SemanticTraj: A New Approach to Interacting with Massive Taxi Trajectories

Area of Science:

• Data Science
• Urban Planning
• Transportation Science

Background:

• Traditional taxi trajectory analysis tools require manual geospatial selection, making complex queries time-consuming.
• Linking GPS data to semantic information like street names is a significant challenge in existing systems.

Purpose of the Study:

• To introduce SemanticTraj, a novel method for semantic management and visualization of taxi trajectory data.
• To enable intuitive, keyword-based querying and interactive exploration of taxi trip information.

Main Methods:

• A textualization process converts GPS points into semantic documents, including street/POI names and descriptive speed terms.
• A corpus of taxi documents is indexed for flexible semantic queries using a text search engine.
• Semantic labels and meta-summaries are integrated with visualizations in the SemanticTraj prototype.

Main Results:

• SemanticTraj facilitates easy answering of complex queries, such as trips between specific locations and times.
• The system provides enhanced visualizations for quick and easy study of taxi trajectories.
• User feedback and a preliminary study indicate the system's usability and effectiveness.

Conclusions:

• SemanticTraj offers an efficient and intuitive approach to discovering knowledge from large-scale taxi trajectory datasets.
• The semantic enrichment and visualization capabilities significantly improve the accessibility and utility of trajectory data for domain experts and the public.