Towards a better understanding of the cognitive destination image of the Basque Country based on the analysis of UGC
Download as PPT, PDF0 likes480 views
The document analyzes the cognitive destination image of the Basque Country through user-generated content (UGC) and social networks, emphasizing the influence of consumer opinions on perceptions of travel destinations. It details a methodology for modeling this image using text mining tools to extract insights from various digital media sources. The findings underscore the importance of cultural resources and the impact of communication strategies by destination management organizations on shaping potential visitors' views.
Towards a better understanding of the cognitive destination image of the Basque Country based on the analysis of UGC
- 1. Towards a better understanding of the cognitive destination image of the Basque Country based on the analysis of UGC Ainhoa Serna Ph.D. Aurkene Alzua, Ph.D Jon Kepa Gerrikagoitia, Ph.D ENTER 2014 Research Track Slide Number 1
- 2. Agenda • Introduction – Problem – Previous research • Method • Evaluation results ENTER 2014 Research Track Slide Number 2
- 3. Introduction Opinions and UGC • The opinions and experiences are central to almost all human activities and are key factors in influencing our behavior. Our beliefs and perceptions of reality, and the decisions we make, are largely conditioned by how others see and evaluate the world. • When we have to make a decision often seek the opinions of others (Liu, 2012). • Consumers no longer rely exclusively on official sources to obtain information about destinations and services. Visitors take into higher condideration UGC because those comments are based on experience (Pan et al., 2008) obtaining a more complete image of the destination and its products (Stepchenkova, 2013; Pang et al., 2011). ENTER 2014 Research Track Slide Number 3
- 4. Introduction Social Networks • With the explosion of Web 2.0 consumers have a place where to share their experiences (Çakmak, 2012). Thus, users opinion gains weight against traditional marketing techniques (Ditoiu et al, 2012). • Social networks are affecting very directly to brands and e-commerce. 65% of active Facebook accounts follow a brand (Llantada, 2013). Facebook reaches 85% of Internet users and Twitter 32%. • Social networks support the complete cycle of tourism: inspiration, planning, booking, stay and content generation based on experiences at the destination. ENTER 2014 Research Track Slide Number 4
- 5. Introduction Social Networks • Much of the UGC is generated by social networks like Twitter, Facebook, ... millions of users use them every day and this information is valuable for modeling the vision of the tourist destination. • Volume of UGC (reviews, forums, blogs, twitter, comments, and posts on social networks) has reached astronomical proportions (Liu, 2012) making imposible the work of monitoring manually the opinions (Subrahmanian, 2009). ENTER 2014 Research Track Slide Number 5
- 6. Introduction Destination Image • The UGC and social networks become a valuable source to get insights about the image of the destination. • The analysis of the discourses on the web provides an approach to extract information and knowledge associated with how destination is perceived, motivations, and underlying discourses. • A proper management of perceptions or images that a potential visitor can have on a destination contributes to create a different positioning (Stepchenkova and Morrison, 2008), accordingly the destination image is presented as a key factor (Echtner and Richie, 1991, 1993; Chon, 1991). • Despite the great importance of the destination image (Gartner, 1986, 1987), there is not a universally accepted and validated model (Crompton, 1977; Baloglu & McCleary, 1999a). » Choosen reference model, Gómez, M. et al (2013). ENTER 2014 Research Track Slide Number 6
- 7. Introduction Aim • Our aim is (what): • to model the cognitive destination image of the Basque Country by visitors, • having as reference the conceptual model proposed by Gomez, M. et al (2013) • based on data collected from relevant digital media: destination sites, general social networks, traveler blogs, reviews, ... etc. ENTER 2014 Research Track Slide Number 7
- 8. Introduction Aim • This work involves (how): • automatic semantic identification of opinions from social networks • using a text mining tool adapted to the domain of the destination, • developed by CICtourGUNE and Mondragon Unibertsitatea. ENTER 2014 Research Track Slide Number 8
- 9. Introdution Previous Research • Destination image has been defined as an individual’s overall perception or total set of impressions of a place (Hunt, 1975; Phelps, 1986; Fakeye and Crompton, 1991) • Research has been conducted since the early 1970s (Hunt, 1975; Mayo, 1973) and comprises hundreds of studies to date with several assement methods – Most tourism scholars agree that destination image holds at least two distinctive components – cognitive (knowledge and beliefs) and affective (feelings towards a destination). – Other authors proposed that destination images are formed by four interrelated components: cognitive, affective, evaluative and behavioural (Boulding, 1956). – Echtner and Ritchie (2003) argue that a destination image is processed in a three dimensional space: namely functional, psychological and holistic. ENTER 2014 Research Track Slide Number 9
- 10. Introdution Previous Research • From the cognitive perspective (M.Gomez et al, 2013) purpose a destination image model based on 4 dimensions: Natural and Cultural Resources, Infrastructures and Socio-economic environment, Social constraints and atmosphere constraints • The literature considers the amount and variety of secondary data sources as an external variable that contributes significantly to the formation of the image (Gartner & Hunt, 1987; Um & Crompton, 1990; Bojanic, 1991, Gartner, 1993; Font, 1997; Baloglu, 1999): • (Stepchenkova , 2012) used UGC and social networks (Flickr) to create maps representing the projected and perceived image of a destination. • (Andrade, 2010) uses destination websites to build the destination image by the tourist. • (Xiang & Gretzel, 2010) have used the eWoM, specifically travel blogs, in the evaluation of a destination image. ENTER 2014 Research Track Slide Number 10
- 11. Methodology • The model will be generated according to next steps: @1. Defining and selecting relevant data @2. Extracting categories from the selected data sources @3. Transforming the categories into a model ENTER 2014 Research Track Slide Number 11
- 12. Methodology Sources 3.1 Defining and selecting task relevant data – A prior analysis was carried out taking into account the quality and quantity of the data from a double perspective, the projected image by the DMO and the perceived image by the visitor. ENTER 2014 Research Track Slide Number 12
- 13. Methodology Sources ENTER 2014 Research Track Slide Number 13
- 14. Methodology Data Gathering 3.2 Extracting categories from the selected data sources •3.2.1 Data gathering – The data extraction is an automatic process achieved by a software program ENTER 2014 Research Track Slide Number 14
- 15. Methodology Data Gathering Different data gathering techniques are used depending on the source: • Minube provides an API (Application Programme Interface) allowing searches at territory (Araba, Bizkaia, Gipuzkoa) and city (Bilbao, San Sebastián, Vitoria-Gasteiz) disaggregation level. • Data from twitter si extracted using the Twitter4J API (Java library) • Facebook is queried through the Facebook Graph API. • The remaining data sources are dumped using web crawling and scraping techniques developed ad-hoc. ENTER 2014 Research Track Slide Number 15
- 16. Methodology Entity recognition and categorization 3.2.2 Entity Categorization •The process of grouping entities into entity categories is called entity categorization. •First entities are extracted achiving the named entity recognition (NER) process (Hobbs and Riloff, 2010; Mooney and Bunescu, 2005; Sarawagi, 2008). •Once the recognized entities are classified a list of categories, entities and number of occurrences is obtained. ENTER 2014 Research Track Slide Number 16
- 17. Methodology Entity recognition and categorization • Enhanced and adapted instruments by the research group: • An ontology to obtain the categories. • A text mining tool based on the Wordnet lexical database that implements an algorithm for learning generalized association rules (hierarchy for categorization and pairs for association) ENTER 2014 Research Track Slide Number 17
- 18. Methodology 3.3 Transforming the categories into a model Matching of the categories and subcategories with the four dimensions of the model Gomez, M et al. (2013) automatically. 1.The first dimension (REC) represents the natural and cultural resources (attractions) including nature, architecture, cultural attractions, customs and cultural activities. 2.The second dimension (INF) represents infrastructures and the socioeconomic environment: accommodation, entertainment, shopping, development and value for price. 3.The third dimension (SOC) represents Social Conditions: sustainability, cleaning, security, pollution ... 4.The fourth dimension (ATM) represents the Atmosphere: relax, absence of mass, “peace and rest”. ENTER 2014 Research Track Slide Number 18
- 19. Evaluation Results ENTER 2014 Research Track Slide Number 19
- 20. Evaluation Results • The resulting model allows us to approach the cognitive destination image of the tourist about the Basque Country based on the user-generated content about experiences related to basque territory. ENTER 2014 Research Track Slide Number 20
- 21. Evaluation Results Museo Guggenheim San Juan De Gaztelugache Torres Isozaki Bosque de Oma ENTER 2014 Research Track Playa de Sopelana Santuario de Urkiola Slide Number 21
- 22. Evaluation Results ENTER 2014 Research Track Slide Number 22
- 23. Evaluation Results • Language: About 70% of the visitors in Euskadi are Spanish speakers, the sample of the elaborated discourse taken into account is almost completely in this language. • The projected image in twitter and facebook uses the English to translate original messages in spanish. The projected image is fostered by embassadors who are residents and basque language (eu) appears in discourses. ENTER 2014 Research Track Slide Number 23
- 24. Evaluation Results • TripAdvisor is the network with the most number of comments with little influence and presence of the DMO (Destination Management Organization) but spontaneous user-generated content, thus it is an important data source to know the perceived image of the destination. • The DMO does an important effort of communication and promotion by Facebook and twitter. • The type of discourse is mostly related to the source, by the type of messages or the constraints of the channel. The more elaborated discourses show more and better described entities. ENTER 2014 Research Track Slide Number 24
- 25. Evaluation Results The categorization process obtains as results a table by data source with the identified entities, number of occurrences, and the corresponding category ENTER 2014 Research Track Slide Number 26
- 26. Evaluation Results Twitter Euskadi Turismo i_Euskadi (TV Euskadi) ENTER 2014 Research Track Slide Number 27
- 27. Conclusions and Future lines • TripAdvisor is the social network with the most number of purely user-generated comments. • TripAdvisor has little influence and presence of the DMO. • The DMO does an important effort promoting the destination using Facebook and twitter, having this activity a direct influence on the projected image. ENTER 2014 Research Track Slide Number 28
- 28. Conclusions and Future lines • The type of discourse is related to the source, by the type of messages or the constraints of the channel. More elaborated discourses -> better described entities. • All the analyzed sources have a major impact on the first dimension concerning natural and cultural resources. ENTER 2014 Research Track Slide Number 29
- 29. Conclusions and Future lines • The DMO´s efforts are focused on the first dimension, everything related to resources such as Nature, Architecture, Cultural Attractions, Traditions, Cultural Activities. • The broadcasting of the event agenda by the DMO has a big presence in facebook and twitter. ENTER 2014 Research Track Slide Number 30
- 30. Conclusions and Future lines • The cultural attributes of the destination are relevant in the discourses. • There is a significant interest in the subcategory of cultural activities: theater, exhibition, jazz, ... etc. ENTER 2014 Research Track Slide Number 31
- 31. Conclusions and Future lines • The third and fourth dimensions may be considered in the communication strategy of the DMO. • Within the natural and cultural resources, that gastronomy becomes the most relevant category by the quantity of related terms and concepts that appear especially in the user-generated content by visitors in all sources. ENTER 2014 Research Track Slide Number 32
- 32. Conclusions and Future lines • The named entity recognition and categorization process provides many variables that are out of the scope of this paper and will be studied in future works, like the sentiment of the speech, the most important points of interest that are commented, attractiveness of a place,…. ENTER 2014 Research Track Slide Number 33
- 33. Conclusions and Future lines • To sum up, the analysis of the usergenerated content using domain oriented text mining tools provides an approach to understand the destination in the way what people says and perceives and what DMO communicates. ENTER 2014 Research Track Slide Number 34
Editor's Notes
- #4: The user-generated have allowed substantial changes in the dynamic of the travel and tourism sector.
- #5: The user-generated content UGC (User Generated Content), and in particular, the online comments have allowed substantial changes in the dynamic of entire sectors. Such is the case of the travel and tourism sector. Consumers no longer rely exclusively on official sources to obtain information about destinations and services. Thanks to new technologies and the growing number of platforms that support communication between different stakeholders, nowadays consumers can access vast amounts of information about products from different sources. According to some studies, visitors take into higher condideration UGC because those comments are based on experience (Pan et al., 2008). This allows consumers to obtain a more complete image of the destination and its products (Stepchenkova, 2013; Pang et al., 2011). With the explosion of Web 2.0, blogs and social networks, consumers have a place where to share their experiences with different brands giving their opinions, positive or negative about any product or service (Çakmak, 2012). Thus, users opinion gains weigh against brands and traditional marketing techniques (Ditoiu et al, 2012). According to the National Statistics Institute (INE) of Spain, 80% of Internet users access the net to learn about products, brands and services. In another study conducted in 2009 by the Association for Research in Media (AIMC), 75.5% of Spanish Internet users have been documented on the internet before making a purchase of products or services.
- #6: The user-generated content UGC (User Generated Content), and in particular, the online comments have allowed substantial changes in the dynamic of entire sectors. Such is the case of the travel and tourism sector. Consumers no longer rely exclusively on official sources to obtain information about destinations and services. Thanks to new technologies and the growing number of platforms that support communication between different stakeholders, nowadays consumers can access vast amounts of information about products from different sources. According to some studies, visitors take into higher condideration UGC because those comments are based on experience (Pan et al., 2008). This allows consumers to obtain a more complete image of the destination and its products (Stepchenkova, 2013; Pang et al., 2011). With the explosion of Web 2.0, blogs and social networks, consumers have a place where to share their experiences with different brands giving their opinions, positive or negative about any product or service (Çakmak, 2012). Thus, users opinion gains weigh against brands and traditional marketing techniques (Ditoiu et al, 2012). According to the National Statistics Institute (INE) of Spain, 80% of Internet users access the net to learn about products, brands and services. In another study conducted in 2009 by the Association for Research in Media (AIMC), 75.5% of Spanish Internet users have been documented on the internet before making a purchase of products or services.
- #7: The user-generated content UGC (User Generated Content), and in particular, the online comments have allowed substantial changes in the dynamic of entire sectors. Such is the case of the travel and tourism sector. Consumers no longer rely exclusively on official sources to obtain information about destinations and services. Thanks to new technologies and the growing number of platforms that support communication between different stakeholders, nowadays consumers can access vast amounts of information about products from different sources. According to some studies, visitors take into higher condideration UGC because those comments are based on experience (Pan et al., 2008). This allows consumers to obtain a more complete image of the destination and its products (Stepchenkova, 2013; Pang et al., 2011). With the explosion of Web 2.0, blogs and social networks, consumers have a place where to share their experiences with different brands giving their opinions, positive or negative about any product or service (Çakmak, 2012). Thus, users opinion gains weigh against brands and traditional marketing techniques (Ditoiu et al, 2012). According to the National Statistics Institute (INE) of Spain, 80% of Internet users access the net to learn about products, brands and services. In another study conducted in 2009 by the Association for Research in Media (AIMC), 75.5% of Spanish Internet users have been documented on the internet before making a purchase of products or services.
- #8: The user-generated content UGC (User Generated Content), and in particular, the online comments have allowed substantial changes in the dynamic of entire sectors. Such is the case of the travel and tourism sector. Consumers no longer rely exclusively on official sources to obtain information about destinations and services. Thanks to new technologies and the growing number of platforms that support communication between different stakeholders, nowadays consumers can access vast amounts of information about products from different sources. According to some studies, visitors take into higher condideration UGC because those comments are based on experience (Pan et al., 2008). This allows consumers to obtain a more complete image of the destination and its products (Stepchenkova, 2013; Pang et al., 2011). With the explosion of Web 2.0, blogs and social networks, consumers have a place where to share their experiences with different brands giving their opinions, positive or negative about any product or service (Çakmak, 2012). Thus, users opinion gains weigh against brands and traditional marketing techniques (Ditoiu et al, 2012). According to the National Statistics Institute (INE) of Spain, 80% of Internet users access the net to learn about products, brands and services. In another study conducted in 2009 by the Association for Research in Media (AIMC), 75.5% of Spanish Internet users have been documented on the internet before making a purchase of products or services.
- #10: and as the mental portrayal of a destination (Crompton, 1979; Woodside and Ronkainen, 1993; Kotler et al, 1993; Middleton, 1994; Milman and Pizam, 1995; Alhemoud and Armstrong, 1996; Seaton and Bennett, 1996). (structured surveys of human respondents, mixed-methods approach textual data from blogs and websites)
- #12: Our aim is to model the cognitive destination image of the region of Euskadi in Spain by visitors, having as reference the conceptual model proposed by Gomez, M. et al (2013) based on data collected from different digital media: destination sites, general social networks, traveler, blogs, reviews, ... etc. To achieve this it is necessary to investigate the meaning of the data to facilitate the generation of the model that helps the understanding the cognitive image of the destination.
- #13: In this analysis different channels were taken into account: channels focused on tourism, channels conducted by DMO's, DMO ́s specific channels in general social networks, ...
- #15: In this step the information is extracted either through API's ("Application Programming Interface") or web scraping techniques depending on the interfaces of data retrieval offered by the data providers. just providing the user account, in this case “TvEuskadi”. The Twitter API limits access to the latest 3.200 states of a user, therefore this is the recovered sample from the 7.266 tweets published from this channel until 26/04/2013. This API shows in a simple and consistent way the Facebook Social Graph (Facebook Social Graph). The social graph repesents graph objects (eg, people, photos, events, pages, etc..) and the connections among them (eg, friend relationships, shared content, labels forums, etc..). The sample periodicity is defined according to the activity of the channel in terms of creation of new contents.
- #17: Furthermore, in Wordnet, nouns, verbs, adjectives and adverbs are grouped into sets of cognitive synonyms named synsets. This process is necessary because in natural language text, people often write the same entity in different ways. As can be observed in the sample text (written in Spanish), categorization is based on listed entites. There are some location names as "Borda de Urbía", “Arantzatzu”, “Aitzkorri”. Also common names such as “hora”, “caminata”, “casona”, “café”, “sidra”, “patxaran”, “bocadillo”, “plato combinado”, “vistas”, “foto”, “tranquilidad”, “precio” that are entities that will belong to different categories. Among these common names there are concepts we group in the gastronomy category: “café”, “sidra”, “patxaran”, “sandwich”, “plato combinado”. In addition, “paisaje”, “vistas” are part of the natural environment and natural resources. In our actual experiment the text mining tool discovered a large number of interesting and important non- taxonomic conceptual relations, the atmosphere concept appears in many different and rich ways qualified as pleasant, charming, good, great, cozy, intimate, caring, lively, relaxed, disco, comfortable, enjoy the atmosphere, lively atmosphere, friendly surfer, footballing environment, unique atmosphere, cosmopolitan, sailor, summery and festive. The same way the term magic appears related to magical day, magical forest, climate magical, magical, magical labyrinth, magical atmosphere / magical moment, magical sunset, magical mountains. All this terms generated by the user match with a dimension of the model. In the same way, the “people” term appears qualified as great, good, friendly, simple, nice, unique, endearing, enjoyable.
- #19: Refering the previous example from minube, we can see that the experience has to to with more than one dimension of the model. The location names are concepts aligned with the first dimension of the model, natural and cultural resources. The concepts refered to mobility, duration and price impact in the second dimension. Finally, Concepts like “buen lugar para reposar ” are matched with de atmosphere dimensión.
- #20: Entidades – common and proper nouns Adjetivos y adverbios para la polaridad
- #21: The resulting model allows us to approach the cognitive destination image of the tourist about the Basque Country based on the user-generated content about experiences related to basque territory. The model implements the four dimensions that will capture the conceptual structures of the cognitive image of the destination by the perspective of the DMO and the visitor
- #24: As about 70% of the visitors in Euskadi are Spanish speakers, the sample of the elaborated discourse taken into account is almost completely in this language.
- #29: In addition, this source may be taken into account by the DMO as promotional channel. In order to validate the Gomez, M. et al (2013) model it is important to underline that the perceived cognitive image based on user-generated content has a great impact on the first dimension
- #33: The presented conclusions are partial observations and interpretations that can be stated from such a valuable source and results.