Math Data
A system for representing, storing, and sharing mathematical objects and benchmarks.
- Suitable for
- Master's thesis
- Supervision
- Zafeirakis Zafeirakopoulos
Modern science produces increasingly large volumes of data consisting of mathematical (structured) objects. Despite this trend there is no standard way of representing mathematical data. Each group develops in-house formats adequate for their particular application, making data sharing and reuse costly.
The mathdata project addresses several fundamental problems:
- Reinventing the wheel — existing data is reproduced each time instead of reused.
- Babel effect — knowledge about the same object from different sources is hard to combine.
- Modern tools — the math community cannot leverage modern data storage and analytic methods.
- Credibility — self-selected benchmarks reduce the credibility of algorithm comparisons.
- Reproducibility — results cannot be independently reproduced if data is not shareable.
Goals
- Develop a DSL (MathDataLang) for describing mathematical data
- Implement a database of mathdata objects
- Build a system for automatic benchmarking
- Create a community around MathData
Project Parts
Authentication — owner, editor, user, and external roles; LDAP and cross-database authentication.
Backend — git-based storage for reproducibility; scalable distributed architecture.
Frontend — web interface for querying and submitting data.
DataSanitizer — validates instances against MathDataLang definitions.
Deliverables
- MathDataLang DSL specification and reference implementation
- Working prototype of the database (backend + basic frontend)
- Automatic benchmarking system connected to the database
- Documentation and example datasets
Resources
References
-
M. D. Wilkinson et al. The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3:160018, 2016. DOI 10.1038/sdata.2016.18
-
A. Heinle and V. Levandovskyy. The SDEval benchmarking toolkit. ACM Communications in Computer Algebra, 49(1):1–9, 2015. DOI 10.1145/2768577.2768578 · PDF (self-hosted by the SymbolicData project)