Science AI Journal
Social Sciences · Research topic

Open research questions in Human Mobility and Location-Based Analysis

52 unresolved questions extracted from the limitations and future-work sections of 829 Human Mobility and Location-Based Analysis papers in our library. Each links back to the study that raised it.

What the literature leaves open

  • Therefore, our future research will focus on developing data-driven approaches to auto- matically specify these thresholds or refining the WEPI measure into a threshold-free variant.

    Mining Sustained Emerging Spatio-Temporal Co-occurrence Patterns from Dynamic Spatial Databases · 2026 · DOI

  • 6.1 Model limitation 6.1.1 Boundary effects in graph model We find higher residual ratio in some spatial units at the edge of our study area. This may be due to graph-based models having fewer node connections at the edges, which makes it difficult to effectively transmit informa- tion from the surrounding area. Therefore, for the edge regions, expanding the buffer zone outward is necessary to achieve effective predictions. 6.1.2 Potential in graph self‑supervision learning Currently, UMIFM is only trained with task supervision, despite abundant intrinsic urban regularities (e.g., spatial auto-correlation and human mobility patterns) that can lend the model to self-supervised learning. In the future, we will use self-supervision to build more generaliz- able model, improve cross-city generalization, and enable few-shot task transfer. 6.2 Experimental limitation 6.2.1 The challenge of baseline selection Currently, few methods target arbitrary multi-scale infor- mation fusion similar to UMIFM, which makes baseline selection difficult. In the future, we will continue to track developments in this area and build standardized multi- scale baselines to rigorously evaluate and improve UMIFM. 6.2.2 Comprehensive ablation experiments We did not perform ablation studies on individual fea- tures or graph encoders and instead focused on the Fig.16 GATCN Ma et al. Urban Informatics (2026) 5:7 Page 16 of 17 comparison between different scales. Nonetheless, the choice of features and graph encoders is crucial to UMIFM’s design, and we plan to investigate their con- tributions in future work. 6.2.3 Cross city transfer learning We only performed our method in Beijing under the graph transduction learning (Blum & Chawla, 2001). In this setting, information from neighboring nodes can potentially influence the results. However, this influ- ence is not same as the data leakage caused by insuf- ficient experiments or models. If the model’s intended use is cross-scale conversion only within a single city, encoding that city’s spatial context is appropriate and effectively, which can be regarded as learning “single- city features”. The longer-term goal for UMIFM is able to transfer between different cities, which can achieve the inductive learning (Hamilton et al., 2017). This requires the model to learn urban general knowledge in the absence of specific city context, thus requiring more data for training or data augmentation (e.g., sub- graph sampling). In the future, we will further focus on ’Inductive Learning’, and evaluate UMIFM’s cross-city performance.

    Urban multi-scale information graph and fusion model · 2026 · DOI

  • Beyond extracting explanatory features, future work could explore the incorporation of explicit decision hierarchies, conditional dependencies, or sequential modeling structures into the multi-task framework, in order to better reflect poten- tial ordering or conditioning relationships among travel mode, departure time, and commuting distance observed in real-world decision-making processes.

    Joint analysis of citizens’ commuting behaviors: a multi-task deep learning approach · 2026 · DOI

  • The accuracy and precision requirements of AI-generated annotations in GeoMarX for different geospatial applications (urban planning, disaster management, environmental monitoring, social services) have not been specified or validated. Benchmarking AI annotation accuracy against application-specific tolerance thresholds is needed.

    GeoMarX: empowering geospatial marking and annotation for complex targets in urban environments · 2026 · DOI

  • The paper presents a comparison method between AI-assisted point-prompted and rectangle-prompted segmentation versus manual annotation, but lacks systematic criteria for determining when to use each labeling method based on target area complexity. A formal complexity assessment framework for selecting appropriate geospatial annotation methods is missing.

    GeoMarX: empowering geospatial marking and annotation for complex targets in urban environments · 2026 · DOI

  • System performance scalability under varying server capacity and network conditions for large-scale geospatial annotation projects has not been empirically evaluated. Quantitative assessment of GeoMarX scalability thresholds (e.g., number of concurrent users, project area size, annotation volume) is needed.

    GeoMarX: empowering geospatial marking and annotation for complex targets in urban environments · 2026 · DOI

  • GeoMarX has not yet integrated heterogeneous geospatial big data sources including street view imagery, POI, road networks, and social media data despite their demonstrated effectiveness for semantic interpretation of urban environments. The technical approach for handling differences in data formats, structures, sources, and update frequencies needs specification.

    GeoMarX: empowering geospatial marking and annotation for complex targets in urban environments · 2026 · DOI

  • GeoMarX lacks human feedback mechanisms for iterative improvement of AI-generated annotations in geospatial marking tasks. Integration of user feedback loops to iteratively refine AI model performance on complex urban targets requires development and validation.

    GeoMarX: empowering geospatial marking and annotation for complex targets in urban environments · 2026 · DOI

  • The SAM model currently used in GeoMarX is pre-trained with fixed parameters not specifically tailored for geospatial tasks, limiting performance on urban environment annotation. Fine-tuning the SAM model or alternative foundation models specifically for domain-specific geospatial annotation tasks (e.g., urban informal settlements, complex building boundaries) needs to be conducted.

    GeoMarX: empowering geospatial marking and annotation for complex targets in urban environments · 2026 · DOI

  • Future research should explore place-based identity, aging-in-place, and spatial justice using the 500-m spatial forecasts to understand how shrinking communities negotiate resilience, belonging, and access to care.

    Demographic decline and resurgence in the aging century - grid-level population tendency grasped by artificial intelligence · 2026 · DOI

  • The model does not restrict relationships between variables (e.g., that TOTAL_U15 + TOTAL_1565 + TOTAL_A65 should equal TOTAL) because a suitable and efficient loss function was not found, and it is uncertain whether prediction performance would improve with such restrictions.

    Demographic decline and resurgence in the aging century - grid-level population tendency grasped by artificial intelligence · 2026 · DOI

  • While previous studies on commuting patterns during the COVID-19 pandemic have documented the change in commuter inflows to employment centers, the compositional aspects of those changes and their relations to the types of economic activities have been understudied.

    Investigating the Internal Difference in Commuting Flows within an Employment Center During COVID-19 in Seoul · 2025 · DOI

  • The findings underline the importance of incorporating temporary population models into urban planning, transport management, and emergency preparedness, and provide a methodological basis for further research on temporary populations and commuting in the context of spatial and functional urban analysis.

    Intra-settlement Redistribution of Daytime and Nighttime Population · 2025 · DOI

  • In Malaysia, Sarawak is projected to become an aging state by 2028, but the spatial patterns of senior citizen distribution remain underexplored, especially in the context of long‐term urbanization.

    Aging in Motion: Mapping the Dynamic Interplay Between Urban Growth and Senior Citizen Density in Sarawak, Malaysia (1980–2020) · 2025 · DOI

  • Understanding how geographical environments influence peoples’ subjective experiences of daily activities is of great potential for improving subjective well-being (SWB), a subject that is presently limited by a lack of available data and proper statistical methods.

    Periodicity and Variability in Daily Activity Satisfaction: Toward a Space-Time Modeling of Subjective Well-Being · 2023 · DOI

  • Few studies have used longitudinal imagery of Google Street View (GSV) despite its potential for measuring changes in urban streetscapes characteristics relevant to health, such as neighborhood disorder.

    Measuring changes in neighborhood disorder using Google Street View longitudinal imagery: a feasibility study · 2023 · DOI

  • This work uses driving licence, voter, social services, and birth records to append address locations to Unemployment Insurance data, a process that could be replicated with administrative records in other US states and countries with sporadic address data from various agencies.

    Constructing monthly residential locations of adults using merged state administrative data · 2022 · DOI

  • Our study investigated this significant gap in the literature on temporal aspects of human mobility behavior-how many days constitute a period long enough to capture individuals' highly organized activity episodes and how they vary among individuals with heterogeneous demographic and social-economic characteristics.

    How Short Is Long Enough? Modeling Temporal Aspects of Human Mobility Behavior Using Mobile Phone Data · 2019 · DOI

  • However, due to the unwillingness of suburban in-migrants to change their registration (approximately 100,000 unregistered residents), the commuting based on the census data is insufficient to capture the real pattern of daily mobility in the hinterland of Bratislava.

    Daily commuting in the Bratislava metropolitan area: case study with mobile positioning data · 2018 · DOI

  • Downloaded by [The Aga Khan University] at 03:50 18 October 2014 which will eventually be required to supply inputs for land development models of the kind now coming into general use.

    HOUSEHOLD ACTIVITY PATTERNS AND LAND USE · 1965 · DOI

  • Yet little is known about which places they surface, which they ignore, and whether these patterns vary across communities and users and translate into real-world economic consequences.

    Large language models create an uneven informational layer over cities · 2026

  • Sev- eral return trips did not meet this threshold, leading to insufficient data density for their inclusion in the subse- quent clustering analysis.

    Identifying environmental stress factors in urban cycling using multimodal human sensing and machine learning · 2026 · DOI

  • However, existing approaches typically characterize mobility patterns as static clusters or short-term variability, leaving the lifecycle dynamics of transit participation underexplored.

    Understanding Long-Term Dynamics of Individual Metro Usage: A Hidden Semi-Markov State Framework with Survival Analysis · 2026

  • The results show that retrieval augmentation provides a useful operational memory for cold-start urban demand forecasting when parametric graph generalization alone is insufficient.

    Bridge: Retrieval-Augmented Spatiotemporal Modeling for Urban Delivery Demand · 2026

  • Designing mechanisms that suppress structural patterns: re- current routes, anchor constellations, and group co-location, while preserving high-resolution spatio-temporal utility, remains an open and pressing research problem.

    How Tough Is Location Anonymization? Re-identifying 100K Real-User Trajectories in Japan · 2026 · DOI

Most-cited papers in Human Mobility and Location-Based Analysis

Most recent work

Find a gap in your own Human Mobility and Location-Based Analysis sub-topic

This page shows what the Human Mobility and Location-Based Analysis literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.

Open the Research Gap Finder →

Related topics in Social Sciences

Teacher Education and Leadership Studies
832 open questions
Criminal Justice and Corrections Analysis
784 open questions
Online and Blended Learning
745 open questions
Science Education and Pedagogy
597 open questions
Higher Education Research Studies
588 open questions
Mathematics Education and Teaching Techniques
563 open questions
Intimate Partner and Family Violence
519 open questions
Early Childhood Education and Development
486 open questions
52 open questions have been extracted from the limitations and future-work passages of 829 Human Mobility and Location-Based Analysis papers in our library. Each one below links back to the study that raised it, so you can read the original claim in context.

Tools for your next paper

Gaps & Literature SearchFind open questions in any field — 120,000+ open gaps mined from our own library.Pre-CheckIs your paper ready? A Tier 1–5 acceptance probability in seconds.Journal FinderA ranked shortlist of target journals from a 17,500-venue index.Duplicate Publication CheckerCheck prior-publication & salami-slicing risk across 8 sources.Citation GeneratorFreeA DOI or title → APA, MLA, Vancouver, Chicago, BibTeX, RIS. No signup.Graphical Abstract MakerFreeTurn your findings into a colorblind-safe graphical abstract.

Compare the category — Honest roundups of the AI research tools, ours listed alongside the alternatives.

Command palette

Jump anywhere, run any action.