Text mining into customers’ minds
(2019) FEKH29 20191Department of Business Administration
- Abstract
- Title: Text mining into customers’ minds
Seminar date: 4 June 2019
Course: FEKH29 Business Administration: Bachelor Degree Project in Marketing Undergraduate Level, 15 credits
Authors: Viktor Alfredsson, Jakob Ferenius & Rebin Naderi
Advisor: Burak Tunca
Keywords: Brand associations; brand association networks; text mining; customer reviews; machine learning
Purpose: To develop a text mining method to derive brand association networks from unstructured online customer reviews and to explore how this method can assist managers in developing more effective brand management strategies.
Methodology: Customer reviews were collected using a web extraction system. The reviews were processed using text mining... (More) - Title: Text mining into customers’ minds
Seminar date: 4 June 2019
Course: FEKH29 Business Administration: Bachelor Degree Project in Marketing Undergraduate Level, 15 credits
Authors: Viktor Alfredsson, Jakob Ferenius & Rebin Naderi
Advisor: Burak Tunca
Keywords: Brand associations; brand association networks; text mining; customer reviews; machine learning
Purpose: To develop a text mining method to derive brand association networks from unstructured online customer reviews and to explore how this method can assist managers in developing more effective brand management strategies.
Methodology: Customer reviews were collected using a web extraction system. The reviews were processed using text mining techniques, including tokenization, stop words removal, lemmatization, and feature extraction. This resulted in numerous brand associations for each brand, which were classified along the dimensions of type, strength, valence, uniqueness, and relevance. The associations’ relationships with each other, and with the brand, were modeled through networks. The derived brand association networks were analyzed between the hotel brands, as well as between satisfied and unsatisfied customers. To generate actionable managerial insights, an importance-performance analysis was conducted to identify the most valuable associations for a hotel brand to focus on.
Theoretical perspectives: The theoretical foundation of the thesis lies in theory on brand associations and customer based brand equity. The conceptualization of brand associations included five dimensions: type, valence, strength, uniqueness, and relevance.
Empirical foundation: The study focused on the hotel industry. 58,709 customer-generated hotel reviews for Hilton, Marriott, Wyndham, and Holiday Inn hotels in New York were collected from Hotels.com.
Conclusions: Several brand associations such as room, hotel, staff, stay, location, clean, and friendly could be derived with the methodology. The derived brand associations were further organized through brand association networks. Network analyses showed that some of the associations were more central and influential than others, thereby illustrating how managers can use these networks for strategic decision making. Brand associations also formed clusters in the networks. For instance, a cluster of room-related associations was discovered, as well as a staff-cluster and a location-cluster. Overall, the associations in the room-cluster had a negative impact on hotel ratings, while the ones in the location- and staff-clusters had a positive impact. Results also showed that associations that were unique to each brand could be identified, which would allow managers to adjust their branding strategies with regard to competitors. Brand association networks analyzed separately for satisfied and unsatisfied customers showed that satisfied customers talked more about the intangible aspects of the hotel service offerings like friendly, comfortable, and convenient, while unsatisfied customers talked more about the tangible aspects such as room, bed, and desk. Finally, we demonstrated that managers can use the importance-performance matrix to decide which associations should be improved, maintained, decreased, or eliminated. (Less) - Abstract (Swedish)
- Examensarbetets titel: Text mining into customers’ minds
Seminariedatum: 4 juni 2019
Ämne/kurs: FEKH29, Examensarbete kandidatnivå i marknadsföring, 15 högskolepoäng
Författare: Viktor Alfredsson, Jakob Ferenius & Rebin Naderi
Handledare: Burak Tunca
Fem nyckelord: Varumärkesassociationer; varumärkesassociationsnätverk; text mining; kundrecensioner; maskininlärning
Syfte: Att utveckla en text mining-metod för att härleda varumärkesassociationsnätverk från webbaserade kundrecensioner. Vidare var syftet att utforska hur denna metod kan vägleda företagsledare i att utveckla effektivare varumärkesstrategier.
Metod: Kundrecensioner hämtades genom ett automatiserat webextraheringssystem. Texterna i recensionerna... (More) - Examensarbetets titel: Text mining into customers’ minds
Seminariedatum: 4 juni 2019
Ämne/kurs: FEKH29, Examensarbete kandidatnivå i marknadsföring, 15 högskolepoäng
Författare: Viktor Alfredsson, Jakob Ferenius & Rebin Naderi
Handledare: Burak Tunca
Fem nyckelord: Varumärkesassociationer; varumärkesassociationsnätverk; text mining; kundrecensioner; maskininlärning
Syfte: Att utveckla en text mining-metod för att härleda varumärkesassociationsnätverk från webbaserade kundrecensioner. Vidare var syftet att utforska hur denna metod kan vägleda företagsledare i att utveckla effektivare varumärkesstrategier.
Metod: Kundrecensioner hämtades genom ett automatiserat webextraheringssystem. Texterna i recensionerna bearbetades genom text mining-tekniker såsom tokenisering, borttagning av stoppord, lemmatisering och särdragsextrahering. Detta resulterade i flertalet varumärkesassociationer för varje varumärke som klassificerades enligt dimensionerna typ, stryka, valens, unikhet och relevans. Associationernas relation med varandra och med varumärket modellerades via nätverk. De härledda varumärkesassociationsnätverken analyserades mellan hotellkedjorna samt mellan nöjda och missnöjda kunder. För att generera handlingsbara insikter för företagsledare gjordes en viktighets-prestationsanalys som identifierade de mest värdefulla associationerna för hotellet att fokusera på.
Teoretiska perspektiv: Studien tar sin utgångspunkt i varumärkesassociationer och kundbaserat varumärkeskapital. Varumärkesassociationer klassificerades längs med fem dimensioner: typ, valens, styrka, unikhet, och relevans.
Empiri: Studien fokuserade på hotellindustrin. 58 709 kundrecensioner samlades in för Hilton, Marriott, Wyndham och Holiday Inn hotell i New York via Hotels.com genom ett web extraction system.
Resultat: Flertalet varumärkesassociationer såsom room, hotel, staff, stay, location, clean och friendly kunde härledas genom metoden. De härledda associationerna kunde sedan organiseras i varumärkesassociationsnätverk. Nätverksanalyser visade att några av associationerna hade mer centrala och inflytelserika positioner i nätverket än andra, således kan företagsledare använda dessa nätverk för strategiskt beslutsfattande. Varumärkesassociationerna visade sig även forma kluster i nätverken. Till exempel, hittades ett kluster av room-relaterade associationer, såväl som ett staff-kluster och ett location-kluster. Överlag hade associationerna i room-klustret en negativ påverkan på hotell-ratings medan associationerna i location- och staff-klustren hade en positiv påverkan. Resultaten visade även att associationer som var unika för varumärke kunde identifieras, vilket skulle kunna tillåta företagsledare att anpassa sin marknadsföringsstrategi med hänsyn taget till konkurrenterna. Separata varumärkesassociationsnätverk för nöjda och missnöjda kunder visade att nöjda kunder i större utsträckning pratade om immateriella aspekter av hotellerbjudandet såsom friendly, comfortable och convenient medan missnöjda kunder talade mer om materiella aspekter såsom room, bed och desk. Slutligen demonstrerade vi att företagsledare kan använda en viktighets-prestationsmatris för att besluta vilka associationer som skall förbättras, behållas, minskas eller elimineras. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/8988659
- author
- Alfredsson, Viktor LU ; Naderi, Rebin LU and Ferenius, Jakob LU
- supervisor
-
- Burak Tunca LU
- organization
- alternative title
- Introducing a novel method to derive brand association networks from online customer reviews
- course
- FEKH29 20191
- year
- 2019
- type
- M2 - Bachelor Degree
- subject
- keywords
- Brand associations, brand association networks, text mining, customer reviews, machine learning
- language
- English
- id
- 8988659
- date added to LUP
- 2019-07-02 11:57:16
- date last changed
- 2019-07-02 11:57:16
@misc{8988659, abstract = {{Title: Text mining into customers’ minds Seminar date: 4 June 2019 Course: FEKH29 Business Administration: Bachelor Degree Project in Marketing Undergraduate Level, 15 credits Authors: Viktor Alfredsson, Jakob Ferenius & Rebin Naderi Advisor: Burak Tunca Keywords: Brand associations; brand association networks; text mining; customer reviews; machine learning Purpose: To develop a text mining method to derive brand association networks from unstructured online customer reviews and to explore how this method can assist managers in developing more effective brand management strategies. Methodology: Customer reviews were collected using a web extraction system. The reviews were processed using text mining techniques, including tokenization, stop words removal, lemmatization, and feature extraction. This resulted in numerous brand associations for each brand, which were classified along the dimensions of type, strength, valence, uniqueness, and relevance. The associations’ relationships with each other, and with the brand, were modeled through networks. The derived brand association networks were analyzed between the hotel brands, as well as between satisfied and unsatisfied customers. To generate actionable managerial insights, an importance-performance analysis was conducted to identify the most valuable associations for a hotel brand to focus on. Theoretical perspectives: The theoretical foundation of the thesis lies in theory on brand associations and customer based brand equity. The conceptualization of brand associations included five dimensions: type, valence, strength, uniqueness, and relevance. Empirical foundation: The study focused on the hotel industry. 58,709 customer-generated hotel reviews for Hilton, Marriott, Wyndham, and Holiday Inn hotels in New York were collected from Hotels.com. Conclusions: Several brand associations such as room, hotel, staff, stay, location, clean, and friendly could be derived with the methodology. The derived brand associations were further organized through brand association networks. Network analyses showed that some of the associations were more central and influential than others, thereby illustrating how managers can use these networks for strategic decision making. Brand associations also formed clusters in the networks. For instance, a cluster of room-related associations was discovered, as well as a staff-cluster and a location-cluster. Overall, the associations in the room-cluster had a negative impact on hotel ratings, while the ones in the location- and staff-clusters had a positive impact. Results also showed that associations that were unique to each brand could be identified, which would allow managers to adjust their branding strategies with regard to competitors. Brand association networks analyzed separately for satisfied and unsatisfied customers showed that satisfied customers talked more about the intangible aspects of the hotel service offerings like friendly, comfortable, and convenient, while unsatisfied customers talked more about the tangible aspects such as room, bed, and desk. Finally, we demonstrated that managers can use the importance-performance matrix to decide which associations should be improved, maintained, decreased, or eliminated.}}, author = {{Alfredsson, Viktor and Naderi, Rebin and Ferenius, Jakob}}, language = {{eng}}, note = {{Student Paper}}, title = {{Text mining into customers’ minds}}, year = {{2019}}, }