Exploiting Web Analytics Tracking for Bootstrapping a Case-based Recommender System
1. ENTER 2015 Research Track Slide Number 1
Exploiting Web Analytics Tracking
for Bootstrapping a
Case-based Recommender System
Paolo Massa(a)
, Michela Ferron(a),
Adriano Venturini(b)
(a)Fondazione Bruno Kessler, Italy
(b)ECTRL Solutions SRL, Italy
venturini@ectrlsolutions.com
http://www.ectrlsolutions.com
http://www.fbk.eu
2. ENTER 2015 Research Track Slide Number 2
Contents
• The Travel Monitor Project
• Develop a tool to monitor the travel
planning process of the user
• How collected data can be used to
bootstrap the recommender system
• Current status and future works
3. ENTER 2015 Research Track Slide Number 3
Travel Monitor
• Tool to analyze the users’ behviours while planning a
travel, when he is interacting with the hotel or
destination websites, or in place using a mobile
devices or local infopoints positioned on site.
• Exploit navigation data to understand:
• Effectiveness of the used systems
• Users’ interests
• Choices and decisions
• Buying behviours
4. ENTER 2015 Research Track Slide Number 4
Travel Monitor: goals
Understand the users:
To identify the different characteristics, needs
Understand the travel planning process:
To define more interesting offers
planning marketing campaign
Indentify strong and weak points of the online presence
To support adaptive behviour of the system
Possiblity to exploit the collected data to initialze
the knowledge base of our recommender
system
10. ENTER 2015 Research Track Slide Number 10
Monitoring the user’s
query and results
11. ENTER 2015 Research Track Slide Number 11
From the users’ query we can
understand:
Which are the most interesting categories ?
Does the user find what is looking for ?
It helps DMO to plan marketing of the
destinations, to understand weakness of the
tourist offers
To define categories for structuring their
contents in the CMS
16. ENTER 2015 Research Track Slide Number 16
Analysing travel planner data
DMOs can get:
Profiles of their visitors and their distributions
Which are the most interesting products
according to different types of travelers
Travel planning choices:
which items they select when they plan a
travel
23. ENTER 2015 Research Track Slide Number 23
Suggesto Recommendation
technology
• Derives from the Trip@dvice case-based
reasoning recommender (Ricci et al, 2006)
• It utilizes travel plans (cases) built by other
users in the past to identify possible items
the user is interested in
24. ENTER 2015 Research Track Slide Number 24
Case Model
• Collaborative features:
• Travel Party
• Travel Budget
• Travel Interests
• Chosen items in the travel plans
• Attractions
• Events
• Accommodations
• Interests
• The system exploits similarity among the collaborative
features of past travels to identify items which could fit the
travel plan planned by the current user
25. ENTER 2015 Research Track Slide Number 25
Bootstrap the recommender
To initialze the casebase with an initial set of
«good» cases:
• Representative of the visitors
• With the correct associations between
collaborative features and selected items
26. ENTER 2015 Research Track Slide Number 26
Mapping analytics data to the
collaborative features
From booking search (number of persons)
Travel Party
From booking search (type of accommodations)
Budget
From search of activities
Travel interests
From visited pages
preferred items
28. ENTER 2015 Research Track Slide Number 28
Project status and future
activities
Developed the Travel Monitor tool
Collecting analyitical data
Analysing how systems are used
Described the methodology for bootstrapping
the recommender
Next step is to implement and evaluate the
indentified methodology
29. ENTER 2015 Research Track Slide Number 29
Thanks!
Acknowledgment: This work has been partially funded by
Travel Monitor Project, FESR 2011 funding programme,
CUP C67I12000030008.