TL;DR
Researchers have created XRZoo, the first large-scale, curated dataset of over 12,500 Extended Reality (XR) applications. This comprehensive resource aims to address the critical lack of data for XR software engineering and security research, enabling more robust and reproducible studies across diverse XR platforms and use cases.
What Problem Does This Paper Address?
The rapid growth of Extended Reality (XR) technologies, including Augmented Reality (AR), Mixed Reality (MR), and Virtual Reality (VR), is paving the way for the Metaverse and revolutionizing fields like healthcare, education, and manufacturing. However, advancing research in XR software engineering and security has been significantly hampered by a lack of large, representative, and high-quality datasets. Existing research often relies on small, potentially outdated, or non-representative samples, leading to biased findings and difficulties in reproducing results. The fragmented nature of XR app stores and development platforms further complicates data collection, creating a significant barrier to empirical XR research.
Methods
The authors developed a systematic approach to collect and curate XR applications from nine leading marketplaces. This involved building specialized crawlers to navigate diverse app store architectures, overcome API limitations, and bypass rate restrictions. They employed sophisticated filtering algorithms, including explicit tag analysis and deeper inspection of application binaries for XR-specific libraries, to accurately identify XR applications within general-purpose stores. Metadata aggregation across platforms was a key challenge, requiring significant effort to synthesize and normalize inconsistent data formats. The dataset includes detailed metadata such as application descriptions, categories, release dates, user review counts, and hardware specifications.
Datasets
The paper introduces XRZoo, a newly created dataset comprising 12,528 free XR applications. These applications are sourced from nine distinct app stores:
- Meta Horizon Store & App Lab
- Vision Pro App Store
- VIVEPORT
- Microsoft Store
- SideQuest
- Steam
- Google Play
- iOS App Store
This dataset is unique in its scale and breadth, covering AR, MR, and VR applications across various XR devices (standalone, PC-assisted, and mobile) and use cases.
Key Insights and Ideas
- The "data desert" in XR research is real: The paper highlights a significant gap in available data for XR software engineering, which is crucial for progress in this rapidly evolving field.
- XR fragmentation is a major hurdle: The diversity of XR platforms, hardware, and app stores creates substantial challenges for data collection and standardization.
- XRZoo is a foundational resource: By providing a large, curated dataset, XRZoo aims to democratize XR research, enabling reproducible studies and fostering innovation.
- Temporal trends reveal XR's evolution: Analysis of release dates shows significant growth in XR applications, particularly post-2020, with distinct patterns across different platforms.
- User engagement varies significantly: Review statistics reveal stark differences in user feedback across platforms, with Vision Pro leading in average reviews, while Meta Horizon and Viveport show minimal engagement.
Key Results
The XRZoo dataset contains 12,528 free XR applications. The data collection spanned from October 2024 and covers applications from nine major app stores. The dataset includes metadata such as application descriptions, categories, release dates, user review numbers, and hardware specifications. Temporal analysis shows a surge in XR application releases, with peaks reflecting growth in AR, VR, and MR. Category distribution indicates entertainment as the dominant use case, followed by education and productivity. User review statistics show a wide range of engagement across platforms, with Vision Pro apps having the highest average reviews.
Limitations and Future Work
The authors acknowledge that XRZoo currently focuses on free applications and does not include proprietary or paid apps. Future work will involve expanding the dataset with richer metadata and longitudinal updates. The dataset is actively maintained and publicly released to encourage further research and development in the XR domain.
Summary updated