Name
Title | ||
|---|---|---|
Location Prediction Through Activity Purpose: Integrating Temporal and Sequential Models. |
Abstract | ||
|---|---|---|
Based on the growing popularity of smart mobile devices, location-aware services become indispensable in human daily life. Location prediction makes these services more intelligent and attractive. However, due to the limited energy of mobile devices and privacy issues, the captured mobility data is typically sparse. This inherent challenge deteriorates significant principles in mobility modeling, i.e. temporal regularity and sequential dependency. To tackle these challenges, by utilizing temporal regularity and sequential dependency, we present a location prediction model with a two-stage fashion. Firstly, it extracts predictive features to effectively target the better performer from sequential and temporal models. Secondly, according to the inferred activity, it adopts non-parametric Kernel Density Estimation for posterior location prediction. Extensive experiments on two public check-in datasets demonstrate that the proposed model outperforms state-of-the-art baselines by 10.1% for activity prediction and 12.9% for location prediction. |
Year | Venue | Field |
|---|---|---|
2017 | PAKDD | Data mining,Computer science,Popularity,Temporal models,Mobile device,Artificial intelligence,Location prediction,Machine learning,Kernel density estimation |
DocType | Citations | PageRank |
Conference | 1 | 0.34 |
References | Authors | |
12 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Dongliang Liao | 1 | 8 | 2.44 |
| Yuan Zhong | 2 | 259 | 21.23 |
| Jing Li | 3 | 22 | 6.73 |