Over the past twenty years, the supply and demand of tourism have transformed for the dynamism of the global economy and the evolution of technology. These changes include the severer competition between destinations, more and more international tourists, and the increased frequency of short trips for people (OECD, 2010). Besides, experience economy is a trend emerging into tourism. Innovation occurred in the service delivery and the product development for SMEs within destinations (e.g. restaurants, accommodations) (OECD, 2010); tourists incline to customized and flexible travel products and services (Stamboulis and Skayannis, 2003). In the end, SMEs as service providers co-create service experiences with their customers during the journey. Then tourists gain the value through evaluating the customized service experiences (Sandström et al., 2008).
The destination recommendation system (DRS) or the travel recommendation system is one of the information technologies having impacts on value creation chains of tourism (OECD, 2010). Its functions include not only the travel information provision, but also suggesting end consumers destinations and arouse their desires to visit the places (Skadberg et al., 2005). Therefore, DRS is supposedly an appropriate tool to increase the destinations‘ competence level by establishing a mutual relationship between destinations and tourists, and satisfy tourists all over the world via recommending favorable destinations (Yuksel and Bilim, 2009).
DRS provide an object situation for tourists to review existing destinations.
These destinations are dynamic ranked by DRS according to individual customers
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
for alleviating their burden of searching through thousands and hundreds of tourism products and services on the Internet. Nevertheless, it has been widely recognized that a gap exists between customers‘ expected services and their perceived services provided by information systems as shown in Figure 1 (Pitt et al., 1995). Being an information system, it is vital for DRS to understand what tourists want, i.e. the customers‘ expectations, in order to close this gap. Actually, capturing the personal needs (as shown in Figure 1) of tourists is an essential step to derive the tourist satisfaction of recommenders that aim at providing information of desirable destinations and SMEs within tourism destinations.
Further, tourism is an experience based industry. These experiences are individually unique and emotion attached. It was claimed that insufficient attention has been paid to the service experiences, including the emotional dimensions (Sandström et al., 2008). Therefore, a DSR should have capabilities of capturing tourists‘ needs, especially the emotional needs, and producing recommendation according to the psychological or emotional properties of tourism products and services.
Figure 1. Determinants of Users' Expectations (Pitt et al., 1995)
‧
Since the emotional information and experiences of tours occur and are confined to only individuals, we argued that a DRS have to capture both tourists‘ minds, i.e.
the emotional needs, and the psychological elements of destinations, in order to suggest customers the best fit results. Without capturing tourists‘ minds in recommendations, the DRS would find itself only satisfied restricted market segments that people favor popular attractions.
Reviewing recommendation systems or platforms nowadays, the technologies they used are generally categorized into content-based, collaborative filtering, knowledge-based, and hybrid methods. However, we found all of these technologies are function oriented, such as content-based recommendation deals with the product descriptions (Zhang et al., 2009); collaborative filtering stresses on the rating data (Wang et al., 2006); and knowledge-based infers recommendation results from functional knowledge about customers, items, and the mapping relations between them (Burke, 2002).
Besides, in the previous research, customer expectations and needs were only represented with questionnaires (Quader, 2009; Robledo, 2001) or a query input describing item information and constraints. But for an efficient travel recommendation system, there have to be a uniform representation which can stand for tourist expectations (emphasizing the emotional needs), destinations, and SMEs so that we can manipulate them to do the matching through a uniform comparison of their similarities. In this way, the flexibility of questionnaires appears to be insufficient to reflect these three stakeholder roles‘ real-time images, because the result data is static and limited by the question designers.
For the competition in the flourishing tourism industry, the images perceived
‧
making process, consumers can reduce the number of alternatives through comparing their expectations with the images of destinations and SMEs. Images can also be a key component in the destination positioning process (Echtner and Ritchie, 2003). Destinations and SMEs can also create their own positive images in order to differentiate themselves from competitors by modifying their operations and policies through diagnosing their own images.We believe that images can serve as the uniform representation for destinations, SMEs, and tourists in our DRS. The reasons include: (1) image is an output of the mental picturing process which people will execute before starting a trip; and (2) images have been commonly used in the marketing of tourism destinations (Robledo, 2001); (3) images can be expressed in languages, so that the material for images can be obtained from Web resources and is dynamic and open-ended.
In this research, we investigate two research questions. The first is how to devise a systematic method to measure and model images. The second is what will happen to these images when roles interactions and social events occur. This research proposes a resolution method and system in response to two questions and the method has two main components—Image Modeling and Image Mixing—
that will be illustrated later.
1.3 Research Method
In our system, each tourist, destination, and SME has their own image, which consists of psychological adjectives. We utilize color as a tool to model all the
‧
images of these stakeholder roles.
Color impacts our everyday life. We can treat it as a source of information to make decisions based on the aroused emotions. It is widely accepted that color can be mapped onto emotions (Xin et al., 1998; Kobayashi, 1981, 1992, 2001; Nijdam, 2005; Ou et al., 2004; Suk and Irtel, 2010). To name a few, the red color can stand for festive and hot feelings; green can be associated with a peaceful or health image. Although the reaction to colors varies person by person, researchers have clarified that there exist universal color factors, including warm–cool, heavy–light, active–passive, and hard–soft (Ou et al., 2004).
Kobayashi and the Nippon Color & Design Research Institute developed Color Image Scale (Kobayashi, 1981, 1992, 2001), which specified the meaning of 130 basic colors according to the two factors: warm–cool and hard–soft. These meanings were assigned with 180 image words (adjectives). On the scale, the similarity and dissimilarity of the emotional word meanings are indicated with the distances between the corresponding colors.
With these color-emotion mapping knowledge, we can obtain more information dimensions about our image models than word meanings, such as their inter relations with precise distances on a color space, the possible matching pairs among them according to the color harmony theory. For example, a man with an image model mainly colored red which means festive and dazzling, and hot. He will be pleased with our recommendation—Taichung and Pingxi Sky Lantern, in all likelihood.
Sometimes, things with different images or emotions will be put together to obtain a fresh new image. For example, SMEs will forge an alliance to build distinguishing features for attractiveness and uniqueness. Besides, the images
‧
carried by tourists, SMEs, and destinations will influence each other during their interactions. To understand the consequences of these alliances and interactions to the images, the image mixture should be functioned in our system.
Since images can be represented by colors, the image mixture may have a strong connection to the color mixture. We propose that the additive color mixing method (normally used to do the light mixture) can serve as the tool for image mixing, for the reason that image or impression is virtual. That means there is no need to concern about its spectral composition, which is a consideration when mixing pigment or printed colors which use subtractive color mixing (Broackes, 1992). With this color mixing method, we can not only shrink the distance between the images and the real status of the stakeholder roles by reflecting the real-time interactions between these stakeholder roles, but also foreseen the results of potential SME alliances.
1.4 Purpose and Contribution
The aim of this research is to demonstrate a destination recommendation system that uses images to capture tourists‘ needs or intrinsic motives, and recommends destinations and SMEs which can meet tourists‘ emotional needs. In addition, the design, method, and architecture of this system could be domain-independent and applicable to a wide range of services. It recommends people things what they can be satisfied based on images which represent both objects and humans‘ minds.
The results of the system are dynamic which evolve over time through the expected changes derived from the 3 stakeholders‘ role interactions and the unexpected changes caused by casual events of society. With this approach, we can find good matches between human and objects (tangible, intangible) based on
‧
In addition, none of the field of information system studies has ever utilized images as the representation of customer expectations, and recommended people the destinations and SMEs according to their images. It is difficult because images are psychological related.
Our system is also in line with Service-Dominant Logic (Vargo & Lusch 2004), which is a mindset distinguished from Goods-Dominant Logic.
Service-Dominant Logic highlights several perspectives such as service exchange, operant resources, co-creation of value, and value in-use. With these concepts, people can rethink role relationships, resources integration, IT facilitation, etc. to gain profound understanding about the service system. In the end, they can be innovative means with a combination of integrated operand and operant resources to fulfill demands. In this research, the recommendation information system is also a dynamic service system which evolves over time. Images of tourists, destinations, and SMEs will change according to the interactions between each other. The active image, therefore, can be regards as an operant resource.
Moreover, the good matches can only be found when customers have a willingness to provide their needs or intrinsic motives of destinations and to be involved in the image modeling process, e.g. attendance of trips and feedbacks, which correspond to the concept of value in-use.
1.5 Content Organization
This research is organized as follow brief introductions: