Spatio-Temporal Linkage over Location-Enhanced Services

Abstract

We are witnessing an enormous growth in the volume of data generated by various online services. An important portion of this data contains geographic references, since many of these services are location-enhanced and thus produce spatio-temporal records of their usage. We postulate that the spatio-temporal usage records belonging to the same real-world entity can be matched across records from different location-enhanced services. Linking spatio-temporal records enables data analysts and service providers to obtain information that they cannot derive by analyzing only one set of usage records. In this paper, we develop a new linkage model that can be used to match entities from two sets of spatio-temporal usage records belonging to two different location-enhanced services. This linkage model is based on the concept of k-l diversity -that we developed to capture both spatial and temporal aspects of the linkage. To realize this linkage model in practice, we develop a scalable linking algorithm called ST-Link, which makes use of effective spatial and temporal filtering mechanisms that significantly reduce the search space for matching users. Furthermore, ST-Link utilizes sequential scan procedures to avoid random disk access and thus scales to large datasets. We evaluated our work with respect to accuracy and performance using several datasets. Experiments show that ST-Link is effective in practice for performing spatio-temporal linkage and can scale to large datasets.

Towards Matching User Mobility Traces in Large-Scale Datasets

“…The work of Cecaj et al [16] is more similar to ours, but is also applied to a more limited data source, limiting the potential statistical analysis. We believe that our work is most similar to that of Basık et al [18] who describe a very similar matching procedure augmented by an efficient spatial pre-filtering to reduce complexity. In our case, we believe such pre-filtering would not be applicable as our data comes from a compact but densely populated area, thus all users' trajectories would span the same units when constructing a coarse spatial index for such pre-filtering.…”

“…However, given the scale of the dataset, we will have a realistic estimate on the probabilities of finding false positive matches, highly unlikely when using a dataset containing only a few thousand trajectories. In this respect, our work is most similar to Basık et al [18,42], who utilize two large datasets as well (containing of several hundred thousand and a few million users respectively), but do not present relevant statistics about the matches between these two datasets due to the lack of ground truth data. They do present an extensive analysis on the quality of matching between a synthetic dataset generated from CDR data and the original CDR dataset.…”

“…Following Basık et al, we also use the term alibi to refer to impossible matches [18]. We use the term temporal match for pairs of points which are either a spatial or impossible match (i.e.…”

SLIM: Scalable Linkage of Mobility Data

“…It normalizes the similarity scores based on the size of the histories in terms of the number of time-location bins. Moreover, while it does not penalize the score when one entity has activity in a particular time window but the other does not, it does penalize the existence of cross-dataset activities that are close in time but distant in space (aka alibis [2,3]). This is an essential property that supports mobility linkage.…”

“…• Evaluation. We perform extensive experimental evaluation using real-world datasets, compare SLIM with two existing approaches (STLink [3], GM [43]), and show superior performance in terms of accuracy and scalability.…”