Please enable JavaScript.
Coggle requires JavaScript to display documents.
Chapter 5 - Personalization (Adoptions of Recommendations in Real World…
Chapter 5 - Personalization
Adoptions of Recommendations in Real World
Google announced Search plus Your World (SpYW)
Netflix Example
Hunch Example
Last.fm
Amazon Goodreads
YouTube suggestions
Impact on online and offline behavior
Benefits
Effectiveness of Marketing Campaigns
Make choosing a pleasant experience (instead of "Choice Overload")
Consumption increases
Habits formation occurs
Users feel more engaged
Recommendations
Challenges
Google has no info from FB
FB has no info from Google
-> personalization could be much bigger
Abuse: FB Emotion Experiment. What happens if they try to manipulate us for their reasons?
I don't see what you see
User Modeling
Process in General
how to obtain, understand, exploit information about the user
inferring information about user and representing it in the software
which data is relevant for inferring information about the user
People leave traces on the Web and their computers
not only user's behavior but also interaction with other users
User Modeling Approaches
Customizing
Overlay User Modeling
User model elicitation
Stereotyping
User relevance modeling
e.g. Google
How to model via Twitter
Semantic Enrichment -> link to Semantic Web
Evaluation strategies: How to determine successful personalization
user perspective
provider perspective
Metrics
IR metries
rankings
Recommendations
RecSys Issues
Digital Bubble
Cold start problem
Changing user preferences
Sparsity problem (new item problem)
Lack of Diversity (overfitting)
Use the right context
Collaborative Filtering
Memory vs model based