with intelligent voice processing functions, like Siri and Amazon Echo, have become widespread globally. However, their recording and networking capabilities spur concerns about users’ privacy. In this research, we report the results of a text-mining analysis of customers’ perception of privacy in smart speakers. The corpus comprises over 4,500 reviews of the Echo line of smart speakers collected from Amazon. Smart speakers owners appear to be mostly oblivious to privacy issues since the fraction of those mentioning privacy in their reviews is largely below 3%. The average sentiment towards privacy is positive, though the average hides a significant fraction exhibiting a negative sentiment. However, those negative perceptions do not affect the overall sentiment polarity towards the product, which stays positive.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 See, e.g., the definition appearing on Wikipedia on the page https://en.wikipedia.org/wiki/Smart_speaker
3 See the Reshape to relevance report available at https://www.accenture.com/_acnmedia/PDF-93/Accenture-Digital-Consumer-2019-Reshape-To-Relevance.pdf
4 Google announced that its Google Assistant has surpassed 1 M skills on day 1 of I/O ’19, during the presentation “Intro to Google Assistant: Build your first action,” the video can be found here)
7 Find the full report at https://www.canalys.com/newsroom/canalys-global-smart-speaker-market-2021-forecast
8 See the report at https://files.constantcontact.com/150f9af2201/d7a53ce1-0484-4f3a-ac1d-9c1622600959.pdf
10 Alexa Confidentiality and Data Handling Overview, Amazon White Paper no. 20180720, available at https://d1.awsstatic.com/product-marketing/A4B/White/%20Paper/%20-/%20Alexa/%20Privacy/%20and/%20Data/%20Handling/%20Overview.pdf
11 The report is available at https://www.ipsos.com/ipsos-mori/en-uk/almost-2-3-gb-adults-feel-concerned-about-their-online-privacy
12 See https://dataminer.io
13 The main stopword list is taken from the Snowball stemmer project in different languages and can be found at https://snowballstem.org/projects.html
14 The Snowball stemming algorithms library is available at https://snowballstem.org/algorithms/
15 See the page https://cran.r-project.org/web/packages/wordnet/wordnet.pdf for the reference description of Wordnet package in R
16 See the group’s website at https://turkunlp.org/
17 The SentimentR package is available at: https://cran.r-project.org/web/packages/sentimentr/index.html
18 The VADER package is available at: https://cran.r-project.org/web/packages/vader/index.html
Additional information
Notes on contributors
Guglielmo Maccario
Guglielmo Maccario is a PhD Candidate in Intercultural Relations and International Management at UNINT University of the International Studies of Rome. In 2018 he graduated in Economics and International Management at UNINT. In 2022 Guglielmo supported research activities of the marketing team at ISCTE Business School.
Maurizio Naldi
Maurizio Naldi graduated in Electronic Engineering (magna cum laude) at the University of Palermo and earned his PhD in Telecommunications Engineering and Microelectronics at the University of Rome Tor Vergata in 1998. He is currently a Full Professor of Computer Science at LUMSA University in Rome.