Tourism Management Perspectives 31 (2019) 323–331

Contents lists available at ScienceDirect

Tourism Management Perspectives

journal homepage: www.elsevier.com/locate/tmp

Facilitating tourists' decision making through open data analyses: A novel T

recommender system
⁎
Eleonora Pantanoa, Constantinos-Vasilios Priporasb, Nikolaos Stylosa, , Charles Dennisc
a
Department of Management, School of Economics, Finance and Management, University of Bristol, The Priory Road Complex, Priory Road, Clifton, Bristol BS8 1TU, UK
b
Department of Marketing, Branding & Tourism, Middlesex University, Business School, W107 Williams Building, Hendon Campus, The Burroughs, London NW4 4BT,
England, UK
c
Department of Marketing, Branding & Tourism, Middlesex University, Business School, Williams Building, Hendon Campus, The Burroughs, London NW4 4BT, England,
UK

A R T I C LE I N FO A B S T R A C T

Keywords: A number of studies have recently been published reporting researchers' efforts to create new, more efficient
Recommender system recommender systems to support tourists' decision making. This current research operationalizes a recommender
Tourist decision making process system by filtering user-generated data that is abundantly available online, based on individuals' evaluation
Consumer information processing criteria, to produce a dataset for analysis. Drawing upon an array of predictive models, this research proposes a
Classifier systems
new recommender system able to facilitate the tourist decision making process through successful managing of
Open data analysis
open data. It further presents a rating estimation method using ratings that pertain to online users-specified
criteria (profile). The model is able to predict consumers' ratings of a certain product with high reliability
starting from open data on their profiles.

1. Introduction particular, this technology may influence consumers' behaviours by

providing digital tools for searching, comparing and buying products
The latest advances in information technology are changing the (Kang, Mun, & Johnson, 2015). Hence, an insightful and interactive
practice of marketing (Quinn, Dibb, Simkin, Canhoto, & Analogbei, digital space created by consumers (user-generated content) is freely
2016). Specifically, continuous advances in digital technology, via provided, offering consumers recommendations and prompting them to
evolutionary hardware capabilities (e.g. powerful lightweight personal purchase particular products, services, and brands.
computers, broadband networks), sophisticated software applications This novel technological enrichment provides innovative pervasive
(e.g. social network platforms) and the rapid diffusion of smart tech- spaces for supporting consumers' access to information and dis-
nologies in the vast range of global population, have dramatically af- semination of consumers' knowledge (including experience, opinions,
fected consumers' behaviours (Maity & Dass, 2014; Pantano & Priporas, reviews, etc.) via social networking services (SNSs) (Balaji, Khong, &
2016). Chong, 2016). Therefore, internet channels are shifting towards online
Many researchers have acknowledged consumers' tendencies to networks characterized by users' interactive knowledge generation and
search for innovative experiences and technologies (Arts, Frambach, & sharing. These digital spaces offer substantial open data that can be
Bijmolt, 2011; Lowe & Alpert, 2015; Pantano & Viassone, 2014). As a really beneficial to the public and consumers, in particular. However,
consequence, innovation and technology adoption theories have been natural limitations in the human brain's ability to process large volumes
exploited in marketing research to predict consumers' usage of parti- of information packages, make the selection stage in the consumer
cular technology (Al-Quesi, Dennis, Alamanos, & Jayawardhena, 2014; proposition acquisition process a very challenging task that can lead to
Davis, 1989; Dennis, Jayawardhena, & Merrilees, 2009; Ng, 2016; information overload, non-optimized resource management (e.g.
Venkatesh, Thong, & Xu, 2012), experience (Blazquez, 2014; Dennis, wasted time, money and/or effort), misguided decision making pro-
Brakus, Gupta, & Alamanos, 2014; Verhoef et al., 2009), and purchase cesses, physical exhaustion and distress or to a combination of these
behaviour (Alnawas & Aburub, 2016; Pantano & Priporas, 2016). In repercussions (Colace, De Santo, Greco, Moscato, & Picariello, 2015;

⁎
Corresponding author at: Department of Management, School of Economics, Finance and Management, University of Bristol, The Priory Road Complex, Priory
Road, Clifton, Bristol BS8 1TU, UK
E-mail addresses: e.pantano@bristol.ac.uk (E. Pantano), C.Priporas@mdx.ac.uk (C.-V. Priporas), n.stylos@bristol.ac.uk (N. Stylos),
C.Dennis@mdx.ac.uk (C. Dennis).

https://doi.org/10.1016/j.tmp.2019.06.003
Received 2 November 2018; Received in revised form 15 April 2019
2211-9736/ © 2019 Elsevier Ltd. All rights reserved.
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

Willemsen & Johnson, 2011). To address these issues in the tourism Hall, & Pal, 2016).
industry, recommender systems analyse, delineate and mirror tourists' Abundant information available on the Internet increases con-
characteristics to create meaningful associations of human profiles and sumers' awareness of product specifications and availability of alter-
respective needs by leveraging the power of open databases (Oliveira, natives yet amplifies the difficulty of making a choice out of a propo-
Delgado, & Assaife, 2017; Pantano, Priporas, & Stylos, 2017). In fact, it sition set (Bonhard & Sasse, 2006). Thus, consumers try to find ways to
has been reported that conduct of open data analyses improve matching filter internet resources to decrease information overload and at the
of available product propositions to tourists' needs (Okazaki, Díaz- same time make appropriate choices (Chen, 2008; Li & Du, 2011). This
Martín, Rozano, & Menéndez-Benito, 2015; Zhang, Zhou, Kehoe, & process is a challenging one though, because internet users exchange
Kilic, 2016). This approach that can potentially generate useful opinions with others that may not belong in their personal social net-
knowledge for both consumers' and companies' decision-making in work of contacts, thus raising concerns about strangers' credibility
general, could also lead to happy tourists and improved business per- (Cheung & Thadani, 2012). Hence, source credibility reflects the ex-
formance (Liu & Shih, 2005; Nguyen & Cao, 2015). pertise and trustworthiness of the message source and the extent to
The aim of this study is to explore the effectiveness of a novel re- which this source could be perceived as believable and competent (Chu
commender system algorithm for successfully predicting tourists' pre- & Kamal, 2008; Davis & Khazanchi, 2008). Recommendations coming
ferred choices in their decision making. To this end, this research is from trusted sources and friends in particular, greatly contribute to
grounded on information gain theory data mining (Liao, Chu, & Hsiao, loosening social e-shopping-related constraints (Harris & Dennis, 2011).
2012; Sudheep, Idikkula, & Alexander, 2011) and proposes an advanced Moreover, it is suggested that the number of customer reviews of a
recommender system based on a rating estimation method through product plays a key role in motivating consumers to engage with e-
ratings pertaining to users-specified criteria (profile). shopping via retail portals and social media.
Theoretically, the paper investigates the growing phenomenon of Human cognitive capacity is finite, and individuals can process only
technology as facilitator of consumer information processing and de- part of the plethora of information transmitted by media and other
cision making, by offering an application in the tourism field, i.e. sources. Consumers therefore can retain and respond to only a subset of
tourism destinations. From a practical viewpoint, it demonstrates how the total data received, according to individual criteria (Johnson,
marketing professionals can exploit open data analysis to influence Bellman, & Lohse, 2003). Hence, recommendations included in product
tourists' decision-making processes. Thus, the study makes also a reviews are said to considerably facilitate the selection process (Park,
methodological contribution which may be of interest to the re- Kim, Choi, & Kim, 2012; Zhang, Zhou, & Zhang, 2010). Influential
searchers involved in marketing and management decision-making communications no longer derive primarily from advertising but from a
(quantitative analysis) and to tourism scholars who are interested in variety of consumer-focused media activities (Lawlor, Dunne, &
enhancing their arsenal of computational tools while researching Rowley, 2016). Online recommendation sources range from reviews
tourists' behaviour. published in online portals to customized recommendations created by
The remainder of the paper is organised in three main sections. electronic decision-making aids, namely recommender systems. In fact,
First, it focuses on the consumer information and decision-making recommender systems are so influential to consumers that researchers
process, followed by a review of actual recommender systems employed have suggested they should also be extensively utilized to validate ex-
for supporting tourists. This part also introduces predictive models, also perts' opinions, thus supporting opinion formers, and opinion leaders in
known as Support Vector Machines (SVM) (Suykens & Vandewalle, their recommendations to consumers (Adomavicius & Tuzhilin, 2001;
1999) as employed in open data analysis and novel recommender sys- García-Crespo, López-Cuadrado, Colomo-Palacios, González-Carrasco,
tems development. Second, a case study example of the tourism sector & Ruiz-Mezcua, 2011).
is provided to illustrate the usefulness of the proposed recommender Current progress in digital information technologies is offering a
system. Finally, implications are discussed to illustrate the significance wide range of systems able to support consumers in information pro-
of the findings for academics and practitioners, respectively. cessing, thus influencing their purchase choices, for example, holiday
planning. The tourism industry is widely adopting intelligent digital
2. Theoretical background recommender systems to assist the large number of online platforms in
tourism industries (e.g. Tripadvisor, Trivago, booking.com). This takes
2.1. Consumer information processing and decision making place in order to provide more pertinent and centred information, and,
eventually, enriched tourism experiences, by means of web-mining,
Consumer generated content (CGC) and social networks are the context-aware systems, and autonomous agents (Gretzel, 2011).
building blocks of social commerce (Amblee & Bui, 2011). During the
last few years, social networks have been dramatically changing mar- 2.2. Recommender systems
keting. On one hand social networks have become one of the leading
information sources for consumers (Davis & Khazanchi, 2008), and on Recommenders as a stimulating marketing technology offer im-
the other, they are fast-integrating e-commerce services to sell products portant value to both consumers and firms. Recommenders assist con-
directly to internet users (Senecal & Nantel, 2004). In this context, sumers in learning about products/services through large choice sets,
electronic word-of mouth (e-WOM) facilitates consumers' desires to whilst at the same time, benefit firms by converting browsers to buyers,
share opinions and experiences in order to help others in their pur- promoting cross-selling and increasing loyalty by providing a custom
chases, as well as for personal prestige (Park & Kim, 2008). E-WOM browsing experience (Lee & Hosanagar, 2019). Recommender systems
makes the online exchange of opinions an easy and cost-effective pro- have been developed for various industries to tackle the problem of
cess. It overcomes the obstacles encountered in traditional word-of- information overload (Sun, Guo, & Zhu, 2019). A recommender system
mouth through scalability, speed of diffusion, multi-way asynchronous consists of personalized information-filtering technology, able to filter a
information exchange, and measurability of the format and quantity of set of items based on consumers' preferences, and thus predict a pos-
information released (Cheung & Thadani, 2012; Park & Kim, 2008). sible preference (Ghazanfar, 2015; Pantano et al., 2017). The effec-
Information theory, which is mainly based on mathematics, statis- tiveness of recommender systems derives from the ability to learn
tics and information engineering provides the theoretical under- consumers' favourites by analysing their past behaviour responses. In
pinnings of classificatory systems and machine learning (MacKay, other words, the system learns what consumers prefer starting from
2003). Furthermore, machine learning supports decision analysis and their previous choices (machine learning) and identifies future pre-
predictive modelling through decision trees and data mining, with main ferences (Borras, Moreno, & Valls, 2014; Noguera, Barranco, Segura, &
aim to better reach and represent choices and decisions (Witten, Frank, Martinez, 2012; Park et al., 2012). Indeed, these systems can be

324
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

exploited to provide additional suggestions about places to visit or Starting from a certain set of examples for training purposes (labelled
products of interest dynamically, by gathering real time data on the by a Boolean value 1/0, in/out, true/false, etc.), an SVM training al-
state of the consumers (e.g. location) and adapting accordingly (context- gorithm renders a model that supports true/false as outcomes. In par-
awareness) (Borras et al., 2014). These features characterize intelligent ticular, an SVM can be considered as a mathematical representation of
and autonomous agents able to (i) analyse the behaviour of a user, (ii) the examples of the set as single points in space, plotted in order to
learn automatically his/her preferences, and (iii) provide advice (re- make the Boolean examples (i.e. true and false) divided into two as
commendations) according to the consumer's profile characteristics distant as possible areas in the space. Other instances are further added
(Borras et al., 2014). into that space, while the SVM predicts the value as true or false ac-
Recommender systems are classified by their applications of colla- cording to the specific area in the space they fall. Additionally, SVMs
borative, context-aware and hybrid methods (Ghazanfar, 2015). Col- may perform a non-linear classification efficiently using the kernel
laborative systems consider simultaneously the given consumer interest trick, thus mapping their inputs into high-dimensional feature spaces.
profile with the profiles of other consumers with similar interests (Yang, Formally, a support vector machine builds a hyper-plane, which can
Cheng, & Dia, 2008). The main limit of collaborative recommender be used for classification, regression, or other tasks. Intuitively, a good
systems concerns their need for more information on users to make separation is achieved by the hyper-plane that has the largest distance
recommendations, thus their efficiency is limited in the cases of new to the nearest training-data point of the true and false classes.
users or new items (He, Parra, & Verbert, 2016). Context aware systems The first step consists of the definition of D, as the set of data used to
adapt the possible recommendation to users' current states (contextual train the system:
situation), by requiring the proper match between user preferences and
D = {(→
x i , yi ) | →
x i ∈ Rn, yi ∈ {true, false} } ; (1)
contextual factors (Baltrunas, Ludwig, Peer, & Ricci, 2012). They often
use obtrusive methods to collect contextual factors on consumers' actual →
Since any hyper-plane can be written as the set of points x i sa-
states (He et al., 2016). A further evolution of these recommender tisfying the equation:
systems is based on the emotion-aware concept, which exploits emo- ⎯→
⎯ ⋅→
tions as contextual factors by collecting data from consumers shared w x − b = 0; (2)
online via tweets, posts, etc. (Narducci, De Gemmis, & Lops, 2015). SVM estimates the maximum-margin hyper-plane able to divide the
Finally, hybrid systems combine the collaborative and content aware space in two separate regions distinguishing the points having yi = true
filtering approaches (Al-Hassan, Lu, & Lu, 2015; Shinde & Kulkarni, from those having yi = false (when the training data are linearly se-
2012). parable).
Recommendations based on consumer reviews and users' profiles The hyper-planes can be further defined as:
are more influential than promotional content created by marketers,
⎯→
⎯ ⋅→
w x −b=1
due to the higher credibility of the former compared to the latter
⎯→
⎯ ⋅→
w x − b = −1,
(Cheung & Thadani, 2012; Harris & Dennis, 2011). Scholars and prac- (3) (4)
titioners face a challenge to predict information that meets users' re-
The distance between the two emerging hyper-planes is expressed as
quirements, representing a critical issue for data extraction techniques. ⎯ ‖ (margin); thus, minimizing ‖⎯→
2/‖⎯→
w ⎯ ‖, the system maximizes the
w
Going one step further, recommender systems have the ability to ana-
distance. Two conditions have to be further set to avoid obtaining any
lyse the preferences of online users. Based on the profile characteristics
point into the margin (in other words, to clearly separate the two re-
of existing and potential users they may suggest appropriate choices to
gions in the space):
potential users (Felfernig, Jeran, Ninaus, Reinfrank, & Reiterer, 2013).
Thus, electronic recommender systems contribute to enhancing the ⎯→
w⎯ ⋅→
x − b ≥ 1 for →
x belonging to the"true"class
trustworthiness of recommendations (Bobadilla, Ortega, Hernando, & ⎯→
⎯w ⋅→
x − b ≤ −1 for →x belonging to the"false"class,
Gutiérrez, 2013). However, the integration of location information,
which is being used in some recommender systems, in conjunction with if yi = true = +1 and yi = false = −1, the condition to satisfy is:
real-time user-centric data drawn from mobile smart devices, offers
yi (⎯→
⎯ ⋅→
w x − b) ≥ 1, (5)
opportunities for advanced recommendation capabilities (Knijnenburg,
Willemsen, Gantner, Soncu, & Newell, 2012). The above-mentioned model is an example of SVM working for
Boolean classes, but it can be extended to easily predict more than two
2.3. Predictive models classes of values, in other words this example predicts a Boolean vari-
able, but it can be improved to predict more values. To achieve this
Predictive modelling is a process using data mining techniques to goal, the new SVM (multiclass SVM) is reduced into multiple binary
identify a causal relationship between a “dependent” variable (target) classification problems (Duan & Keerthi, 2005). In particular, this
and “independent” variables, based on the formulation of a statistical process is based on two main methods: (i) the development of a ma-
model. Thus, the aim of the predictive model is to predict the future chine that distinguishes between the value of one variable and all the
values of the dependent variables, drawing upon the past values. other (one-versus-all), in this case the classifier with the highest output
When numerical datasets are available, literature suggests several function assigns the class to all the new instances; and (ii) the devel-
established predictive models; for instance, neural networks (NN) opment of a machine that distinguishes between pair of classes (one-
(Horikawa, Furuhashi, & Uchikawa, 1992), k-nearest neighbour (k-NN) versus-one), in this case, each classifier assigns the instance to one of
(Weinberger & Saul, 2009), support vector machines (SVM) (Chapelle, the two categories, then the vote for the assigned class is increased by
Haffner, & Vapnik, 1999), and so on. The selection of the best predictive one vote, and finally the categories with the most votes determines the
method is considered a challenge for scholars and practitioners, re- instance classification.
quiring deep analysis of the actual context, the analysis of the nature of Mathematica™ is a powerful software package that has the ability to
the data (i.e. strings, numbers, etc.), and the computational cost in generate the classification function and predict the value of the de-
terms of capability of performing the data and time for executing the pendent variable by automatically choosing the most appropriate
task (for instance in some cases the huge volume of data cannot be methods through an internal algorithm according to the expected out-
analysed by a traditional computer and requires higher-performances comes. In particular, this function allows the evaluation of a huge
machines). variety of data sets, including numerical, textual, sound, image and a
In particular, Support Vector Machines analyse data and identify combination of all (Mathematica, 2015). In this case Mathematica™
clear patterns, which can be effectively used for classification analyses. serves as the computational platform for data analysis and prediction

325
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

Fig. 1. An overview of the predictive (recommender) system for selecting the product that best match users' requirements. (Source: The current authors).

purposes. In the next section, the methodological steps for this study are consisting an 18-bit string (which means that the cardinality of the set is
illustrated on a step-by-step basis. 18, as 18 is the number of possible topics identifying a certain tourist on
TripAdvisor, in other words n = 218). Thus, our predictive system aims
at identifying the 218 values of the target; Fig. 3 represents the table of
3. Methodology
rules (where the blank shape represents a value of “0” and the black one
the value of “1”).
3.1. Case study of new recommender systems for the tourism sector
TripAdvisor allows users to rate each attraction with stars (from 0 to
5). For convenience, we can give to each attraction the value 1 if
As aforementioned, digital recommender systems can analyse open
tourists assigned 5 stars and 0 if they gave 0 stars, if considering
data in order to select the information that might better fit consumers'
S = {0,1} as the two possible values of the tourist attraction, the
preferences (through the prediction model), thus recommending pro-
function f assigning to each data a value of 0 or 1 will be:
ducts with a higher likelihood to fit into consumers' interests. A new
recommender system based on open data analysis is illustrated in Fig. 1. f: I → S (6)
In this new recommender system, data from different sources are
processed and converted into a format well-matched with the predictor (x1, x 2⋯, x1) ∈ I → S ∈ S (7)
system. Data are also cleaned, normalized and optimized in advance to where xi, ∀i=1,⋯,18.
make the learning and prediction processes faster and even more ac- Thus, the task of the system is to identify the function describing
curate. Once the learning phase completes and a fresh dataset from new these relationships, in other words the goal of the predictive system is
users is acquired, the system carries on with suggesting a fitting holiday to define the 218 target values to predict the values of the element of I
destination via its predictive algorithm. The prediction is based on the based on the characteristics of the individual TripAdvisor users).
principle that users with similar needs, interests and requirements To implement this theoretical framework in the Disneyland case,
would probably choose similar products, thus the prediction will be Mathematica™ software read the dataset, and then built the set of rules
more accurate if initial data used for learning are also accurate. for the training set consisting of 250 items of review data, linking the
In the tourism destinations area, a relevant application could be input data with the expected results. Finally, it built for x times the
based on the selection of open data from TripAdvisor to predict the prediction function, and compared the findings applying the result
extent to which a tourist will like a certain destination. In our case, a emerging from the prediction function with the target value of all the
famous tourism attraction was selected, i.e. Disneyland Park in Paris. data in the sample.
This was originally opened as ‘Euro Disney’ theme park in Paris on April The experiment ran based on the development of 200 classifying
12, 1992. It is part of the wider Disneyland complex in Paris that also functions. In other words, the system used 200 classifying functions to
includes the Walt Disney Studios Park and seven Disney owned hotels. validate the SVM model by determining the successful cases for the
The whole Disney complex extends over approximately 5510 acres classifying function as proposed by Mathematica. The reliability of the
area. Some 320 million individuals have visited Disneyland Park since classifier machine can be further evaluated through the sum of per-
opening, with an annual attendance of more than 10 million visitors centage values of the cases in which the system identified properly the
(DisneyNews, 2017). A theme park is a useful context for a case study. value of 1 and 0 were determined. The proportion of successful cases in
In particular, Disneyland theme parks are known world-wide and at- which the classifier machine identified properly the value of 1 is 0.69,
tract visitors of a wide range of ages, with most of them being fairly while the proportion of success of 0 is 0.58, which lead to a total value
well-educated (Geissler & Rucks, 2011; Toyoda, 2014). of 1.268 (Fig. 4).
Fig. 5 graphically shows the extent to which our classifier machine
3.2. Data analysis predicts properly the value (prevision) of the known data (target)
(Fig. 5) by considering 20 random results, where the blank shape re-
In order to test our recommender systems' ability to identify a trend presents the value of “0” and the black one the value of “1”.
in visitor's evaluations of their lived experiences in Disneyland Park
(Paris), two datasets were formed: one set of 263 Trip Advisor users 4. Research findings
expressing a totally positive (5 stars, excellent) and another one con-
sisting of 263 different users providing totally negative evaluations (0 Utilizing equations/functions 1 to 7 we reach the graph appearing
stars, terrible) of the chosen attraction, based on their online profiles. in Fig. 4. The values exceed 1.12 in only two cases, while the best
The user profiles offer information on 18 topics (Fig. 2 shows the 18 prediction appears on the 184th case, as the proportion of successful
categories; each user needs to choose at least three to create a Tri- identification of 1 is 0.77 and the successful identification of 0 is 0.44,
pAdvisor profile). for a global value of 1.214. The reliability of the estimated values
For convenience, we chose to represent these elements in binary equals 0.707. To achieve this result, Mathematica adopted the Random
mode: a value equal to 1 is an indication of interest to the specific topic, Forest method. A Random Forest is a particular approach that generates
0 otherwise. Similarly, a value equal to 0 represents a negative eva- different Decision Trees by randomly cutting out sub-samples of the
luation, whereas 1 a positive one (as suggested by Pantano et al., 2017). data observations and sub-sections of the data variables, and then al-
In mathematical terms, we consider I as the set of instances lowing the emerging decision tree models to converge on the better

326
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

Fig. 2. The 18 possible characteristics defining a tourist to be chosen to create a profile on TripAdvisor.

Hall, 2001).
The subsequent evaluation of the confusion matrix (error matrix)
allows visualizing the performance of the emerging supervised learning
machine, where each row represents the instances in the predicted
class, and each column denotes the instances in the actual class (Fig. 6).
From this matrix, the reliability value is 0.607.
The proposed recommender algorithm is thus able to predict con-
sumers' preferences for a certain product (in this case potential visita-
tion of a tourism attraction/destination), based upon the open data
freely accessible via SNSs like TripAdvisor. The results show that the
proposed method has a high predictive power and increases the accu-
racy of recommendations, and thus might usefully support tourists
decision-making process. While literature studies on recommender
systems are mainly based on the proposal of systems with a higher
prediction accuracy, our system would be oriented to provide more
precise solutions starting from new input variables (i.e. consumers' fa-
cial expression and related emotion) that requires a limited time of
performance. In this way, our results reinforce the value of re-
commender systems in terms of value and time complexity reduction, as
solicited by recent researches (Bag, Kumar, Awasthi, & Tiwari, 2019).
Hence, the appropriate destinations can be recommended to those
tourists with those sets of interests.

5. Discussion and conclusion

Technology and its applications in informatics consist an area of

rapid developments during the last two decades (Chen, Zhang, Hu,
Taleb, & Sheng, 2015). It has been widely recognized that the emer-
gence of social media platforms, which was matched by the expansion
of the smartphones and tablets market (Fotiadis & Stylos, 2017), has
greatly influenced consumers' priorities with regards to receiving,
Fig. 3. Table of rules for the classifier function (aiming at identifying the 218 storing, processing and evaluating information (Pantano & Priporas,
target values). 2016). Consumers' needs for efficient data management and effective
individual decision making resulted to the introduction of software
solution (in this case, the system predicts from Breiman-Cutler en- applications for recommendations (Gavalas, Konstantopoulos,
semble of decision trees). Recent approaches to decision trees (as Mastakas, & Pantziou, 2014).
random forests) allow for multi-class classification, as well as ordinal The introduction of recommender systems in web commerce sites is
multi-class classification, prediction (Cardoso & Costa, 2007; Frank & promoted by the combination of remarkable increases in computers'
calculating power, volume decrease of computer devices and new

327
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

Fig. 4. Experiment results based on building 200 classifying functions.

Fig. 6. Confusion matrix.

(Khusro, Ali, & Ullah, 2016).

Although past studies emphasize the need for supporting consumers'
decision-making utilizing recommender systems (Barragáns-Martínez,
Fig. 5. Table of rules for 20 random results. Costa-Montenegro, & Juncal-Martínez, 2015), researchers have argued
that personalized online applications in various consumer markets are
software capabilities, which creates new opportunities for delivering scarce (Wang, 2015; Yeh & Cheng, 2015). The present paper proposes a
and customizing information on demand (i.e. cloud computing) new recommender algorithm for tourists by exploiting open data. It
(Hashem et al., 2015). Recommender systems facilitate consumers in responds to the call of Gretzel (2011), who highlights the need for more
high-quality decision-making buying processes in e-commerce settings, efficient recommender systems in delivering highly customized re-
since the options are more tailored to their needs and preferences, less commendations to online platform users with several tourism-related
time consuming, require less effort and thus, potentially result in higher choices (i.e. from the choice of a restaurant, to the choice of a gift, to a
user satisfaction (Farokhi, Vahid, Nilashi, & Ibrahim, 2016; Scholz, book, to an activity for the weekend with friends, etc.). In particular,
Dorner, Schryen, & Benlian, 2017). The interest on behalf of academics this paper presents an estimation method that is based on ratings per-
and practitioners in this area continues unabated due to growing user taining to users-specified criteria (profile) and, ultimately, suggesting a
demands, despite the numerous research and development advances product with a major consumer adoption likelihood.

328
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

In summary, this proposed novel recommender system is able to: 1) consumers built on open data analytics, which to the authors' knowl-
extract tourist profiles via TripAdvisor (including the 18 above men- edge, is novel in the tourism marketing literature. In particular, it
tioned topics) to produce a dataset that can be processed by the SVM; 2) presents a rating estimation method using ratings that pertain to users-
collect knowledge concerning tourism destinations via TripAdvisor; 3) specified criteria (profile) to suggest a product with a major likelihood
convert heterogeneous data, such as images, texts, and videos that can of consumers purchase. User behavior is complex and dynamic, since
be processed in order to run via the predictive system; 4) be properly consumers' preferences can change over time (Ansari, Essegaier, &
trained; and 5) create a representation of users' data as points in space; Kohli, 2000; Mobasher, Cooley, & Srivastava, 2000; Scholz, Dorner,
hence, every point in space will correspond to the most appropriate Franz, & Hinz, 2015). It would therefore be interesting for forthcoming
classification, i.e. in this case a location. All in all, the proposed com- studies to research the effect on quality of recommendations when
putational framework predicts the tourism destination that would po- tourists recommend the same product/service many times in applica-
tentially best fit system users' requirements as entered in the algorithm. tions where they are allowed to do so more than once.
Thus, this paper builds upon the work of Pantano et al. (2017) by
clearly showing the applicability of the proposed predictive model in References
delivering highly efficient and reliable recommender systems within a
tourism destinations and theme parks context. Adomavicius, G., & Tuzhilin, A. (2001). Using data mining methods to build customer
This study contributes to the literature on recommender systems by profiles. Computer, 34(2), 74–82.
Al-Hassan, M., Lu, H., & Lu, J. (2015). A semantic enhanced hybrid recommendation
providing additional findings based on open data analysis concerning approach: A case study of e-Government tourism service recommendation system.
the facilitation of consumer decision making processes. It offers further Decision Support Systems, 72, 97–109.
evidence about recommender systems as Memon et al. (2015), who Alnawas, I., & Aburub, F. (2016). The effect of benefits generated from interacting with
branded mobile apps on consumer satisfaction and purchase intentions. Journal of
highlighted the need for high-quality recommender systems to render Retailing and Consumer Services, 31, 313–322.
optimized recommendations to tourists across a wide spectrum of ap- Al-Quesi, K., Dennis, C., Alamanos, E., & Jayawardhena, C. (2014). Website design quality
plications. The recommender system algorithm proposed in this study and usage behaviour: Unified theory of acceptance and use of technology. Journal of
Business Research, 67(11), 2282–2290.
shows a high predictability of tourists' choices, thus demonstrating its Amblee, N., & Bui, T. (2011). Harnessing the influence of social proof in online shopping:
high effectiveness in providing suggestions. Therefore, it further en- The effect of electronic word of mouth on sales of digital microproducts. International
hances the researchers' toolbox with an effective alternative to other Journal of Electronic Commerce, 16(2), 91–114.
Ansari, A., Essegaier, S., & Kohli, R. (2000). Internet recommendation systems. Journal of
recommender algorithms provided in previous research studies
Marketing Research, 37(3), 363–375.
(Baltrunas et al., 2012; Chu, Wang, Zheng, Wang, & Tan, 2016; Moreno, Arts, J. W. C., Frambach, R. T., & Bijmolt, T. H. A. (2011). Generalizations on consumer
Valls, Isern, Marin, & Borràs, 2013; Nilashi, Bin Ibrahim, Ithnin, & innovation adoption: A meta-analysis on drivers of intention and behaviour.
Sarmin, 2015). This study also responds to the call of Pantano et al. International Journal of Marketing Research, 28(2), 134–144.
Bag, S., Kumar, S., Awasthi, A., & Tiwari, M. K. (2019). A noise correction-based approach
(2017) for testing the applicability of the Random Forest decision trees to support a recommender system in a highly sparse rating environment. Decision
technique. The findings not only provide support for the implementa- Support Systems, 118, 46–57.
tion of the particular computational approach, but also demonstrate for Balaji, M. S., Khong, K. W., & Chong, A. Y. L. (2016). Determinants of negative word-of-
mouth communication using social networking sites. Information & Management,
a specific tourism attraction the high degree of visitors' preferences 53(3), 528–540.
predictability based on the recommendations produced by other visitors Baltrunas, L., Ludwig, B., Peer, S., & Ricci, F. (2012). Context relevance assessment and
to that tourism attraction. On the practical side, organisations/desti- exploitation in mobile recommender systems. Personalized Ubiquitous Computing, 16,
507–526.
nations can acquire useful information on their products/services, can Barragáns-Martínez, B., Costa-Montenegro, E., & Juncal-Martínez, J. (2015). Developing
better reach their clientele and can design products/services that meet a recommender system in a consumer electronic device. Expert Systems with
their clientele's needs and wants. Also, the proposed system adds value Applications, 42(9), 4216–4228.
Blazquez, M. (2014). Fashion shopping in multichannel retail: The role of technology in
in the form of recommendations, as it avoids information overflow and enhancing the customer experience. International Journal of Electronic Commerce,
can thus increase users' satisfaction which to a large extent depends on 18(4), 97–116.
the first interactions with the system and at the same time is an im- Bobadilla, J., Ortega, F., Hernando, A., & Gutiérrez, A. (2013). Recommender systems
survey. Knowledge-Based Systems, 46, 109–132.
portant factor for a recommender systems' success and acceptance (Su &
Bonhard, P., & Sasse, M. A. (2006). Knowing me, knowing you’—Using profiles and social
Khoshgoftaar, 2009). networking to improve recommender systems. BT Technology Journal, 24(3), 84–98.
Notwithstanding that the current study introduced a functional real- Borras, J., Moreno, A., & Valls, A. (2014). Intelligent tourism recommender systems: A
life recommender system using a novel approach, there are some un- survey. Expert Systems with Applications, 41(15), 7370–7389.
Cardoso, J. S., & Costa, J. F. (2007). Learning to classify ordinal data: The data replication
avoidable limitations. The study used a relatively small data set, method. Journal of Machine Learning Research, 1393–1429.
whereas larger data sets might offer a better validation of the proposed Chapelle, O., Haffner, P., & Vapnik, V. N. (1999). Support vector machines for histogram-
recommender system. Also, this study focused on a specific tourism based image classification. IEEE Transactions on Neural Networks, 10(5), 1055–1064.
Chen, M., Zhang, Y., Hu, L., Taleb, T., & Sheng, Z. (2015). Cloud-based wireless network:
setting, so future works could further test the applicability of the re- Virtualized, reconfigurable, smart wireless network to enable 5G technologies. Mobile
commender system to other types of tourism destinations. In any case, Networks and Applications, 20(6), 704–712.
the authors believe that, given the focus of this paper that is to propose Chen, Y. F. (2008). Herd behavior in purchasing books online. Computers in Human
Behavior, 24(5), 1977–1992.
a new recommender system, the limitations mentioned do not sub- Cheung, C. M., & Thadani, D. R. (2012). The impact of electronic word-of-mouth com-
stantially undermine the value of this study. munication: A literature analysis and integrative model. Decision support systems,
The methodological approach followed in this paper responds to the 54(1), 461–470.
Chu, S. C., & Kamal, S. (2008). The effect of perceived blogger credibility and argument
call of Williams, Dwivedi, Lal, and Schwarz (2009) ‘for authors [of
quality on message elaboration and brand attitudes: An exploratory study. Journal of
technology adoption studies] to make greater use of the theoretical and Interactive Advertising, 8(2), 26–37.
methodological variety available to them’ (Williams et al., 2009, p.9). Chu, Y., Wang, H., Zheng, L., Wang, Z., & Tan, K. L. (2016). TRSO: A tourism re-
commender system based on ontology. International Conference on Knowledge Science,
Indeed, the current study offers both quantitative researchers working
engineering and management (pp. 567–579). Springer International Publishing.
in decision sciences and, more specifically, scholars investigating Colace, F., De Santo, M., Greco, L., Moscato, V., & Picariello, A. (2015). A collaborative
tourism destination selection an additional valuable tool for predictive user-centered framework for recommending items in online social networks.
analysis and decision-making process evaluation. As the paper seeks to Computers in Human Behavior, 51, 694–704.
Davis, F. D. (1989). Perceived usefulness, perceived ease of use and user acceptance of
open new avenues in decision-making, it has tried to balance the information technology. MIS Quarterly, 13(3), 319–340.
technical aspect of the tool proposed with the case study context to Davis, A., & Khazanchi, D. (2008). An empirical study of online word of mouth as a
cover both the wider audience of computational analysts, as well as the predictor for multi-product category e-commerce sales. Electronic markets, 18(2),
130–141.
more specific research group of tourism scholars, respectively. Dennis, C., Brakus, J. J., Gupta, S., & Alamanos, E. (2014). The effect of digital signage on
In sum, the present work provides a new recommender system for

329
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

shoppers' behaviour: The role of the evoked experience. Journal of Business Research, Engineering Applications of Artificial Intelligence, 26(1), 633–651.
67(11), 2250–2257. Narducci, F., De Gemmis, M., & Lops, P. (2015). A general architecture for an emotion-
Dennis, C., Jayawardhena, C., Merrilees, W., & Wright, L. T. (2009). E-consumer beha- aware content-based recommender system. Proceedings of the 3rd workshop on emotions
viour. European Journal of Marketing, 43(9/10), 1121–1139. and personality in personalized systems (EMPIRE) (pp. 3–6). ACM.
DisneyNews (2017). 25 Interesting Disneyland Paris Statistics and Fun Facts (June 2017). Ng, M. (2016). Factors influencing the consumer adoption of Facebook: A two country
Retreived from: https://disneynews.us/disneyland-paris-statistics-fun-facts/, on study of youth markets. Computers in Human Behavior, 54, 491–500.
October, 14, 2017. Nilashi, M., Bin Ibrahim, O., Ithnin, N., & Sarmin, N. H. (2015). A multi-criteria colla-
Duan, K. B., & Keerthi, S. S. (2005). Which is the best multiclass SVM method? An em- borative filtering recommender system for the tourism domain using expectation
pirical study. Multiple classifier systems. Lecture Notes on Computer Science, 3541, maximization (EM) and PCA–ANFIS2. Electronic Commerce Research and Applications,
278–285. 14(6), 542–562.
Farokhi, N., Vahid, M., Nilashi, M., & Ibrahim, O. (2016). A multi-criteria recommender Noguera, J. M., Barranco, M., Segura, R., & Martinez, L. (2012). A mobile 3D-GIS hybrid
system for tourism using fuzzy approach. Journal of Soft Computing and Decision recommender system for tourism. Information Sciences, 215, 37–52.
Support Systems, 3(4), 19–29. Okazaki, S., Díaz-Martín, A. M., Rozano, M., & Menéndez-Benito, H. D. (2015). Using
Felfernig, A., Jeran, M., Ninaus, G., Reinfrank, F., & Reiterer, S. (2013). Toward the next twitter to engage with customers: A data mining approach. Internet Research, 25(3),
generation of recommender systems: Applications and research challenges. Multimedia 416–434.
services in intelligent environments: Springer International Publishing81–98. Oliveira, J., Delgado, C., & Assaife, A. C. (2017). A recommendation approach for con-
Fotiadis, A. K., & Stylos, N. (2017). The effects of online social networking on retail suming linked open data. Expert Systems with Applications, 72, 407–420.
consumer dynamics in the attractions industry: The case of ‘E-da'theme park, Taiwan. Pantano, E., & Priporas, C. V. (2016). The effect of mobile retailing on consumers' pur-
Technological Forecasting and Social Change, 124, 283–294. chasing experiences: A dynamic perspective. Computers in Human Behavior, 61,
Frank, E., & Hall, M. (2001). A simple approach to ordinal classification. In L. De Raedt, & 548–555.
P. Flach (Eds.). Machine learning: ECML 2001 (pp. 145–156). Berlin: Srpinger. Pantano, E., Priporas, C. V., & Stylos, N. (2017). ‘You will like it!’ Using open data to
García-Crespo, Á., López-Cuadrado, J. L., Colomo-Palacios, R., González-Carrasco, I., & predict tourists' response to a tourist attraction. Tourism Management, 60, 430–438.
Ruiz-Mezcua, B. (2011). Sem-fit: A semantic based expert system to provide re- Pantano, E., & Viassone, M. (2014). Demand pull and technology push perspective in
commendations in the tourism domain. Expert Systems with Applications, 38(10), technology-based innovations for the points of sale: The retailers evaluation. Journal
13310–13319. of Retailing and Consumer Services, 21(1), 43–47.
Gavalas, D., Konstantopoulos, C., Mastakas, K., & Pantziou, G. (2014). Mobile re- Park, D. H., Kim, H. K., Choi, I. Y., & Kim, J. K. (2012). A literature review and classi-
commender systems in tourism. Journal of Network and Computer Applications, 39, fication of recommender systems research. Expert Systems with Applications, 39,
319–333. 10059–10072.
Geissler, G. L., & Rucks, C. T. (2011). The overall theme park experience: A visitor sa- Park, D. H., & Kim, S. (2009). The effects of consumer knowledge on message processing
tisfaction tracking study. Journal of Vacation Marketing, 17(2), 127–138. of electronic word-of-mouth via online consumer reviews. Electronic Commerce
Ghazanfar, M. A. (2015). Experimenting switching hybrid recommender systems. Research and Applications, 7(4), 399–410.
Intelligent Data Analysis, 19, 845–877. Quinn, L., Dibb, S., Simkin, L., Canhoto, A., & Analogbei, M. (2016). Troubled waters: The
Gretzel, U. (2011). Intelligent systems in tourism: A social science perspective. Annals of transformation of marketing in a digital world. European Journal of Marketing, 50(12),
Tourism Research, 38(3), 757–779. 2103–2133.
Harris, L., & Dennis, C. (2011). Engaging customers on Facebook: Challenges for e-re- Scholz, M., Dorner, V., Franz, M., & Hinz, O. (2015). Measuring consumers' willingness to
tailers. Journal of Consumer Behaviour, 10(6), 338–346. pay with utility-based recommendation systems. Decision Support Systems, 72, 60–71.
Hashem, I. A. T., Yaqoob, I., Anuar, N. B., Mokhtar, S., Gani, A., & Khan, S. U. (2015). The Scholz, M., Dorner, V., Schryen, G., & Benlian, A. (2017). A configuration-based re-
rise of “big data” on cloud computing: Review and open research issues. Information commender system for supporting e-commerce decisions. European Journal of
Systems, 47, 98–115. Operational Research, 259(1), 205–215.
He, C., Parra, D., & Verbert, K. (2016). Interactive recommender systems: A survey of the Senecal, S., & Nantel, J. (2004). The influence of online product recommendations on
state of the art and future research challenges and opportunities. Expert Systems with consumers' online choices. Journal of Retailing, 80(2), 159–169.
Applications, 56, 9–27. Shinde, S. K., & Kulkarni, U. (2012). Hybrid personalized recommender system using
Horikawa, S. I., Furuhashi, T., & Uchikawa, Y. (1992). On fuzzy modeling using fuzzy centering-bunching based clustering algorithm. Expert Systems with Applications, 39,
neural networks with the back-propagation algorithm. IEEE Transactions on Neural 1381–1387.
Networks, 3(5), 801–806. Su, X., & Khoshgoftaar, T. M. (2009). A survey of collaborative filtering techniques.
Johnson, E. J., Bellman, S., & Lohse, G. L. (2003). Cognitive lock-in and the power law of Advances in Artificial Intelligence, 4. https://doi.org/10.1155/2009/421425.
practice. Journal of Marketing, 67(2), 62–75. Sudheep, E. M., Idikkula, S. M., & Alexander, J. (2011). Design and performance analysis
Kang, J.-Y. M., Mun, J. M., & Johnson, K. K. P. (2015). In-store mobile usage: of data mining techniques based on decision trees and naive bayes classifier for
Downloading and usage intention toward mobile location-based retail apps. employment chance prediction. Journal of Convergence Information Technology,
Computers in Human Behaviour, 46, 210–217. 6(5), 89.
Khusro, S., Ali, Z., & Ullah, I. (2016). Recommender systems: Issues, challenges, and Sun, L., Guo, J., & Zhu, Y. (2019). Applying uncertainty theory into the restaurant re-
research opportunities. Information Science and Applications (ICISA), 2016, commender system based on sentiment analysis of online Chinese reviews. World
1179–1189. Wide Web, 22, 83–100.
Knijnenburg, B. P., Willemsen, M. C., Gantner, Z., Soncu, H., & Newell, C. (2012). Suykens, J. A., & Vandewalle, J. (1999). Least squares support vector machine classifiers.
Explaining the user experience of recommender systems. User Modeling and User- Neural Processing Letters, 9(3), 293–300.
Adapted Interaction, 22(4–5), 441–504. Toyoda, Y. (2014). Recontextualizing Disney: Tokyo Disney resort as a kingdom of
Lawlor, M., Dunne, Á., & Rowley, J. (2016). Young consumers'brand communications dreams and magic. Social Science Japan Journal, 17(2), 207–226.
literacy in a social networking site context. European Journal of Marketing, 50(11), Venkatesh, V., Thong, J. Y. L., & Xu, X. (2012). Consumer acceptance and use of in-
2018–2040. formation technology: Extending the unified theory of acceptance and use of tech-
Lee, D., & Hosanagar, K. (2019). How do recommender systems affect sales diversity? A nology. MIS Quarterly, 36(1), 157–178.
cross-category investigation via randomized field experiment. Information Systems Verhoef, P. C., Lemon, K. N., Parasuraman, A., Roggeveen, A., Tsiros, M., & Schlesinger, L.
Research. https://doi.org/10.1287/isre.2018.0800. A. (2009). Customer experience creation: Determinants, dynamics and management
Li, F., & Du, T. C. (2011). Who is talking? An ontology-based opinion leader identification strategies. Journal of Retailing, 85(1), 31–41.
framework for word-of-mouth marketing in online social blogs. Decision support sys- Wang, C. H. (2015). A market-oriented approach to accomplish product positioning and
tems, 51(1), 190–197. product recommendation for smart phones and wearable devices. International
Liao, S. H., Chu, P. H., & Hsiao, P. Y. (2012). Data mining techniques and applications–a Journal of Production Research, 53(8), 2542–2553.
decade review from 2000 to 2011. Expert Systems with Applications, 39(12), Weinberger, K. Q., & Saul, L. K. (2009). Distance metric learning for large margin nearest
11303–11311. neighbor classification. Journal of Machine Learning Research, 10(Feb), 207–244.
Liu, D. R., & Shih, Y. Y. (2005). Integrating AHP and data mining for product re- Willemsen, M. C., & Johnson, E. J. (2011). Visiting the decision factory: Observing cog-
commendation based on customer lifetime value. Information & Management, 42(3), nition with MouselabWEB and other information acquisition methods. A handbook of
387–400. process tracing methods for decision research, 21–42.
Lowe, B., & Alpert, F. (2015). Forecasting consumer perception of innovativeness. Williams, M., Dwivedi, Y., Lal, B., & Schwarz, A. (2009). Contemporary trends and issues
Technovation, 45-46, 1–14. in IT adoption and diffusion research. Journal of Information Technology, 24(1), 1–10.
MacKay, D. J. (2003). Information theory, inference and learning algorithms. Cambridge Witten, I. H., Frank, E., Hall, M. A., & Pal, C. J. (2016). Data mining: Practical machine
University Press. learning tools and techniques. Morgan Kaufmann.
Maity, M., & Dass, M. (2014). Consumer decision-making across modern and traditional Yang, W.-S., Cheng, H.-C., & Dia, J.-B. (2008). A location-aware recommender system for
channels: E-commerce, m-commerce, in-store. Decision Support Systems, 61, 34–46. mobile shopping environments. Expert Systems with Applications, 34, 437–445.
Mathematica, W. (2015). Modern Technical Computing; Wolfram Research. IL, USA: Inc.: Yeh, D. Y., & Cheng, C. H. (2015). Recommendation system for popular tourist attractions
Champaign. in Taiwan using Delphi panel and repertory grid techniques. Tourism Management, 46,
Memon, I., Chen, L., Majid, A., Lv, M., Hussain, I., & Chen, G. (2015). Travel re- 164–176.
commendation using geo-tagged photos in social media for tourist. Wireless Personal Zhang, D., Zhou, L., Kehoe, J. L., & Kilic, I. Y. (2016). What online reviewer behaviors
Communications, 80(4), 1347–1362. really matter? Effects of verbal and nonverbal behaviors on detection of fake online
Mobasher, B., Cooley, R., & Srivastava, J. (2000). Automatic personalization based on reviews. Journal of Management Information Systems, 33(2), 456–481.
web usage mining. Communications of the ACM, 43(8), 142–151. Zhang, Z. K., Zhou, T., & Zhang, Y. C. (2010). Personalized recommendation via in-
Moreno, A., Valls, A., Isern, D., Marin, L., & Borràs, J. (2013). Sigtur/e-destination: tegrated diffusion on user–item–tag tripartite graphs. Physica A: Statistical Mechanics
Ontology-based personalized recommendation of tourism and leisure activities. and its Applications, 389(1), 179–186.

330
E. Pantano, et al. Tourism Management Perspectives 31 (2019) 323–331

Eleonora Pantano, PhD, is a Senior Lecturer in Marketing Nikolaos Stylos, PhD, SFHEA, MCMI is Senior Lecturer in
at Middlesex University Business School, UK. Prior to Marketing and Programme Director of MSc Marketing at
joining Middlesex, she was post doc research fellow at the University of Bristol, UK. He has a PhD in Business
University of Calabria (Italy) and researcher at Technical Administration (Marketing), an M.B.A, an M.Sc. in
University of Eindhoven (The Netherlands). She held a Mechanical Reliability Engineering and a Post Graduate
Ph.D. in “Psychology of Programming and Artificial Diploma in Mechanical Engineering. His research interests
Intelligence” and a Master Science in Business and include consumer behavior, tourism marketing, as well as,
Management Engineering at University of Calabria (Italy). environmental management and sustainable development.
Her research activities explore marketing management and Dr. Stylos has published in various international journals
mainly relate to consumers acceptance of new services and including Journal of Travel Research, Tourism Management,
to customer experience in modern and ancient retail for- International Journal of Contemporary Hospitality
mats, design of new customers solution in retail industry, Management, Computers in Human Behavior, Technological
implementation of business and retail models in terms of Forecasting and Social Change, and Journal of Cleaner
innovation and technology management, and success factors of innovation management Production. He has worked on EU funded projects and served as a professional manage-
with emphasis on retail industry. Her findings appear in numerous international journals ment and technical consultant for more than a decade. (email: n.stylos@bristol.ac.uk).
such as Computers in Human Behavior, Information Technology & People, and
International Journal of Information Management (e-mail: E.Pantano@mdx.ac.uk).
Charles Dennis is Professor of Consumer Behaviour at The
Business School, Middlesex University (UK); and Associate
Constantinos-Vasilios Priporas, PhD, MCIM, FEMAB, is a Editor (Retailing) of the European Journal of Marketing. His
Senior Lecturer in Marketing at Middlesex University main teaching and research area is (e-)retail and consumer
Business School, UK. His research interests include con- behaviour – the vital final link of the Marketing process.
sumer behavior and strategic marketing with main em- Charles is a Chartered Marketer, elected a Fellow of the
phasis on tourism, food and retailing. He has published in Chartered Institute of Marketing for helping to modernise
several international academic journals and conferences, the teaching of the discipline. Charles was awarded the Vice
including Tourism Management, Journal of Travel Research, Chancellor's Award for Teaching Excellence for improving
Journal of Hospitality and Tourism Research, Computers in the interactive student learning experience at Brunel
Human Behavior, The Services Industries Journal, EuroMed University. Charles has published in journals such as
Academy of Business, EMAC. In addition, he co-edited a Journal of Business Research and European Journal of
book. He is a member of several professional bodies and he Marketing. Books include Marketing the e-Business, (1st &
is editorial board member of the Journal of Customer 2nd editions) (joint-authored with Dr. Lisa Harris); and
Behaviour and has acted as a guest editor, reviewer, and research monograph Objects of Desire: Consumer Behaviour in Shopping Centre Choice
track chair in academic journals and conferences. (e-mail: (Palgrave). His research into shopping styles has received extensive coverage in the
C.Priporas@mdx.ac.uk). popular media including TV appearances with Sir Trevor McDonald OBE and Adrian
Edmondson (email: C.Dennis@mdx.ac.uk).

331