Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Research article
Insights on next-generation manufacturing of smart devices using
text analytics
Suchithra Rajendran
a
,
b
,
*
, Emily Pagel
a
a
Department of Industrial and Manufacturing Systems Engineering, University of Missouri Columbia, MO 65211, USA
b
Department of Marketing, University of Missouri Columbia, MO 65211, USA
ARTICLE INFO
Keywords:
Smart device
Next-generation manufacturing
Online customer reviews
Text analytics
Internet of Things (IoT)
SWOT analysis
Industrial Engineering
Mechanical engineering
Systems engineering
Manufacturing Engineering
Information Science
ABSTRACT
With the mass expansion in technological user-friendly products, there is an increasing demand for smart devices,
resulting in a highly competitive novel market. To ensure sustainable success, these products must remain robust
and be perceived positively by customers. With the development of Web 2.0, individuals are able to make
knowledgeable purchasing decisions, specifically with the availability of millions of online customer reviews.
Companies manufacturing smart devices can utilize this unstructured data to analyze the customers' perceptions
of their products and identify potential improvements. To the best of our knowledge, this paper is the first to
propose next-generation manufacturing insights for companies producing smart devices by determining the
current strengths, weaknesses, opportunities, and threats (SWOT) of these gadgets using text analytics. A three-
stage methodology is utilized, consisting of bigram and trigram examination, topic identification, and SWOT
analysis. After online review extraction, comments for each smart device are separated into positive, neutral, and
negative categories, based on the customer ratings. Text analytic tools are then used to determine the most
frequently occurring bigrams and trigrams to provide topics for conducting the SWOT analysis. Using the SWOT
technique results, numerous next-generation smart device manufacturing recommendations are presented.
1. Introduction
During this technological age, consumers are regularly exposed to an
abundance of innovative devices that continue to advance the boundaries
of artificial intelligence. The introduction of smart devices has facilitated
the communication and control of numerous products using the Internet
of Things (IoT). Intelligent devices provide users with an interactive
experience while assisting in basic tasks and managing their homes.
These devices can understand simple commands sent by users via Blue-
tooth or button operations to create a more efficient and autonomous
lifestyle. Consumers are inviting technology, like smart hubs and devices,
into their homes at an extremely high rate. In a study conducted by
Poushter (2016), a median of 87% reported smart device ownership
across 11 countries.
Smart devices provide customers with a user-friendly, interactive
experience in the comfort of their own home. With simple voice com-
mands, manual operation using buttons, or virtual instructions through
IoT, the user can control an entire IoT ecosystem within their home
(Lazar et al., 2015). Everyday mundane tasks, such as setting the ther-
mostat, can be performed automatically or standardized with scheduling
abilities. These smart products could add convenience to the users' daily
lives in numerous ways that extend past house operations. Smartphones
have revolutionized communication and the capabilities of the internet,
including the collaboration of smart devices through smartphone appli-
cations, which is a conventional device feature.
The diversity of smart devices in the market also proves as an
advantage for attracting a wide array of individuals. With a vast market
for these devices, customers can find a product that is most suitable for
their desired intent. Smart devices also often have customizable features
(e.g., manual or automatic control options) to accommodate a broader
market. These gadgets are created to provide an efficient experience for
clients, while utilizing the expansive capacity of the internet.
Consequently, the expanding popularity of these smart devices de-
mands a market analysis to assess and manage the existing faults of the
products. Some may be attributed to the lack of knowledge of these de-
vices, which can significantly disrupt a customer's individual experience.
This ignorance also creates a low serviceability rate for smart devices
(Chi, 2018). With new technology, it is crucial to provide adequate re-
sources for understanding the product's capabilities, which has been a
major downside for manufacturers. This weakness affects customer
* Corresponding author.
E-mail address: RajendranS@missouri.edu (S. Rajendran).
Contents lists available at ScienceDirect
Heliyon
journal homepage: www.cell.com/heliyon
https://doi.org/10.1016/j.heliyon.2020.e04491
Received 21 February 2020; Received in revised form 24 April 2020; Accepted 14 July 2020
2405-8440/©2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Heliyon 6 (2020) e04491
satisfaction and is only a single example among many others identified
through this research.
The rise in Web 2.0 tools has altered the way companies and busi-
nesses gather information on their products' performances and the con-
sumers' perceptions. Consumers can now communicate their personal
opinions regarding the quality of products to a global audience as user-
generated content persistently increases (Gao et al., 2017). The client
experience and satisfaction levels posted online circulate as viable in-
formation for potential shoppers. The increased accessibility of the
internet has provided individuals an opportunity to discuss a product's
standard of quality. Although online customer reviews (OCRs) are purely
based on specific experiences and are only a small component of the
client population, they are still considered the second most reliable
source of product information (Nielsen Holdings, 2012;Srinivas and
Rajendran, 2019). OCRs could assist the technological market in
increasing sales by providing insights into potential managerial recom-
mendations that improve their product.
In this study, we provide next-generation manufacturing recommen-
dations for smart devices by analyzing online customer reviews of several
intelligent devices and identifying the key features that people prefer and
dislike. Ten different smart devices are selected based on their price,
intended usage, review volume, and rating distribution. A three-stage
methodology is used consisting of bigram and trigram examination, topic
mining, and SWOT analysis. Based on this approach, next-generation
manufacturing recommendationsare provided for smart device companies.
The remaining paper is organized as follows. Section 2presents the
review of smart devices, online customer reviews, and next-generation
manufacturing in detail. Section 3discusses the methodology, and Sec-
tion 4describes the data in-depth. The results are presented in Section 5,
while the discussion based on the results is provided in Section 6. The
conclusion and future work are given in Section 7.
2. Literature review
2.1. Smart devices
User-friendly technology, like smart devices, may assist in everyday
mundane tasks in a customer's home or workplace. Ur et al. (2013)
analyzed the ease of using three different smart items within a home based
on three separate client's experiences. It was concluded that several fea-
tures limited the adaptability of these intelligent devices, including
network connectivity and accessibility to manual features. Although their
research identified the current weaknesses of the smart devices, it only
consisted of three customer's experiences rather than a large sample pop-
ulation, unlike thestudy mentioned in our paper. The current inadequacies
of smart products were alsodiscussed through research conductedby Lazar
et al. (2015). After several customers purchased a gadget chosen by the
researchers, information was collected regarding the advantages of using
the product and the existingchallenges. The authors found that individuals
were unable to performregular maintenance on devices dueto their lack of
product knowledge. However, their study established conclusions based
only on one smart product rather than multiple.
Although smart devices possess several weaknesses that negatively
impact a customer's satisfaction, the study by Mattern (2003) highlighted
several potential advantages of intelligent devices. Mattern (2003)
summarized many positive aspects, including their ability to communi-
cate with one another via the Internet of Things. This feature is very
marketable and encourages individuals to purchase several items rather
than one. Characteristics like these that are well perceived by consumers
may be transferrable to other products and are considered for managerial
recommendations proposed in our study.
2.2. Online customer reviews
With the growing popularity of social media websites, consumers are
now able to openly discuss about products online with acquaintances or
strangers. Erkan and Evans (2016) concluded that through the analysis of
electronic word-of-mouth (eWOM), the influence of people's conversations
on purchase intent could be examined. This influence is potentially crucial
to sales, which has caused businesses to revolutionize marketing. Alongside
the expansion of online ordering, companies are attempting to acclimate
based on positive and negative consumerperceptions evaluated viareviews
that have been posted to the internet. Fu et al. (2015) provided a better
understanding of the motivations behind a consumer's decision to partake
in eWOM communication, which has become more important as consumers
begin to place more significance on eWOM while assessing a product.
Cheung and Lee (2012) identified the key factors that motivate con-
sumers to propagate positive eWOM rather than negative feedback on
social networking sites. With over 200 samples from a consumer review
community, an empirical test was conducted to identify the multiple fac-
tors that contribute to a consumer's decision to participate in positive
eWOM. The authors concluded that reputation, sense of belonging in a
community, and the delight of helping other consumers significantly
correlated to consumers' eWOM intention. The results of their research
were also useful in understanding the behaviors and fluctuations of online
consumer-opinion platforms members. It was shown through the conclu-
sions of their study that this expanding form of communication, electronic
word-of-mouth, represents a new and imperative marketing phenomenon.
Traditional (offline) word-of-mouth has shown to play a substantial
role in customers' purchasing decisions (Balaji et al., 2016;Rajendran,
2020). The introduction of Web 2.0 has increased consumers' outlook for
assembling unbiased item details from several individuals through
eWOM rather than relying solely on a company's reputation. To further
understand the complexity and implications of eWOM, Hennig-Thurau
et al. (2004) used an online sample of 2,000 consumers to gather infor-
mation on the structure and relevance of the motives of consumers' on-
line expressions. The authors suggested that consumers' interest in social
interaction, desire for economic incentives, personal concern for other
individuals, and the potential to enhance their self-worth are the primary
factors leading to eWOM behavior. By identifying these factors, busi-
nesses can modify their marketing strategies or product design to address
the negative eWOM reactions.
2.3. Next-generation manufacturing
Manufacturing, alongside technology, has dramatically evolved and
provoked an increase in market competition. Alvi and Labib (2001)
examined the next generation of manufacturing methodologies to iden-
tify beneficial paradigms that help companies to remain robust and
competitive. The authors also classified potentially harmful market
behavior, such as introducing products to the public ahead of time and
not allowing proper time for market research. They concluded that
strategies like these often lead to a low serviceability rate that can
negatively affect a customer's experience with the product.
The management of next-generation manufacturing with technology
was analyzed by Soliman and Youssef (2001) to understand its impact
further. An example of this evolving technology is the Internet of Things
(IoT), which can serve as a communication path for devices connected to
the internet. Their study assessed the implementation of IoT within
manufacturing and found that it allowed companies to collect large
amounts of valuable data that could be used to improve their process.
Another technology that has shown to benefit next-generation
manufacturing is virtual reality (Rubio et al., 2005;Chong et al., 2018;
Karvouniari et al., 2018). Virtual reality revolutionizes process simulation
and can identify potential machine faults that have financial paybacks
(Rubio et al., 2005). These advancements in technology give manufacturing
companies the opportunity to become more proactive and sustainable.
2.4. Contributions to the literature
We contribute to the existing literature multi-folds. Although analysis
of the service sector has been widely discussed in the literature (e.g.,
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
2
Rajendran et al., 2015;Smith and Srinivas, 2019;Rajendran and Rav-
indran, 2019;Kambli et al., 2020;Srinivas, 2020), relatively fewer pa-
pers provide product examination and design recommendations.
However, several articles in the literature highlight the necessity to
consider proposing product design recommendations as these outcomes
result in insights for better manufacturing (Singh and Tucker, 2017;Kim
and Noh, 2019;Yang et al., 2019). In addition, prior research has proven
a strong influence of customer opinion on the decision-making process of
next-generation manufacturing (e.g., Gao et al., 2017;Lee, 2018;Zhu and
Pham, 2020).
To the best of our knowledge, this study is the first to analyze online
customer reviews of smart devices through text analytics and provide
next-generation manufacturing insights based on the investigation. Next,
despite several articles focusing on identifying the current weaknesses of
the smart devices, most of them either considered limited client experi-
ences (rather than a large sample population) or analyzed only the per-
formance of one device. Whereas, in this paper, we examine more than
33,000 customer reviews of numerous smart gadgets, and evaluate rat-
ings over an extended period that gives the smart gadget producers the
ability to draw many conclusions.
Moreover, this research examines multiple products that are diverse
with regard to their application, size, and price. Identifying the topics
that individuals positively and negatively perceive in each product en-
ables companies to potentially understand the key features that people
prefer and dislike in each of these devices. Even though there are very
limited papers on proposing managerial recommendations for the next
generation of manufacturing, their findings did not capture the voice of
the customers of other dissimilar products.
3. Methodology
The overview of the approach (Figure 1) displays the three main steps
within this research. We extract the reviews related to different smart
devices from several social networking sites, perform an in-depth
customer review analysis, and then propose recommendations with a
SWOT analysis. These managerial insights are based on text analytics
performed during review analysis.
The overview of methodology (Figure 2) thoroughly outlines each
step taken to complete the analysis of all customer reviews and the
composition of managerial recommendations. To accurately assess the
extracted online reviews, data pre-processing is conducted (Section 3.1),
and the separation of reviews based on a rating scale is provided in
Section 3.2. This categorization allows for more straightforward text
analytics of the bigrams and trigrams found within the positive, negative,
and neutral reviews. These bigrams and trigrams are then used to identify
probable topics from client reviews that made an impact on the satis-
faction level. After finding several topics, a SWOT analysis is conducted
to determine the current strengths, weaknesses, opportunities and threats
of each smart device. The strengths and weaknesses are attributed to
internal factors like Wi-Fi connectivity that affect customer's perception
of quality, and the opportunities and threats are external factors like a
competitive gadget's features. From the SWOT analysis, managerial rec-
ommendations are proposed to improve the overall quality of intelligent
devices under study.
3.1. Web-scraping and data pre-processing
The proposed approach will begin by using a web scraper to extract
over 33,000 publicly available online customer reviews from various
online sources. This data will then be analyzed to identify the plausible
subject matters using topic modeling. In this case, we extract the time of
the online post, the given client rating from one to five, the review, and
verification of purchase. Each review may not focus on a single aspect of
the gadget. Therefore, each feedback is separated into individual sen-
tences, and the statements are treated as independent comments.
Following the extraction, Python®is used to condense the data by
detecting and removing duplicate reviews. Subsequently, the sentences
are further pre-processed for text analytics by tokenizing, removing un-
necessary special characters and non-English words, stemming inflected
words, converting all characters to lowercase, and removing stop words.
This is accomplished through modules that are readily available in the
Figure 1. Overview of the approach.
Figure 2. Overview of the methodology.
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
3
Python®natural language toolkit (NLTK). Along with the default stop
words provided in the NLTK, we also developed a set of custom stop
words to filter unnecessary words that do not add value for the topic
identification.
3.2. Separating reviews based on rating scale
After extracting and pre-processing the data, reviews for each product
are separated based on the ratings established by the individual. These
ratings are distributed between one and five and are dependent on the
level of customer satisfaction. A rating scale is created to classify the
topics associated with a positive or negative sentiment. Every 1–2 star
rating is categorized as a negative review, a 3-rating as neutral, and 4–5
as positive. The categorization of reviews provides a clear divide in
commonly occurring positive and negative topics based on the ratings.
3.3. Bigram and trigram analysis
The following section displays the bigram and trigram analysis dis-
cussed in Jurafsky and Martin (2014).
Suppose if R1;R2;…;RNare set of any random words, the probability
that these words are occurring in a sequence (i.e., PðR1…RNÞ) is denoted
by PðRÞ, as given using Eq. (1).
PðRÞ¼PðR1…RNÞ(1)
The probability of occurrence of R2after R1in a sequence is given by
Eqs. (2) and (3). Similarly, the likelihood of occurrence of R3after R2and
R1consecutively is given by Eq. (4).Eq. (5) derives PðR1;R2;R3Þusing
constraints (3) and (4).
PðR2jR1Þ¼PðR1;R2Þ
PðR1Þ(2)
PðR1;R2Þ¼PðR2jR1ÞPðR1Þ(3)
PðR1;R2;R3Þ¼PðR3jR1;R2ÞPðR1;R2Þ(4)
PðR1;R2;R3Þ¼PðR1ÞPðR2jR1ÞPðR3jR1;R2Þ(5)
Applying this expression to Nwords, the probability of the occurrence
of RNimmediately after the word sequence R1…RN1is presented by Eqs.
(6) and (7).
PðR1…RNÞ¼PðR1ÞPðR2jR1ÞPðR3jR1;R2Þ…PðRNjR1;R2…RN1Þ(6)
PðR1…RNÞ¼ Y
N
n¼1
PrðRnjR1…Rn1Þ(7)
Applying the chain rule of conditional probability to the word se-
quences under study, Eq. (9) gives the likelihood of the occurrence of N
words in any sequence.
PðR1…RNÞ¼PðR1ÞPðR2jR1ÞPR3
R2
1…PRN
Rn1
1(8)
PðR1…RNÞ¼ Y
N
i¼1
PRN
RN1
1(9)
Alternatively, for the bigram model, the Markovian assumption pre-
viously discussed in Eq. (10) is considered.
PRN
RN1
1¼PðRNjRN1(10)
where PðRNjRN1Þis estimated using the ratio of the bigram count of
RN1and RN(represented by
ν
ðRN1RNÞ) to the sum of the frequency of
all the bigrams containing only RN1(P
r
ν
ðRN1RÞ), as provided in
Eq. (11).
PðRNjRN1Þ¼
ν
ðRN1RNÞ
Pr
ν
ðRN1RÞ(11)
Similarly, the trigram model to forecast the conditional probability of
the N
th
word in a sequence is given by constraint (12).
PðRNjRN1;RN2Þ¼
ν
ðRN2RN1RNÞ
Pr
ν
ðRN2RN1RÞ(12)
3.4. SWOT analysis
The SWOT analysis shown in Figure 3 briefly outlines the categori-
zation of the identified strengths, weaknesses, opportunities and threats
of the smart devices under study. The strengths and weaknesses of each
product are internal factors that affect the perceived quality of the
product, whereas opportunities and threats are external factors. External
factors are often attributed to the current competitors' strengths. This
analysis is used to create several managerial insights that, if imple-
mented, would enhance the quality of the applicable smart gadgets.
4. Data
In total, 33,345 reviews of ten different intelligent devices and smart
hubs posted by consumers from various online sources are extracted. The
smart devices considered are Alexa Remote, Amazon Smart Plug, Logi-
tech Harmony Hub, Nest Thermostat, Ring Alarm Kit, Samsung Smart
Things, Schlage Connect, Sengled LED Lightbulb, Sonos Speaker, and
WINK Hub.
As mentioned earlier, all online customer reviews extracted for each
product are separated into positive, negative, and neutral reviews, based
on a rating system out of five. To gather competitive intelligence, we
identify the strengths and weaknesses of numerous smart devices and
hubs and further analyze reviews pertaining to these traits. The strengths
and weaknesses of the gadgets are respectively based on customer
compliments and complaints, while the opportunities and threats are
identified based on examining the online reviews of the many other
existing competitors. After establishing the key topics, we put forth
several managerial recommendations to the manufacturing companies
that can aid them in their planning decisions to improve client satisfac-
tion, product reputation, and sales.
Table 1 represents data fields that are obtained from the online
customer reviews for all the ten smart devices under study. These fields
are chosen based on relevance to the customer's perception of the quality
and credibility of the review. The data points “rating”and “review”are
essential when analyzing the overall perceived quality of each smart
gadget. To draw conclusions regarding the intelligent device's perfor-
mances over a period of time, the “date”record is also extracted.
4.1. Data analysis
Ten different products with varying abilities, prices, and lifetime on
the market are chosen for this study. A diverse assortment of smart
gadgets introduces additional features that can potentially be integrated
with others. The selection of products is crucial in this study to provide
well-informed managerial insights. The smart devices are labeled as
SD#1 –SD#10 for the purpose of de-identification.
Table 2 displays the total number of reviews extracted for the ten
smart devices under study, as well as the total number of negative,
neutral, and positive reviews. It is clear that the majority of the reviews
are positive (i.e., 4 stars). The percentage of authentic purchases is
based on the classification of “verified purchase”displayed with each
online customer review. Figure 4 outlines the total distribution of re-
views as the percentage of negative, neutral, and positive reviews for
each smart device under study. Through this figure, it is shown that most
reviews for the gadgets under investigation are positive. Highly-rated
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
4
devices are further studied to establish all well-perceived qualities that
are relevant to other competitive products.
Figure 5 represents the average rating across the four quarters of
2017, 2018, and 2019. Based on the trends identified in this graph, it is
evident that the overall ratings decrease over time across all products.
This steady decline could be attributed to the deteriorating long-term
performance of smart devices that would have initially been highly
commended due to many other factors. Often, smart gadgets require
software updates that are meant to improve the performance of the de-
vice but may negatively affect the customer's experience. Table 3 pre-
sents the word and sentence statistics of the individual reviews for all ten
smart devices under study. While observing the standard deviation
values, it is evident that the average number of words in a sentence
Strengths Weaknesses
Internal
Smart device features that satisfy
people’s expectations and lead to
positive OCRs (e.g., versatilit y)
Qualities and identities of the smart
product that lack satisfaction among
customers (e.g., poor connectivity)
Opportunities Threats
External
Potential changes that the gadget can
adopt to enhance reputation (e.g.,
smartphone application capability)
Difficulties that the product faces due to
compet itors (e.g. , ot her device’s ab ility
to connect with varying Wi-Fi
frequencie s)
Figure 3. SWOT analysis framework.
Table 1. Extracted data description.
Field Explanation
Date Contains the date that review is posted online by the individual
Title Contains a title of the client review, and in general, the overall feeling
towards the product
Rating Contains numerical value given by the customer regarding the
product's perceived quality
Review Contains the full review posted by individuals
Verified
Purchase
States whether the review is posted by a customer whose purchase is
verified by website
Table 2. Review distribution.
Product Number of Reviews % of Authentic Reviews
Positive (>¼4) Negative (<¼2) Neutral (¼3) Total
SD#1 1056 460 187 1703 82.90
SD#2 1657 748 267 2672 91.54
SD#3 3166 740 206 4112 97.01
SD#4 766 643 123 1532 84.01
SD#5 3432 439 120 3991 99.97
SD#6 4909 231 635 5775 98.07
SD#7 3294 878 220 4392 99.54
SD#8 4976 452 226 5654 62.86
SD#9 1369 423 221 2013 97.98
SD#10 1291 150 60 1501 85.69
Figure 4. Distribution of reviews.
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
5
remains almost the same for all smart items. In contrast, the average
sentences in each review fluctuate across products, with a range of more
than four.
5. Results
5.1. Topics summary
Table 4 presents the topics identified for both positive and negative
reviews for all ten smart products. These topics are inferred through the
most common bigrams and trigrams found within the positive and
negative online customer reviews.
5.2. SWOT analysis
The Fishbone diagram (Figure 6) outlines the five leading causes of
customer dissatisfaction with smart gadgets that are considered in this
study. Through text analytics, it appears that the capability, durability,
connectivity, size dimensions, and serviceability of the products are
mostly negatively perceived by customers. Problems with connectivity
are most commonly associated with Wi-Fi and unreliable connection
with other smart devices over IoT. Serviceability also poses issues for
individuals due to the lack of technical support and product specialists.
Some intelligent devices under study are newly introduced products, and
hence, there is a lack of knowledge of their long-term performance,
which negatively affects some customers' experience with the item. Many
negative reviews possess content on the size dimensions of several
products, which often cause mobility problems. Another factor of client
dissatisfaction is the product's capabilities, including automatic control,
sensor detection, and a lack of manual control. The device's durability
also impacts customer satisfaction. In addition, people complain of
inefficient battery life and short product relevance life.
The link analysis (see Figure 7) is created as a visual representation of
the most commonly identified topics within positive customer reviews.
These words are most relevant to the positive qualities of the smart de-
vices under study. The size of the bubble directly correlates to the fre-
quency of the topic within the individual reviews. From this figure, it is
shown that versatility, installation, features, responsiveness, QR code
availability, and sound quality all positively impact the customers'
perception of quality of the gadgets under study. People compliment
various features of the products like a microphone and Bluetooth options,
which can be easily integrated into other smart products. The respon-
siveness of the voice control feature is also highly praised throughout the
online reviews. Several of these devices contain speaker systems that
have a high sound quality, as well.
After completing text analytics, we conduct the overall SWOT anal-
ysis for all the devices under study, as presented in Figure 8. The iden-
tified deficiencies violate several quality control standards, which
negatively influence the reputation of the product. The following con-
clusions concerning the eight dimensions of quality are made after
completion of the SWOT analysis.
Customers often complain about the smart device's unreliable
connection to Wi-Fi routers, which may pose many issues in regards to
product performance. This unreliability challenges the eighth dimension
of quality –conformance to standards. Smart products often require a
stable Wi-Fi connection to fulfill all their functions, so any interruption of
this connection results in the nonconformance of standards and a nega-
tive perception of its quality. For example, some people experience
connectivity and functioning issues with their smart door locks, due to
which they are forced to hire a locksmith to enter into their house or
break open the door, resulting in significant collateral damages.
Customers also criticize the insufficient battery life of smart items,
which defies the third dimension of quality –durability. The smart
device's inability to withstand time creates a negative perception of its
strength. By increasing battery life, people would become more satisfied
and relay their satisfaction via eWOM.
The incapability to move the device conveniently after installation
pose problems for thousands of customers. Individuals regularly express
their grievances about the size and weight of the product. If an individual
chooses to relocate the device for any reason, their perception of its
quality may be dependent on ease of reinstallation, so it is a critical
aspect to consider when designing the product.
Another common complaint among clients is inadequate technical
support for smart gadgets, which challenges the fourth dimension of
quality - serviceability. Since smart devices are newly introduced to the
technology market, sufficient technical support often does not exist. If
these device manufacturers aspire to have success longevity, adequate
support must be offered to customers in various forms to assure
accessibility.
The last major weakness identified through the SWOT analysis is the
lack of manual control, which violates the sixth dimension of quality –
features. Although these devices are often meant to be fully automated,
manual control is also expected by people for ease of use. By offering this
additional feature, individuals can choose between automatic and
manual depending on the desired outcome, which could positively
impact their experience with the device and improve market appeal.
6. Discussions and managerial implications
Candi et al. (2017) categorized the product design aspects under three
dimensions –functional, aesthetic, and symbolic. While the functional
Figure 5. Average rating over time.
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
6
aspect focusses on whether the product does the intended task, the aes-
thetics is centered around the beauty of the device. The symbolic char-
acteristic emphasizes on the self-image and value perceived by the
customer as a result of using the gadget. The results obtained under the
current study could also be categorized under one of these three di-
mensions. For instance, the topics “connectivity”and “capability”could
be classified under product functionality, “size dimension”is related to
the aesthetics, and “features”are associated with the symbolic attribute.
Hu and Liu (2004) suggested that online customer reviews descrip-
tion can be related to two aspects of a product; explicit and implicit.
However, studies centered around the explicit feature portrayal of
products are more extensive as they are easy to extract from the un-
structured online data (Gao et al., 2017). This research contributes to the
current body of the literature by analyzing both dimensions of the smart
devices. Even though factors, such as price, shape, and vendors, have
been identified as key topics in prior studies (e.g., Candi et al., 2017;Gao
et al., 2017;Singh and Tucker, 2017;Jiang et al., 2019), the reviews
particularly associated with smart devices, focusses on features, dura-
bility, connectivity, and issues with automation.
Upon analysis, certain dimensions of quality are not currently satis-
fied by the products under study. Although automation is growing in
popularity, many consumers are unfamiliar and uncomfortable with its
capabilities. By providing the consumers with more manual control of
scheduling, security preferences, and/or temperature fluctuations, they
will feel more comfortable allowing devices to make changes in their
homes when they are away. Smart products, like smart thermostats, can
change the temperature of the user's home without command, which may
lead to concern. Allowing consumers to choose the automatic or manual
Table 3. Word and sentence statistics.
Product Average Words per Sentence Average Sentences per Review
Mean STD Mean STD
SD#1 4.61 0.02 6.46 5.01
SD#2 5.27 0.02 5.21 5.02
SD#3 4.35 0.02 4.54 3.05
SD#4 4.51 0.03 5.01 4.01
SD#5 4.31 0.02 2.36 2.04
SD#6 4.39 0.04 4.32 3.16
SD#7 4.52 0.02 6.32 5.93
SD#8 4.62 0.02 6.74 4.12
SD#9 4.79 0.03 4.12 3.31
SD#10 4.41 0.05 5.27 4.13
Table 4. Summary of positive and negative topics.
Product Topics from Positive Reviews Topics from Negative Reviews
SD#1 IR blaster, capability of smartphone application, setup Technical support, connectivity with smart device, cable box, large dimension
SD#2 Connectivity with other smart devices, setup, features, QR code Product durability, automatic control, technical support, motion sensor
SD#3 Bluetooth, installation, connectivity to other smart hub and smart devices Reliability, customer service, battery life, factory reset, large
SD#4 Setup, responsiveness, versatility, sound quality Technical support, connectivity with smart devices, product specialist, heavy
SD#5 Microphone, connectivity with speakers, voice control, responsiveness Battery life, customer service, automatic setting
SD#6 Installation, voice control, QR code scanning, accessories Connectivity with Wi-Fi, customer service, responsiveness
SD#7 Temperature control, installation, money-saving, versatility Customer service, connectivity with Wi-Fi, common wire –power source
SD#8 Sound quality, installation, accessories, security system, motion sensor Customer service, security system, disarming alarm, extended ringing
SD#9 QR code, installation, price, connectivity with smart hubs Customer service, connectivity with Wi-Fi, product specialist, motion sensor
SD#10 Setup, sound quality, Bluetooth connectivity, features Technical support, access to online music libraries, heavy, automatic control, battery life
Figure 6. Fish bone diagram.
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
7
state of the device might build people's confidence in the device's
capabilities.
The product's characteristic of frequently disconnecting from the
internet posed a concern to many individuals across every device in this
study and results in thousands of negative ratings and reviews. The un-
reliability of this connection can affect the device's performance and
dependability. For example, when owners are traveling, and their smart
security system disconnects from the internet, they are unable to monitor
their homes remotely. A solution would be to connect multiple routers
within a home to increase Wi-Fi connections or install a hot spot
connection within the device. If the device could consistently connect to
the internet, its performance would improve.
To become more competitive, smart devices could increase versatility
and connect to routers with different wireless frequencies, like 2.4 GHz
and 5 GHz. This additional versatility will expand the product's market
and result in increased sales. Moreover, a smart gadget that is compatible
with multiple frequencies is more attractive to customers who have
varying expectations, constraints, and needs.
Within this study, it is found that many consumers respond well to the
smartphone application feature of certain products. This feature could be
utilized virtually by any device connected to the internet. With the
increased popularity of smartphones, an application gives consumers the
freedom to control the devices from anywhere with limited effort. The
smartphone application could also display data provided by the smart
gadget, such as the current temperature of the home, security footage, or
device status.
Many consumers complain about the low durability of products and
insufficient battery life. With moderate to high prices, consumers expect
a certain quality of products, so providing a long-lasting battery is
necessary. Improving battery life will result in an increased life span, and
therefore, could improve the customers' perception of its quality.
As new household smart devices emerge into the market, products
could become more compatible with them and offer grouping and
scheduling options. Through IoT, these devices could schedule on and off
states to maximize energy efficiency within a home. This is especially
useful with lighting and temperature, but has other applications as well.
The field of analyzing OCR for proposing next-generation
manufacturing insights is still in the developmental stage. Although
prior articles emphasize the need to consider OCRs for making product
design recommendations (e.g., Candi et al., 2017;Gao et al., 2017;Jiang
et al., 2019), manufacturing companies do need to explicitly specify the
next-generation design changes that have been made as a result of the
examining the OCRs. Discussing the implementation of the
next-generation design insights would be helpful for academicians for
validating their OCR analysis results.
7. Conclusions
Prior studies have repeatedly suggested that online customer reviews
significantly affect an individual's purchasing decisions. eWOM also al-
lows companies to understand further their customer's needs, prefer-
ences, and dislikes. To the best of our knowledge, this study is the first to
provide next-generation manufacturing recommendations for existing
Figure 7. Positive link analysis.
Figure 8. Summary of SWOT analysis.
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
8
smart devices through online review analysis. We develop a three-stage
methodology, consisting of bigram and trigram examination, topic ide-
nitification, and SWOT analysis. Based on this approach, next-generation
manufacturing recommendations are provided for companies making
smart devices.
To begin this study, 33,345 online customer reviews for ten diverse
smart gadgets are collected from several online sources. The reviews are
then separated depending on star rating, and positive and negative topics
are inferred through commonly occurring bigrams and trigrams. Upon
completion, several meaningful topics are concluded, and a SWOT
analysis is conducted to highlight the key factors that influence customer
satisfaction. Numerous managerial recommendations with respect to
durability, additional features, versatility, connectivity, and manual
control are presented.
Declarations
Author contribution statement
Suchithra Rajendran: Conceived and designed the experiments; Per-
formed the experiments; Analyzed and interpreted the data; Contributed
reagents, materials, analysis tools or data; Wrote the paper.
Emily Pagel: Performed the experiments; Analyzed and interpreted
the data; Wrote the paper.
Funding statement
This research did not receive any specific grant from funding agencies
in the public, commercial, or not-for-profit sectors.
Competing interest statement
The authors declare no conflict of interest.
Additional information
No additional information is available for this paper.
Acknowledgements
The authors would like to thank the reviewers for their valuable
feedback that helped us to improve the earlier versions of the paper.
References
Alvi, A.U., Labib, A.W., 2001. Selecting next-generation manufacturing paradigms—an
analytic hierarchy process based criticality analysis. Proc. Inst. Mech. Eng. Part B J.
Eng. Manufact. 215 (12), 1773–1786.
Balaji, M.S., Khong, K.W., Chong, A.Y.L., 2016. Determinants of negative word-of-mouth
communication using social networking sites. Inf. Manag. 53 (4), 528–540.
Candi, M., Jae, H., Makarem, S., Mohan, M., 2017. Consumer responses to functional,
aesthetic and symbolic product design in online reviews. J. Bus. Res. 81, 31–39.
Cheung, C.M., Lee, M.K., 2012. What drives consumers to spread electronic word of
mouth in online consumer-opinion platforms. Decis. Support Syst. 53 (1), 218–225.
Chi, M.H., 2018, March. Smart IC technologies for smart devices in IoT applications. In:
2018 China Semiconductor Technology International Conference (CSTIC). IEEE,
pp. 1–4.
Chong, L., Ramakrishna, S., Singh, S., 2018. A review of digital manufacturing-based
hybrid additive manufacturing processes. Int. J. Adv. Manuf. Technol. 95 (5–8),
2281–2300.
Erkan, I., Evans, C., 2016. The influence of eWOM in social media on consumers' purchase
intentions: an extended approach to information adoption. Comput. Hum. Behav. 61,
47–55.
Fu, J.R., Ju, P.H., Hsu, C.W., 2015. Understanding why consumers engage in electronic
word-of-mouth communication: perspectives from theory of planned behavior and
justice theory. Electron. Commer. Res. Appl. 14 (6), 616–630.
Gao, H.Y., Tian, W.E.I., Yang, S.X., 2017. Features Mining and Analysis of Smart Bracelet
Based on Online Reviews. DEStech Transactions on Computer Science and
Engineering (smce).
Hennig-Thurau, T., Marchand, A., Marx, P., Gwinner, Kevin P., Walsh, Gianfranco,
Gremler, Dwayne D., 2004. Electronic word-of-mouth via consumer-opinion
platforms: what motivates consumers to articulate themselves on the internet?
J. Interact. Market. 18 (1), 38–52.
Hu, M., Liu, B., 2004, July. Mining opinion features in customer reviews. AAAI 4 (4),
755–760.
Jiang, H., Kwong, C.K., Kremer, G.O., Park, W.Y., 2019. Dynamic modelling of customer
preferences for product design using DENFIS and opinion mining. Adv. Eng. Inf. 42,
100969.
Jurafsky, D., Martin, J.H., 2014. Speech and Language Processing, 3. Pearson, London.
Kambli, A., Sinha, A.A., Srinivas, S., 2020. Improving campus dining operations using
capacity and queue management: a simulation-based case study. J. Hospit. Tourism
Manag. 43, 62–70.
Karvouniari, A., Michalos, G., Dimitropoulos, N., Makris, S., 2018. An approach for
exoskeleton integration in manufacturing lines using Virtual Reality techniques.
Procedia CIRP 78, 103–108.
Kim, H.S., Noh, Y., 2019. Elicitation of design factors through big data analysis of online
customer reviews for washing machines. J. Mech. Sci. Technol. 33 (6), 2785–2795.
Lazar, A., Koehler, C., Tanenbaum, J., Nguyen, D.H., 2015, September. Why we use and
abandon smart devices. In: Proceedings of the 2015 ACM International Joint
Conference on Pervasive and Ubiquitous Computing. ACM, pp. 635–646.
Lee, H.J., 2018. A ghost in the shell? Influences of AI features on product evaluations of
smart speakers with customer reviews. J. Inf. Technol. Services 17 (2), 191–205.
Mattern, F., 2003, April. From smart devices to smart everyday objects. In: Proceedings of
Smart Objects Conference, pp. 15–16.
Nielsen Holdings, N.V., 2012. Global consumers' trust in 'Earned' advertising grows in
importance | Nielsen. Retrieved November 12, 2017, from. http://www.nielsen.com/
us/en/press-room/2012/nielsen-global-consumers-trust-in-earned-advertising-g
rows.html.
Poushter, J., 2016. Smartphone ownership and internet usage continues to climb in
emerging economies. Pew Res. Center 22, 1–44.
Rajendran, S., Srinivas, S., Saha, C., 2015. Analysis of operations of port using
mathematical and simulation modelling. Int. J. Logist. Syst. Manag. 20 (3), 325–347.
Rajendran, S., Ravindran, A.R., 2019. Inventory management of platelets along blood
supply chain to minimize wastage and shortage. Comput. Ind. Eng. 130, 714–730.
Rajendran, S., 2020. Improving the performance of global courier &delivery services
industry by analyzing the voice of customers and employees using text analytics. Int.
J. Logist. Res. Appl. 1–21.
Rubio, E.M., Sanz, A., Sebasti
an, M.A., 2005. Virtual reality applications for the next-
generation manufacturing. Int. J. Comput. Integr. Manuf. 18 (7), 601–609.
Singh, A., Tucker, C.S., 2017. A machine learning approach to product review
disambiguation based on function, form and behavior classification. Decis. Support
Syst. 97, 81–91.
Smith, D., Srinivas, S., 2019. A simulation-based evaluation of warehouse check-in
strategies for improving inbound logistics operations. Simul. Model. Practice Th. 94,
303–320.
Soliman, F., Youssef, M., 2001. The impact of some recent developments in e-business on
the management of next generation manufacturing. Int. J. Oper. Prod. Manag. 21,
538–564.
Srinivas, S., 2020. A machine learning-based approach for predicting patient punctuality
in ambulatory care centers. Int. J. Environ. Res. Publ. Health 17 (10).
Srinivas, S., Rajendran, S., 2019. Topic-based knowledge mining of online student
reviews for strategic planning in universities. Comput. Ind. Eng. 128, 974–984.
Ur, B., Jung, J., Schechter, S., 2013. The current state of access control for smart devices
in homes. In: Workshop on Home Usable Privacy and Security (HUPS). HUPS 2014.
Yang, B., Liu, Y., Liang, Y., Tang, M., 2019. Exploiting user experience from online
customer reviews for product design. Int. J. Inf. Manag. 46, 173–186.
Zhu, M., Pham, H., 2020. An empirical study of factor identification in smart health-
monitoring wearable device. IEEE Trans. Comput. Soc. Syst. 7 (2), 404–416.
S. Rajendran, E. Pagel Heliyon 6 (2020) e04491
9
