Online Consumers’ Opinions Analysis for Marketing Strategy

Evaluation

Elena Ektoros

, Andreas Gregoriades

and Michael Georgiades

Cyprus University of Technology, Limassol, Cyprus

Primetel PLC, Limassol, Cyprus

Keywords: Micro-blogs, Tweeter Analytics, Product Positioning, Marketing Strategy.

Abstract: With over two billion users having access to social media accounts, people increasingly choose to express

themselves online. Electronic word of mouth generates large amounts of data, making it a valuable source for

big data analytics. This provides organisations with key capabilities for improved decision-making through

mining insights directly from online sources. In this work we gathered and analysed the sentiment of

consumers’ tweets regarding the release of two smartphone products. Tweeter data was collected using a

custom-made Android application. The research question addressed in this study focused on whether the

marketing positioning strategy of the company under investigation was successful after the release of two of

its new products. To evaluate this, we compared the product positioning strategy of the firm before and after

the release of the product. Consumers’ opinions were analysed to identify possible discrepancies between

planned consumers’ reactions and sentiments, as strategized by the company, and how these were altered with

the release of the product.

1 INTRODUCTION

A new challenge for companies is how to discover

hidden information in big data sources to remain

competitive advantage through effective data

processing (Khade, A. A., 2016). Micro-blogs

represent one type of big data that is available for

analysis and complements other forms of human

interaction (Chamlertwat, W., et al., 2012). They

gained popularity from the inherent need of people to

express their views on a wider scale. The internet

empowers consumers to gain access to information

sources, enabling them to be “active co-producers of

value” (deChernatony 2000) and sometimes referred

as prosumers. Consumers give their opinion freely

through micro-blogs and “reviews”. Micro-blogs are

viewed as an electronic word-of- mouth (eWOM)

which could trigger discussions on products or

services. Companies realised the potential from

eWOM analysis and use micro-blogging as a part of

their marketing strategy (Jansen et al., 2009).

This work aims to analyse whether Huawei’s

product release, positioning and marketing strategy

for the Huawei P20 and Huawei P20 Pro products was

successful by evaluating tweets before and after their

release. To answer this research question, we utilised

public opinion data obtained via a custom-made

Twitter collection tool.

2 LITERATURE REVIEW

To improve the process of new product release that

aim to satisfy new customers’ needs, it is important to

evaluate alternative ways to retrieve information

regarding customers’ opinions that could highlight

those needs (Jung, J. J., 2012). According to a

statement made by Scott Cook, co-founder of Intuit,

“A brand is no longer what we tell a customer it is- it

is what customers tell each other it is” (Nayab G et al,

2016). Twitter and other social networks became

valuable resource for mining sentiment in fields such

as customer behaviour. Around 20% of microblogs

mention a brand name (Jansen, B. J., et al., 2009),

hence, companies should include in their marketing

strategy the management of brand perception on

Twitter and other social media platforms. Several

studies investigate the use of Twitter and other social

networks to mine consumer-sentiment in field as

customer behaviour.

266

Ektoros, E., Gregoriades, A. and Georgiades, M.

Online Consumers’ Opinions Analysis for Marketing Strategy Evaluation.

DOI: 10.5220/0007838802660273

In Proceedings of the 21st International Conference on Enterprise Information Systems (ICEIS 2019), pages 266-273

ISBN: 978-989-758-372-8

The role of online communities, particularly in the

context of new product development, has been

discussed by many studies (e.g., Franke and Piller,

2003). Pang (2007) highlighted that online product

reviews enable marketers and manufacturers to gain

more complete understandings of customers. Zhu and

Zhang (2010) proved that online customer reviews

can be a good proxy for communicating customer

experience by word-of-mouth.

In order to achieve successful product

development, product positioning is a critical

technique to help firms better understand the

underlying relationships between product features,

competing alternatives and diverse consumers’ needs

(Lilien and Rangaswamy 2003; Petiot and Grognet

2006; Cha, et al., 2009). Product positioning is not

what companies do to a product but is what

companies do to prospective customers.

In strategic marketing, ‘positioning’ refers to

implementing a set of tactics to ensure that a product

and its characteristics occupy a unique position in the

minds of customers (Lilien and Rangaswamy 2003).

Optimal product positioning corresponds to

determining which attributes and configurations

should be used to satisfy customer requirements

(Kwong, Luo, and Tang 2011). According to (Chih-

Hsuan W., 2015) product positioning is implemented

through a series of steps: (1) visualising competitive

alternatives and product features (2) constructing a

forecasting model to estimate how potential buyers

will react to marketing stimulus and (3) specifying

optimal position of new product(s) and identifying

niche segment(s). A perceptual map is a powerful tool

to visualise the relationships between competing

products and their associated features in a

comprehensible way. Moreover, perceptual maps can

help firms assess the strengths and weaknesses of

competing alternatives. According to (Hair J. et al,

2009), two ways are mainly used to construct a

perceptual map: the first is multi-dimensional scaling

and the second is correspondence analysis. The latter

can use categorical variables to project on a plot the

relationships between benchmarks and associated

product features, without requiring multiple

regression. Due to its simplicity correspondence

analysis is adopted in this study to evaluate Huawei’s

P20 & P20 pro products’ positioning.

Traditional approaches for studying consumer

behaviour, such as marketing survey, interviews,

focus groups, experiments and other, require a great

amount of time and resources. Moreover, some

products, such as smartphones, have a short-term

product life cycle, hence this approach could not be

appropriate due to the time required to perform these

analyses and the lifespan of the product.

Smartphone’s market competition is fierce due to the

constant advancement of technology, hence new

updated versions are hitting the market at lightening

speeds. According to HTC, the average shelf life for

smartphones has decreased from three years in 2007

to around six to nine months in 2011 (Ferreira, 2011).

Another important statistic from Statista.com shows

that the average number of months for people to

change their smartphones is less than two years (≈22

months). Therefore, producers have too little time to

research market by traditional way. Referring to

Technology Adoption Life Cycle, each model of

smartphones has limited time to prove their product

adoption using conventional means.

Twitter represent a useful platform for micro-blog

analysis due to the amount of data that is generated

from the public (Chamlertwat W. et al., 2012). Due to

these platforms, sentiment analysis and topic

extraction have been very hot research fields recently.

These capabilities make it possible to automatically

identify user’s emotions regarding a subject (Al-

Obeidat F. et al., 2018). Therefore, these methods are

more appropriate in fast changing domains such as

the smartphone market.

3 METHODOLOGY

The diagram in figure 1 illustrates the steps followed

to answer our research questions. The first step in the

method was to examine Huawei’s marketing strategy

for both P20 products and their associated position-

ing. An important element of the marketing strategy

was the target customer segment and the keywords

for promoting particular features of new products.

Figure 1: Methodology.

Online Consumers’ Opinions Analysis for Marketing Strategy Evaluation

267

The second step of the methodology concentrated

on the building of an application to dynamically

collect data regarding the target customers of the

company. In this stage an android application was

developed to gather data based on specific

geolocation coordinates of the microblogs and

specific theme.

The next step in the methodology was data pre-

processing, which involved data cleaning,

dimensionality reduction and irrelevant data

elimination. This was a necessary step to enable the

processing of the data and the extraction of

meaningful insights.

Following data pre-processing, the sentiment and

statistical analysis on the dataset was performed. For

these tasks open source tools were used. Specifically,

a frequency analysis of the main themes in the dataset

and polarity of sentiment analysis were performed.

In the final step, the obtained quantitative and

qualitative data was analysed further to map the

observed data against the planned company’s

strategy. During this step, the results obtained were

compared with planned positioning to examine

whether the marketing strategy of the firm matched

the opinions of its customers. Therefore, identifying

if consumers perceived positively the features of the

new products and their quality.

4 STRATEGY ANALYSIS

The analysis of Huawei’s marketing strategy was

necessary to examine if the consumers’ perception of

the new products was analogous to the expectations

of the company prior to the smartphones release.

Based on the conducted analysis, it seems that

Huawei adopted a product and cost differentiation

strategy for the promotion of Huawei P20 and

Huawei P20 Pro. The company aimed to target

consumers that look for high specification

smartphone, usually provided by iPhone, at a lower

price. Based on this, Huawei’s marketing strategy

promoted the unique and competitive features of

photography that Huawei Pro series smartphones has.

The company tried to emphasise on the quality of the

camera resolution in order to differentiate itself in

their target market and use specific keywords to

emphasise on this feature. Samsung and iPhone

constitute Huawei’s major competitors, with, iPhone

as the leading brand owned 19,2% market share on

smartphone sales and Samsung second with 18,4%.

Huawei concentrated its efforts on the triple-camera

feature introduced in the smartphone market for first

time, utilising intelligent photography, hence

differentiating Huawei from the other brands in

smartphone industry. As part of their promotion the

firm uses the "See Mooore" slogan clearly referring

to the Leica triple-lens system presented in all

advertisements that the smartphone has.

The assumption made here is that if the company

manages to satisfy the needs of the market segment

they targeted, this will be reflected in the eWOM of

its consumers. In this study we explore the effect of

the strategy on eWOM, before and after the release of

the products.

5 THE DATA GATHERING TOOL

A data gathering tool was used to collect relevant

tweets for the analysis. A custom Twitter application

was developed for this process for Android mobiles.

To have access to Twitter’s APIs, a mobile app using

Android studio was built. This is a typical procedure

required by Twitter to secure access to Tweeters

database by authorised users. In order to have the

right data from the microblogs, the corresponding

data entries of interest were specified in each query.

i.e. data for sentiment and statistical analysis. The

best feature that Twitter offers for this purpose is the

"Search Tweets". Twitter offers two options to

retrieve tweets. The Rest API and the Streaming API.

For this task the Rest API was used, that allowed for

searching of specific tweets using certain criteria

specified in the queries searched for. The timeframe

of the data gathering was 7 days due to API

limitations. Figure 2 shows a screen shot of the

developed Android app.

The specification of the tweets query keywords

was based on preliminary analysis of the two products

features and company positioning and strategy. For

the two smartphones, terms such as #HuaweiP20,

#HuaweiP20Pro, #Huawei, #SeeMore etc. were used.

These keywords yielded 436 tweets with 140

characters length each. The #SeeMore and other

related keywords related to the products’ positioning

strategy, with emphasis on being perceived by

consumers as a product equipped with improved

mobile photography. This was the key feature of the

firm’s marketing campaign.

Tweets were restricted to English language, and

the target market for this study was the United

Kingdom, since UK is one of the top 10 countries

with active Twitter users worldwide according to

Statista.com. Four main cities in the UK were

targeted: London, Birmingham, Liverpool and

Manchester.

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

268

Figure 2: Screenshot of the developed tweet collector

Android application retrieving tweets for keyword

“Huawei” in London area for a specific date.

The data collection was held between 27th of

March 2018 and 11th of April 2018. On the 27th of

March the Huawei smartphones were official

introduced in an event that took place in Paris. This

date was treated as a key milestone in classifying the

data in two categories: before and after product

release.

In order to address the problem of customers

referring to product features using different terms, it

was necessary to create an ontology to enable the

grouping of similar terms under different categories.

This was manually developed based on products’

features categories obtained from the Huawei’s

website.

6 DATA PRE-PROCESSING

Filtering of tweets was a necessary step prior to the

analysis, to eliminate useless metadata information,

and keep tweets as simple format. The filtering

criteria for the Tweets were: inclusion of related

keywords in tweets without information of author’s

username and other irrelevant punctuation symbols.

The evaluated tweets were all in the English

language.

The data gathering tool downloaded tweets in txt

file-format, categorised based on their location and

hashtag keywords. Tweets were organised in sixteen

folds for all four UK cities. For each city the tweets

were organised in four categories based on their

keyword’s hashtags.

The following process describes the data

processing performed over each of the folders

mentioned above as well as on the whole set of files.

A custom tweet cleaner program was specified as a

batch file executed from Windows command prompt

and whose purpose was to consolidate all the *.txt

files placed within the same folder into a single

merged .txt file (named merged.txt).

Figure 3: Example Tweet after cleansing from irrelevant

content.

Following the file merging procedure, the data

was processed to include one tweet per line for better

text manipulation and that included the elimination of

irrelevant information such as characters, usernames,

links, protocols. An example pre-process tweet is

depicted in figure 3.

7 ANALYSIS

The keywords frequency analysis was performed

using an opensource tool that disclaims automatically

unknown symbols (i.e @, #, emojies), words in other

languages and other misleading information. This

process was repeated for each of the 16 files (4 cities

and four keywords).

Table 1: Ranking of the top 10 words that users mentioned

most frequently.

A/A

Word

Percentage

huaweip20

9.29

pro

7.32

huawei

5.63

p20

3.93

seemoore

2.24

paris

2.24

new

2.01

seemore

1.74

smartphone

1.33

unboxing

1.03

The first analysis was conducted using the

WriteWords.com and yielded 1142 unique words in

all tweets. Most words appeared in tweets more than

once, so the overall sum of the words was 7581.

Words were subsequently classified according to a

number of themes. For instance, words that were

relevant to the Huawei brand included keywords such

as: Huawei P20, Pro, P20, HuaweiMobile etc.

Online Consumers’ Opinions Analysis for Marketing Strategy Evaluation

269

Similarly, words that were referring to the phone’s

camera were filtered based on keywords:

photography, triple-camera, Leica and so on.

The top 10 most popular words, as mentioned by

users in their tweets are listed in Table 1. The word

“HuaweiP20” was mentioned 9,29% of the time and

the word “pro” 7.32%. The words “seemore” and

“seemooore”, were ranked 5

and 6

positions, and

referred to the slogan used during the promotion of

the products. The word Paris ranked in the top ten and

denoted the place of the product release. This

however had no relevance to how consumers

perceived the new product and hence was ignored

during the analysis.

Overall, 1174 out of the 4070 words were directly

related to Huawei. Manchester emerged as the city

with the highest number of relevant tweets. This

could be attributed to the fact that it has the highest

proportion of student population among the rest of the

cities that were examined.

Some of the words were written in different

spelling, such as “top” and “toop”, so we considered

these as identical. Each of the words were categorised

as either positive, negative or neutral. This was

performed using the Opinion Lexicon, which

constitutes a list of positive and negative words or

sentiments, compiled by (Hu and Liu, KDD-2004).

All tweets have been evaluated against their polarity

with 28 classified as positive, 10 negative and 12

neutral. The results from this analysis could indicate

that the Huawei P20 release, triggered a positive

reaction by consumers as indicated from the ratio of

the positive over negative words.

Huawei’s strategy as has been evaluated and

analysed in previous section is mainly based on

product-differentiation. Hence, its marketing plan

was to promote the differentiating feature of their

products. In the P20 and P20pro smartphones this was

the triple-camera. From the tweets’ analysis, forty

unique words were detected in the dataset that

described the camera in a positive way. While, from

the overall tweets mentioning the camera, 15%

referred positively to the triple-camera feature.

7.1 Sentiment Analysis

Sentiment analysis (SA) was a necessary component

of this work. Considering the sentiment analysis

methods available, this work assessed tweets that

reflect public opinion regarding the new products and

how this changed with time. SA literature (Pang and

Lee, 2008) provided methodologies from other

studies that utilised information from public opinions

to extract their polarity and content categorisation

(Gamon et ad., 2005).

SA is used to extract static data patterns and

discover dynamic trends (Lei, N. and Moon, S.K.,

2015) based on the emotional state of the author of a

text. SA is a useful for research into online opinions

due to their ability to automatically measure emotion

in online texts. SA includes algorithms to

automatically detect sentiment in text (Pang and Lee,

2008). Certain algorithms assign an overall polarity

to a text while others identify topics that users

discussed along with the polarity of the sentiment

expressed in topics (Gamon et ad., 2005). Three

common sentiment analysis approaches are: Machine

Learning approaches, Lexicon-based methods and

Linguistic Analysis techniques.

A Machine Learning (ML) approach utilises

machine learning techniques and statistics. According

to (Witten et al.,2016), training texts annotated by

human coders in terms of polarity are used to train an

algorithm to detect features that associate with the

three positive, negative or neutral emotions. The

trained algorithm can then look for the same features

in new texts to predict their polarity (Thelwall et al.,

2011). Sets of words, are also used for the algorithm

training.

Figure 4: Sentiment analysis for the keywords:

“HuaweiP20Pro” (bottom) and “HuaweiP20Pro Camera”

visualised with the sentiment Viz tool.

Positive sentiment

Negative sentiment

Positive sentiment

Negative sentiment

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

270

The Lexicon Approach is divided into two further

techniques. The first one is the dictionary technique

and the second is the corpus-based approach. It starts

with lists of words that are pre-coded for polarity and

strength, and are used as prior knowledge to evaluate

the overall polarity of a paragraph based on

occurrences of these know words (Thelwall et al.,

2011)

The linguistic analysis method exploits the

grammatical structure of text to predict its polarity,

using a lexicon. For example, linguistic algorithms

identify context and idioms as part of the polarity

prediction process (Thelwall et al., 2011).

From the above three techniques in the field of

Sentiment Analysis we selected the ML approach

since it produces more accurate results. There are two

basic ML methods that can be used for polarity

classification, which are considered as the most

efficient and simple to use:

 Naïve Bayes classification

 Support Vector machines

For these techniques it is important to train the

classifier using supervised learning techniques. A set

of training datasets should be used with cases that

already have been assigned as negative, positive or

neutral. In our case we used the tweeter sentiment

corpus.

The Naïve Bayes Model is more commonly used

in cases when we examine the polarity of a sentence

by recognising the polarity of each word individually.

The support vector machines is better for large texts

(phrases) or combinations of words. However, since

in Twitter, we deal with short texts, we initially

decided to use the first model.

Despite its popularity, SA and its techniques have

been criticised of their accuracy and specifically with

regards to the detection of polarity of plain text

contents. For example, in certain domain the accuracy

of sentiment analysis was lower than 50% with less

precision in detecting negative sentiments (Jongeling

et al, 2015). To escape from this problem, we

concentrated on positive and negative words alone, as

they appear in tweets and referred to the product

under study. This process identification was

performed automatically. Plus, we constrained the

search criteria to specific keywords and hence

eliminated irrelevant tweets that could have

influenced the results.

To validate our results with other methods, we

used the SentimentViz tool to obtain the overall

sentiment of consumers with regards to Huawei’s P20

for the same period of analysis. Figure 4 illustrates the

overall sentiment for the “HuaweiP20Pro” keyword.

This represents the distribution of the keywords

across the sentiment scale. The tweets in Figure 4 lean

towards the positive side of the spectrum indicating

that consumers perceived positively the new product

and its differentiating features. Similarly, the camera

keyword shows a positive sentiment as indicated also

in figure 4. Both observations confirm the results

performed from our preliminary analysis. Each tweet

is shown as a circle positioned by sentiment, an

estimate of the emotion contained in the tweet's text.

Circles correspond to tweets. Negative tweets are

drawn as blue circles on the left, and positive tweets

as green circles on the right.

8 RESULTS

From the analysis of the data, we identified

preliminary evidence which could indicate that

Huawei’s marketing strategy had a positive impact on

their targeted users. The product positioning plan of

the firm was to differentiate on price and

technological features such as the use of the Leica

triple-camera. Preliminary results showedthat

consumers received well the new triple camera

feature of P20 and referred to it in their tweets with

positive connotations. The new phones triggered

people’s interest as it was also witnessed by Google

Trends search queries in Germany, Spain and Italy for

the same period, that could also indicate an increase

in reputation.

Secondary data regarding company’s sales

performance, verified our preliminary results

regarding the success of the company’s positioning

strategy by targeting consumers and addressing their

needs. Both products were positively received by

consumers and the trend of this effect was not

declining after the first couple of weeks when

consumers had a chance to experiment with the

products. The overall opinion of users’ tweets

analyzed in this paper was positive throughout the

specified time-frame and the ratio of positive over

negative tweets was relatively constant.

This exercise showed that Twitter could be a

valuable tool for predicting and analyzing costumer

behaviour on product release. People through social

media express their views about their experiences

with products. Sentiment Analysis is an important

tool for textual data investigation and marketing

managers should include these capabilities in their

portfolio of tools during analysis of public opinion.

However, analysis of big data should be executed

with consistency and accuracy to produce useful

results.

Online Consumers’ Opinions Analysis for Marketing Strategy Evaluation

271

9 CONCLUSIONS

Product positioning is key in targeting the right

consumers. Commercial organisations continuously

monitor their product positioning by gathering data

online and offline. Products mispositioning however,

could jeopardise the marketing strategy of a company

and should be avoided. Validating strategies early in

the product release cycle constitute a vital process for

effective sales performance. Therefore, companies in

addition to other information sources, should also

utilise data from the blogosphere to understand

customers’ opinions in real time and accordingly

respond to their needs (Al-Obeidat, F., Spencer, B.

and Kafeza, E., 2018). Data mining can help

enterprises resolve marketing issues and improve

product positioning through quicker analysis of

online consumers opinions.

This work presented a technique for evaluating

product positioning using eWOM analysis. An

application of eWOM analysis was also presented, for

the marketing strategy of two Huawei smartphones.

Limitations of this work lie in the small sample size

which concentrated on specific geographical regions.

For future work the authors are considering

expanding on the methodology to evaluate the impact

of marketing strategies using more sophisticated

sentiment analysis techniques with less false positives

and false negatives rates and hence require less

manual evaluation.

REFERENCES

Jung, J. J., 2008. Taxonomy alignment for interoperability

between heterogeneous virtual organizations. Expert

Systems with Applications, 34(4), pp.2721-2731.de

Chernatony, L. (2000). Succeeding with brands on the

Internet. Brand Management, 8(3), 186-195.

Forbes, 2018 (https://www.forbes.com/companies/twitter/)

Retrieved 10 April 2018

Statista, 2018 (https://www.statista.com/statistics/282087/

number-of-monthly-active-twitter-users/) Retrieved 13

April 2018

Huawei, 2018 (http://www.huawei.com/en/about-huawei/

corporate-information/milestone) Retrieved 14 April

2018

Sagolla, D. (2009). 140 characters: A style guide for the

short form. Hoboken: Wiley.

Jansen, B. J., Zhang, M., Sobel, K., & Chowdhury, A.

(2009). Twitter power: Tweets as electronic word of

mouth. Journal of the American Society for Information

Science, 60(11), 2169-2188.

Zhang, M., Jansen, B. J. and Chowdhury, A., 2011.

Business engagement on Twitter: a path analysis.

Electronic Markets, 21(3), p.161.

Wauters, R. (2009). Twitter spawned 50,000 apps to date,

will open up firehose for more. Retrieved January 6,

2010, from http://www. techcrunch.com/2009/12/09/

twitter-le-web-2009/.

Keller, E. (2007). Unleashing the power of word of mouth:

Creating brand advocacy to drive growth. Journal of

Advertising Research, 47(4), 448–452.

Chamlertwat, W., Bhattarakosol, P., Rungkasiri, T. and

Haruechaiyasak, C., 2012. Discovering Consumer

Insight from Twitter via Sentiment Analysis. J. UCS,

18(8), pp. 973-992.

Lei, N. and Moon, S. K., 2015. A Decision Support System

for market-driven product positioning and design.

Decision Support Systems, 69, pp.82-91.

Jung, J. J., 2012. Evolutionary approach for semantic-based

query sampling in large-scale information sources.

Information Sciences, 182(1), pp.30-39.

Ferreira, A., 2011. Android OS changes smartphone life

cycle. The Vista, February.

Mostafa, M.M., 2013. More than words: Social networks’

text mining for consumer brand sentiments. Expert

Systems with Applications, 40(10), pp.4241-4251.

Olobatuyi, M. E., 2006. A user’s guide to path analysis.

Lanham: University Press of America, Inc. p.32.

Thelwall, M., Buckley, K. and Paltoglou, G., 2011. Senti-

ment in Twitter events. Journal of the Association for

Information Science and Technology, 62(2), pp.406-418.

Huang, J., Thornton, K. M. and Efthimiadis, E. N., 2010,

June. Conversational tagging in twitter. In Proceedings

of the 21st ACM conference on Hypertext and

hypermedia (pp. 173-178). ACM.

Honey, C. and Herring, S. C., 2009, January. Beyond

microblogging: Conversation and collaboration via

Twitter. In System Sciences, 2009. HICSS'09. 42nd

Hawaii International Conference on (pp. 1-10). IEEE.

Franke, N. and Piller, F. T., 2003. Key research issues in

user interaction with user toolkits in a mass

customisation system. International Journal of

Technology Management, 26(5-6), pp.578-599.

Zhu, F. and Zhang, X., 2010. Impact of online consumer

reviews on sales: The moderating role of product and

consumer characteristics. Journal of marketing, 74(2),

pp.133-148.

Lilien, G. L. and Rangaswamy, A., 2003. New Product and

Brand Management: Marketing Engineering

Applications. Prentice Hall.

Boyd, D., Golder, S. and Lotan, G., 2010, January. Tweet,

tweet, retweet: Conversational aspects of retweeting on

twitter. In System Sciences (HICSS), 2010 43rd Hawaii

International Conference on (pp. 1-10). IEEE.

Petiot, J. F. and Grognet, S., 2006. Product design: a vectors

field-based approach for preference modelling. Journal

of engineering design, 17(03), pp.217-233.

Kwong, C. K., Luo, X. G. and Tang, J. F., 2011. A

methodology for optimal product positioning with

engineering constraints consideration. International

Journal of Production Economics, 132(1), pp.93-100.

Pang, B. and Lee, L., 2008. Opinion mining and sentiment

analysis. Foundations and Trends® in Information

Retrieval, 2(1–2), pp.1-135.

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

272

Gamon, M., Aue, A., Corston-Oliver, S., & Ringger, E.

2005. Pulse: Mining customer opinions from free text.

Lecture Notes in Computer Science, 3646, 121–132.

Witten, I. H., Frank, E., Hall, M. A. and Pal, C. J., 2016.

Data Mining: Practical machine learning tools and

techniques. Morgan Kaufmann.

Kim, S. M. and Hovy, E., 2004, August. Determining the

sentiment of opinions. In Proceedings of the 20th

international conference on Computational Linguistics

(p. 1367). Association for Computational Linguistics.

Martinčić-Ipšić, S., Močibob, E. and Perc, M., 2017. Link

prediction on Twitter. PloS one, 12(7), p.e0181079.

Khade, A.A., 2016. Performing customer behavior analysis

using big data analytics. Procedia computer science,

79, pp.986-992.

Jongeling R, S. Datta, and A. Serebrenik. Choosing your

weapons: On sentiment analysis tools for software

engineering research. In Proceedings of the

International Conference on Software Maintenance

and Evolution, pages 531–535, 2015.

Nayab G , Bilal M , Shrafat A, 2016, “A brand is no longer

what we tell the customer it is - it is What customers tell

each other it is: Validation from lahore, Pakistan,

Sci.Int. (Lahore), 28(3),2757-2762.

Chih-Hsuan W, 2015, A market-oriented approach to

accomplish product positioning and product

recommendation for smart phones and wearable

devices, International Journal of Production Research,

53:8, 2542-2553.

Hair, J. F., Black, W. C.,Babin, B. J., & Anderson, R. E.

2009. Multivariate data analysis (7th Ed.). Upper

Saddle River, NJ: Prentice Hall.

Online Consumers’ Opinions Analysis for Marketing Strategy Evaluation

273