- [Nov. 2023] Our study Evaluating geospatial context information for travel mode detection was published in Journal of Transport Geography. We identify common context representations and propose an analytical pipeline to assess the contribution of geospatial context information for travel mode detection. We highlight the importance of network features and also suggest that the majority of land-use features do not significantly contribute to the task.
- [Aug. 2023] Our study Context-aware multi-head self-attentional neural network model for next location prediction was published in Transportation Research Part C. We present a neural network that incorporates location visits, visit times, activity durations, and land use functions to predict an individual’s next location. We also explore key modeling decisions that significantly influence prediction performance.
- [Jul. 2023] Our paper Influence of tracking duration on the privacy of individual mobility graphs was published in Journal of Location Based Services. We quantified the influence of tracking duration on privacy when representing user mobility histories as location graphs.
- [Jul. 2023] Our paper Predicting mobile users’ next location using the semantically enriched geo-embedding model and the multilayer attention mechanism was published in Computers, Environment and Urban Systems. We proposed a next location prediction model that combines location and spatiotemporal information, which achieved state-of-the-art prediction performance.
- [Jul. 2023] We have a new short paper Predicting visit frequencies to new places accepted at GIScience 2023. We proposed a new mobility prediction problem: Predicting the frequency of future visits to a newly visited location, given the historic mobility of a single user.
- [Jan. 2023] Our paper Trackintel: An open-source Python library for human mobility analysis was published in Computers, Environment and Urban Systems. Check out the Python trackintel library for human mobility analysis on GitHub!
- [Dec. 2022] Our study Conserved quantities in human mobility was published in Transportation Research Part C. We showed that individuals implicitly balance new behaviour exploration and existing option exploitation over time, resulting in a conserved number of basic travel mode and activity location combinations.
- [Nov. 2022] We presented our study How do you go where? improving next location prediction by learning travel mode information using transformers at SIGSPATIAL ‘22 in Seattle, USA. Our proposed transformer decoder-based model utilizing location, travel mode and time-related information for next location prediction achieved state-of-the-art performance on large scale tracking datasets.
