Browse Prior Art Database

A System & Method For Capturing User-specific Signatures from Spatio-Temporal Traces

IP.com Disclosure Number: IPCOM000241034D
Publication Date: 2015-Mar-20
Document File: 5 page(s) / 279K

Publishing Venue

The IP.com Prior Art Database

Abstract

On a per use basis, given a continuous stream of spatio-temporal data (CDR traces, GPS traces, etc.), our proposed system generates a compact representation (signature) that captures hangouts and movement patterns of each user distinctly, where Hangouts indicate the time instances at which he was present at specific locations (these locations can be absolute lat-long values or discretized space boxes obtained from a spatial index) and Movement patterns indicate transition times from one hangout to another. This signature could be stored natively in a hashmap or relevant index and can be visualized if necessary (for discoverability).

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 29% of the total text.

Page 01 of 5

A System & Method For Capturing User-specific Signatures from Spatio-Temporal Traces

Problem : Existing techniques for processing spatio-temporal data do not provide a compact representation on a user-centric basis that indicate the user's preferred hangouts and movement

patterns over time.

Proposal : On a per use basis, given a continuous stream of spatio-temporal data (CDR traces, GPS traces, etc.), our proposed system generates a compact representation (signature) that captures hangouts and movement patterns of each user distinctly. This signature could be stored natively in a hashmap or relevant index and can be visualized if necessary (for discoverability).

Known Solutions & Their Drawbacks: Existing works in this domain can be classified and differentiated from our proposed system as follows: 1. Monitoring Users in Urban Cities: These works [1,2,3] highlight the aspect of indexing and visualizing the aggregate spatio-temporal behavior across all users (i.e., non-user-centric representation). Further, these works primarily discuss spatio-temporal indices which map users to them (without distinguishing between users). Instead identifying space-time boxes where large number of users might gather/travel is the goal. In contrast, out model is more user-centric indexing wherein we are trying to identify the hangouts/movement patterns of individual users.

2. Trajectory Reconcilliation: This method [4] describes a method of correlating track data from two or more data sources by extracting the trajectories of moving entities to identify

possible matches. However, this method considers only a single movement pattern/trajectory at any time with space-time boxes. Further, no visualization or storage-space reduction is considered. In contrast, our notion of a signature would take several such traces as input and generate an aggregate ST-Signature which would visualize the movement patterns over a

period of time. In addition, our representation captures the entire trace with almost 90% saving in storage space without any loss of data.

3. Trajectory Clustering: These methods [5,6,7,8] typically group similar trajectories of all users to identify commonalities. Even if it was restricted to clustering all trajectories of a single user, it would just show a representation of the users' movement patterns. However, hangout patterns are not considered nor is meta-data (location and time information) appended to the cluster formation. In contrast, our representation shows a concise representation of the users movement and hangout patterns without losing the individuality element associated with each trajectory.

4. Frequent Pattern Mining from Trajectories: Existing techniques [9] mines trajectories that visit the same sequence of places with similar transition times. However, these representations are not interested in capturing the complete trace; instead just important

patterns within the trace. Further it does not account for hangout informa...