adaptive personalization using social networks --> chung, 2016 Flashcards
article in a nutshell
mobile automated adaptive personalization systems for news consumption
that take advantage of social network data
may be a promising approach to making personalization more effective
past marketing approaches
companty –> customer
standardized product
spray and pray
today’s marketing
company <–> customer
aggregated information - personalized product
engagement marketing
engagement marketing
listen, learn, engage with customers
future interaction building
feels personal at individual level
big data + personalization
proliferation of customer data
personalizes product offerings
on social networks can improve personalizations
5 characteristics of adaptive personalization systems
autonamtic through algorithms
no proactive effort by customer
better outcomes over time - learns
better than self-customization
best with info from social network
core finding
adaptive personality performs better than self-customization
customization
explicit specifications are stated by the user with full user control
personalization
implicit interests are learned by the system and content
tailord to user’s individual characteristics of preferences
customization example
e.g. creating your own playlist based on you music selection and interests
personalization example
e.g. spotify recommends music based off of music history or playlists
peer social influence is stronger for
taste products
ambiguous preferences
cold start problem
serendipity stimulation
changes in taste
fashion peer social influence example
kim kardashian wearing all black balenciaga look
other people started doing it to
4 aspects of the tool made from this study
high personalized real time
filters through feed
keywords and baysian classification algorithms
shared interest are quickly discoveres
what can they infer from the data on reading news stories
whether they skip headline
time spent reading
content of news story
adaptive persolination = better than
self-customization
adaptive personalization outperforms choosing articles based on ?
popularity
adaptive personalization systems improve when using information from peer social network
social influence (induction) > similarity (homophily)
3 recommendations for privacy
greater transparency - platforms
data minimization principle
network privacy
greater transparency - platforms
simplify privacy settings
provide clear explanations of data usage to help users make informed decisions
data minimization principle
reduce the amount of personal data collected while maintaining quality of personalized services
