Location-Based Social Networks Chapter 8 and 9 of the book Computing with Spatial Trajectories Yu Zheng and Xing Xie Microsoft Research Asia Outline • Chapter 8 (Location-based social networks: Users) – – – – Concepts, definition, and research philosophy Modeling user location history Computing user similarity based on location history Friend recommendation and community discovery • Chapter 9 (Location-based social networks: Locations) – Generic travel recommendations • Mining interesting locations and travel sequences • Trip planning and itinerary recommendation • Location-activity recommendation – Personalized travel recommendation • User-based collaborative filtering • Item-based collaborative filtering • Open challenges Social Networks “A social network is a social structure made up of individuals connected by one or more specific types of interdependency, such as friendship, common interests, and shared knowledge.” 3 Social Networking Services A social networking service builds on and reflects the real-life social networks among people through online platforms such as a website, providing ways for users to share ideas, activities, events, and interests over the Internet. 4 Locations Location-acquisition technologies Outdoor: GPS, GSM, CDMA, … Indoor: Wi-Fi, RFID, supersonic, … Representation of locations Absolute (latitude-longitude coordinates) Relative (100 meters north of the Space Needle) Symbolic (home, office, or shopping mall) Forms of locations Point locations Regions Trajectories 5 Locations + Social Networks Add a new dimension to social networks Geo-tagged user-generated media: texts, photos, and videos, etc. Recording location history of users Location is a new object in the network Bridging the gap between the virtual and physical worlds Sharing real-world experiences online Consume online information in the physical world 6 Examples Virtual world Interactions Sharing & Understanding Generating & Consuming Physical world 7 Location-Based Social Networks Locations Sharing Geo-tagged media Virtual Physical worlds Understanding User interests/preferences Location property User-user, location-location, user-location correlations An new dimension: Geo-tag An new object Social networks Expanding social structures Recommendations Users Locations media Sharing Locations Understanding Social networks 8 Scenarios - Sharing Data + Intelligence Microsoft Services Third Party Services Scenarios - Understanding Data Information Knowledge Intelligence Data + Intelligence Microsoft Services Third Party Services Location-Based Social Networks (LBSN) not only mean adding a location to an existing social network so that people in the social structure can share location-embedded information, but also consists of the new social structure made up of individuals connected by the interdependency derived from their locations in the physical world as well as their location-tagged media content Here, the physical location consists of the instant location of an individual at a given timestamp and the location history that an individual has accumulated in a certain period. The interdependency includes not only that two persons co-occur in the same physical location or share similar location histories but also the knowledge, e.g., common interests, behavior, and activities, inferred from an individual’s location (history) and location-tagged data. From Book “Computing With Spatial Trajectories” 11 Categories of LBSN Services Geo-tagged-media-based Geo- Point-location-driven Trajectory-centric LBSN Services Focus Real-time Information Geo-tagged-media-based Media Normal Poor Point location Instant Normal Trajectory Relatively Slow Rich Point-location-driven Trajectory-centric 12 Research Philosophy User Graph Users User-Location Graph User Correlation Trajectories Locations Location Correlation Location-tagged user-generated content Location Graph 13 Research Philosophy Sharing Making sense of the data Effective and efficient information retrieval …… 14 Repla y Shar e Replay travel experiences on a map with a GPS trajectory 15 16 Research Philosophy User Graph Understanding Understanding users Understanding locations Understanding events Location Graph 17 Understanding Users (Chapter 8) User similarity/ link prediction Experts/Influencers detection Community Discovery 18 Understanding Locations (Chapter 9) Generic recommendation Most interesting locations and travel routes/sequences Itinerary planning Location-activity recommenders Personalized recommendation Location recommendations User-based collaborative filtering model Item-based collaborative filtering model Open challenges 19 Understanding Events Anomaly Crowd Behavioral Patterns 20 Mining User Similarity Based on Location History 21 GIS ‘08/Trans. On the Web Grouping users in terms of the similarity between their location histories, and conduct personalized location recommendations. 22 Mining User Similarity Based on Location History Model user location history Geographic spaces Semantic spaces User similarity Museum 1 Cinema 2 Semantic Location history Coffee 3 Geo-Location history GPS trajectories 23 Mining User Similarity Based on Location History Computing user similarity Hierarchical properties Sequential properties Popularity of a location c10 1 High {C } c20 c21 S1 A c30 Low c31 c32 c33 2 2 3 4 5 2 0.5 1 1 B 4 1 c34 S2 A 2 B C 0.5 0.5 4 D D 5 2 0.5 C E 6 2 3.5 2 E 7 F 1 2 F 2 G 𝐴 → 𝐵 → 𝐶, 𝐴 → 𝐵 → 𝐷 → 𝐸 → 𝐹 Stands for a stay point S Stands for a stay point cluster cij 24 GPS Logs of User 1 GPS Logs of User i GPS Logs of User 2 GPS Logs of User i+1 GPS Logs of User n-1 GPS Logs of User n 1. Stay point detection 2. Hierarchical clustering GPS Logs of User 1 GPS Logs of User 2 Layer 1 Layer 1 c10 G1 Layer 2 G2 High G1 {C } c20 c21 A A B Low c c30 c31 c32 c33 B c34 Layer 3 G3 3. Individual graph building Layer 3 Low e a Layer 2 G2 High Stands for a stay point S Stands for a stay point cluster cij Shared Hierarchical Framework e a d b G3 Friend and Location Recommendation Similar Users Retrieval Ranking Locations Location Candidates Discovering User taste inferring u1 u2 . . un L1, L2, …., Ln x1, x2, …, xn y1, y2, …, yn . . z1, z2, …, zn 26 Mining interesting locations and travel sequences from GPS trajectories 27 Mining interesting locations, travel sequences, and travel experts from user-generated travel routes 28 Users: Hub nodes The HITS-based inference model Locations: Authority nodes 29 Location-Activity Recommendation Goal: To Answer 2 Typical Questions Q1: what can I do there if I visit some place? (Activity recommendation given location query) Location query Recommended activity list A recommended location Activity query Q2: where should I go if I want to do something? (Location recommendation given activity query) Recommended location list 31 Problem Data sparseness (<0.6% entries are filled) Exhibition Shopping Forbidden City 5 ? ? Bird’s Nest ? 1 ? Zhongguancun 1 ? 6 Activities Locations Tourism ? 32 Solution • Collaborative filtering with collective matrix factorization U Y= UWT Activities V X = UVT Activities Activities Locations Locations Features Z = VVT – Low rank approximation, by minimizing where U, V and W are the low-dimensional representations for the locations, activities and location features, respectively. I is an indicatory matrix. 33 Research Philosophy User Graph Users User-Location Graph User Correlation Trajectories Locations Location Correlation Location-tagged user-generated content Location Graph 34 New Challenges in LBSNs Papers Heterogeneous networks Locations and users Geo-tagged media and trajectories Authors Conferences Special properties Hierarchy / granularity Sequential property Fast evolving Easy to access a new location User experience/knowledge changes Media Users Locations 35 GeoLife Trajectory Dataset (1.1) Transportation mode Distance (km) Duration (hour) Walk 11,457 5,126 Bike 6,335 2,304 Bus 21,931 1,430 Car & taxi 34,127 2,349 Train 74,449 459 Airplane 28,493 37 Other 10,886 335 Total 187,679 12,041 Version 1.0 Version 1.1 Incremental Time span of the collection 04/2007 – 08/2009 04/2007 – 12/2010 +16 months Number of users 155 167 +12 Number of trajectories 15,854 17,355 +1,501 Number of points 19,304,153 22,294,264 2,990,111 Total distance 600,917 km 1,070,406 km +469,489 km Total duration 44,776 hour 48,349 hour +3,573 hour Effective days 8,977 9,694 +717 Link to the data Conferences ACM SIGSPATIAL Workshop on Location-Based Social Networks LBSN 2011: Nov. 1, 2011, in Chicago (3rd year) Over 40 attendees this year 26 submissions. 10 full papers and 4 short papers 38 Summary Locations and social networks Sharing and understanding New challenges and new opportunities 39 Thanks! Yu Zheng yuzheng@microsoft.com 40