Upload
daniele-quercia
View
2.360
Download
0
Embed Size (px)
DESCRIPTION
Recommending Social Events from Mobile Phone Location DataA city offers thousands of social events a day, and it is difficult for dwellers to make choices. The combination of mobile phones and recommender systems can change the way one deals with such abundance. Mobile phones with positioning technology are now widely available, making it easy for people to broadcast their whereabouts; recommender systems can now identify patterns in people’s movements in order to, for example, recommend events. To do so, the system relies on having mobile users who share their attendance at a large number of social events: cold-start users, who have no location history, cannot receive recommendations. We set out to address the mobile cold-start problem by answering the following research question: how can social events be recommended to a cold-start user based only on his home location?To answer this question, we carry out a study of the rela- tionship between preferences for social events and geography, the first of its kind in a large metropolitan area. We sample location estimations of one million mobile phone users in Greater Boston, combine the sample with social events in the same area, and infer the social events attended by 2,519 residents. Upon this data, we test a variety of algorithms for recommending social events. We find that the most effective algorithm recommends events that are popular among residents of an area. The least effective, instead, recommends events that are geographically close to the area. This last result has interesting implications for location-based services that emphasize recommending nearby events.
Citation preview
_ nets & the city _
Use mobility data …
… to recommend social events
mobile phone location data
location estimations
lessons
1. infer attendace at events
2. recommend in 6 ways
location estimations
lessons
3. measure “quality”
time
distance
1m users (20% population)sample 80K
1. infer attendace at events
location estimations
attendance
distance
1. infer attendace at events
location estimations
attendance
time
1. infer attendace at events
location estimations
attendance
resolutions: time (1 ½ h) & space (350m)
1. infer attendace at events
location estimations
attendance
1. infer attendace at events
location estimations
attendance
it’s not about single individuals. it’s about areas
1. infer attendace at events
location estimations
attendance
On input of area of residence: 1. popular events 2. geographically close3. popular in area of residence4. TF-IDF (similar to 3 expect for less-attended events)5. K-Nearest Locations6. K-Nearest Events
2. recommend in 6 ways
attendance
ranked recommendations
ShakespeareRed Sox
You went to…
lessons
3. measure “quality”
ranked recommendations
ShakespeareRed Sox
You went to…
lessons
3. measure “quality”
ranked recommendations
ShakespeareRed Sox
You went to…
1. Shakespeare2. Cirque…5. Red Sox
1. Shakespeare2. Red Sox…5. Cirque
lessons
3. measure “quality”
ranked recommendations
ShakespeareRed Sox
You went to…
1. Shakespeare2. Cirque…5. Red Sox
1. Shakespeare2. Red Sox…5. Cirque
average percentile rankingHigh Low
lessons
3. measure “quality”
ranked recommendations
lessons
3. measure “quality”
ranked recommendations
Lesson 1: geographically close isn’t the best ;-)
lessons
3. measure “quality”
ranked recommendations
lessons
3. measure “quality”
ranked recommendations
Lesson 2: popular in area rocks ;-)
lessons
3. measure “quality”
ranked recommendations
lessons
3. measure “quality”
ranked recommendations
Lesson 3: geographical patterns matter ;-)
lessons
3. measure “quality”
ranked recommendations
geographical patterns matter
geographically close isn’t the best
‘popular in area’ rocks
Future
Future 1| differential privacy
SpotME if you can
fake your location yet aggregate location data is still OK
promoting location privacy… one lie at a time
Future 2| social nets & space