| name | privacy-systems-literature |
| description | Research map for SIDRA-related literature: decentralized analytics, personal data stores, local-first systems, materialized view maintenance, streaming DP, SQL DP, PAC privacy, secure computation, and privacy-preserving database systems. |
Privacy Systems Literature
Use this skill when writing or revising SIDRA paper sections, comparing related systems, or choosing privacy mechanisms for decentralized analytics.
Read the focused reference files as needed:
references/sidra-literature.md - SIDRA paper's current citation map and missing literature.references/streaming-dp.md - DP for continual/staged/streaming releases.references/sql-dp-systems.md - SQL/query-system DP including elastic sensitivity and Google DP.references/decentralized-analytics.md - PDS/local-first/federated/edge analytics context.references/classic-dp-platforms.md - PINQ, Airavat, GUPT, Chorus, OpenDP/SmartNoise/Tumult.references/secure-analytics.md - MPC, secure outsourced DBs, and DP+cryptography systems.references/ivm-streaming-systems.md - IVM, streaming MVs, Materialize, DBSP, IncShrink, Enzyme.references/privacy-attacks.md - reconstruction, differencing, membership inference, and aggregate disclosure attacks.
High-Level Positioning
SIDRA should be framed as a declarative architecture/compiler for data minimization: raw user data stays in PDS/client stores, while only privacy-filtered or privatized materialized-view deltas reach central storage. This is distinct from:
- Federated learning, which usually protects model training rather than relational analytics.
- Secure computation, which protects computation visibility but can be expensive and does not by itself minimize central materialization.
- Differentially private SQL systems, which usually assume a centralized curator and static/batch queries.
- Stream processors, which handle windows/triggers but generally do not provide PDS placement and column-level privacy semantics.
Citation Discipline
- For claims about the SIDRA implementation, cite local files or
SIDRA.pdf.
- For paper claims, prefer primary sources: arXiv, PVLDB, ACM/USENIX/Microsoft/Google/OpenDP/DuckDB docs.
- Be explicit about dates for fast-moving projects. The current session date is May 4 2026.
