Graph-based Point-of-interest Recommendation with Geographical and - - PowerPoint PPT Presentation

▶
graph based point of interest recommendation with
SMART_READER_LITE
LIVE PREVIEW

Graph-based Point-of-interest Recommendation with Geographical and - - PowerPoint PPT Presentation

Apr 03, 2023 123 likes •374 views

Graph-based Point-of-interest Recommendation with Geographical and Temporal Influences Date : 2015/05/07 Author: Quan Yung, Gao Cong, Aixin Sun Source: ACM CIKM14 Advisor: Jia-ling Koh Speaker: Han, Wang 1 Outline Introduction

slide-1
SLIDE 1

Graph-based Point-of-interest Recommendation with Geographical and Temporal Influences

Date : 2015/05/07 Author: Quan Yung, Gao Cong, Aixin Sun Source: ACM CIKM’14 Advisor: Jia-ling Koh Speaker: Han, Wang

1

slide-2
SLIDE 2

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

2

slide-3
SLIDE 3

Introduction

Motivation:

  • The availability of user check-in data offers the opportunity to facilitate

users’ travels and social interactions.

Goal: Recommend POIs to users who have not visited before.

3

slide-4
SLIDE 4

Introduction

Two observation:

  • Geographical Influence
  • Temporal Influence

2 KM 10 KM

4

slide-5
SLIDE 5

Introduction

Framework:

5

Input:

Check- ins, (Users, time, POI)

  • utput:

Top-k not visited POIs

POIs Recommendation

Step 1: Construct GTAG Step 2: Weight computation

  • f edges

Step 3: Path Selection Step 4: Breadth-first Preference Propagation

Input:

target user, target time, GTAG, k

slide-6
SLIDE 6

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

6

slide-7
SLIDE 7

Approach

Temporal influence on check-ins:

  • 7
  • time difference
  • daily periodic
  • users tend to visit the same

POIs at close time

slide-8
SLIDE 8

Approach

Geographical-Temporal Influence Aware Graph(GTAG):

1. User’s interests vary with time. 2. The check-in interests of a user in the time closer to the target time are more relevant, and more important. 3. If two users have similar temporal interests(the POIs he visited in that time) in two time, they tend to visit the same POIs in the two time. 4. Users tend to visit nearby POIs.

8

Holly 20:00

Silks

Palace

National Palace Museum

time slot correspond to time slot POI links Check-in links

slide-9
SLIDE 9

Approach

Geographical-Temporal Influence Aware Graph(GTAG):

9

slide-10
SLIDE 10

Approach

Weight of edges: (initial = 1 )

10

S2,18:00

L1

S2,21:00 S1,17:00

L2 L3 L4 Ｕ1 Ｕ2 Ｕ S # visit S L

target time = 18:00 H = 2

1 2 1 1 1 1

slide-11
SLIDE 11

Approach

Weight of edges: (initial = 1 )

11

S22 L1 S25 S11 L2 L3 L4 Ｕ1 Ｕ2 L L

α = 1, β = -1

1km 2km 4km 1 1/2 1/4

slide-12
SLIDE 12

Approach

Normalize Weight:

12

S22 L1 S25 S11 L2 L3 L4 Ｕ1 Ｕ2

target time = 18:00 = 0.2, = 0.3 Ex: L1-> S11 : 1/0.3*1+1 = 0.76 L1-> L2 : 0.3*1/0.3*1+1 = 0.23 S11-> U1 : 1/0.2*2+1 = 0.71 S11-> L1 : 0.2/0.2*2+1 = 0.14 U1-> S11 : / = 1

0.71 0.71 1 0.66 0.83 0.33 0.17 0.93 0.07 0.07 0.23 0.14 0.76 0.14 0.46 0.46 0.14 0.14 0.93 0.07

slide-13
SLIDE 13

Approach

Path Selection:

  • Simple path, no repeated node in a path.
  • The path only contain one visited POI node and session node of the target

user, for avoiding long propagation paths.

  • The path terminates when an unvisited POI node met.

The path always start form a target user, valid path length is 3,4 or 6.

13

slide-14
SLIDE 14

0.71 0.71 1 0.66 0.83 0.33 0.17 0.93 0.07 0.07 0.23 0.14 0.76 0.14 0.46 0.46 0.14 0.14 0.93 0.07

Approach

14

Breath-first Preference Propagation:(BPP)

(target user: u1, k = 1)

S22 L1 S25 S11 L2 L3 L4 Ｕ1 Ｕ2

ru1:1 rs11:1 rL1:1 rL2:1 rs22:1 rL3:1

U1 S11 L2 L3

Q:

U1 S11 L2 L3

N: Iteration 1

ru1:-0.71 rs11:2

  • rs11 = 1 + 1 * 1 = 2
  • ru1 = 0 - 1*1*0.71 = -0.71
slide-15
SLIDE 15

0.71 0.71 1 0.66 0.83 0.33 0.17 0.93 0.07 0.07 0.23 0.14 0.76 0.14 0.46 0.46 0.14 0.14 0.93 0.07

Approach

15

Breath-first Preference Propagation:(BPP)

(target user: u1, k = 1)

S22 L1 S25 S11 L2 L3 L4 Ｕ1 Ｕ2

rL1:1 rL2:1 rs22:1 rL3:1

U1 S11 L1 L2 L3

Q:

U1 S11 L2 L3

N:

ru1:-0.29 rs11:2

Iteration 2

  • rL1 = 1 + 2 * 0.14 = 1.28
  • rs11 = 0 - 2*0.14*0.76 = -0.2128
  • rL2 = 1 + 2 * 0.14 = 1.28
  • rs11 = -0.2128 - 2*0.46*0.14 = -0.3416
  • ru1 = -0.71 + 2 * 0.71 = 0.71
  • rs11 = -0.3416 - 2*0.71*1 = -1.762

rL2:1.28 rL1:1.28 rs11:-1.762 ru1:0.71

slide-16
SLIDE 16

Approach

Breath-first Preference Propagation:(BPP)

Until Q is empty,stop. Output: L3(Recommendation)

slide-17
SLIDE 17

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

17

slide-18
SLIDE 18

Experiment

Dataset: Foursquare vs Gowalla(shut down)

  • Foursquare: 342,850 check-ins between Aug.2010 ~ Jul.2011 in Singapore
  • Gowalla: 736,148 check-ins between Feb.2009 ~ Oct.2010 in California and

Nevada

  • Remove POIs which fewer than 5 times be checked and users who check-in

fewer than 5 times.

  • 10% development set, 20% test set, 70% training set.

18

slide-19
SLIDE 19

Experiment

Impact of Parameters:

19

0.08 0.2 3 0.01 0.3 270 290

Preference propagated to user nodes H Impact of session nodes far from target time Geographical influence

slide-20
SLIDE 20

Experiment

Temporal influence:

20

slide-21
SLIDE 21

Experiment

Geographical influence:

21

slide-22
SLIDE 22

Experiment

Effect of the Length time slot:

22

slide-23
SLIDE 23

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

23

slide-24
SLIDE 24

Conclusion

Propose the GTAG to model check-in behaviors and a graph-based preference propagation algorithm for POI recommendation. The solution exploits both geographical and temporal influences. Conduct experiments over two real-world LBSN datasets, and the result outperforms than the others.

24