Introducing SCQL: Secure Collaborative Query Language for Privacy-Preserving Data Analysis

On March 29, during the inaugural "YinYu Open Source Community Open Day," the privacy‑computing framework YinYu announced a product upgrade and open‑sourced the SCQL feature.

SCQL provides a simple, user‑friendly BI analysis capability that helps small and medium institutions quickly address urgent long‑tail data security analysis needs, marking the first industry implementation of privacy‑preserving data analysis from AI to BI.

In the context of an evolving data‑element strategy, developing privacy‑computing technologies such as secure multi‑party computation (MPC) is essential for trustworthy large‑scale data flow. Traditional privacy‑computing solutions are often too complex and costly for smaller data volumes, prompting the need for a more accessible, cost‑effective approach.

At the event, developers, application partners, and academic experts discussed building more usable privacy‑computing technology and a healthier ecosystem. YinYu framework lead Wang Lei highlighted SCQL, a multi‑party secure data analysis system that uses MPC as its foundation and SQL as the analysis language, enabling straightforward multi‑party data analysis tasks.

Key technical points of YinYu SCQL (Secure Collaborative Query Language):

- Uses SQL as the interface, built on YinYu's MPC kernel SPU, supporting multi‑party joint data analysis. - Provides abstractions such as virtual databases, tables, and users for project‑level permission control. - Implements field‑level data permission control (CCL). - Offers API integration for upstream platforms, enabling independent product encapsulation. - Supports common operators: JOIN, GROUP BY, aggregation functions (sum, min, max, avg), arithmetic (+, -, *, /, INTDIV), comparison (<, <=, >, >=, =, !=), and IN.

SCQL focuses on multi‑party data collaboration scenarios, leveraging the SPU device abstraction to create a SQL‑like secure analysis language.

The language inherits SQL's popularity, ease of learning, and maturity while extending semantics to describe secure computations across multiple data sources using familiar statements like SELECT FROM, JOIN ON, and GROUP BY.

Specific advantages include:

From usage and integration: easy connection to local data sources, SQL‑like experience, simple API for low‑cost multi‑party collaborative analysis.

From data access control: column‑level permission via YinYu CCL, allowing fine‑grained data authorization.

From functional scope and flexibility: supports common data sources (built‑in MySQL, future PostgreSQL, Hive, CSV), a wide range of analysis operators and functions, suitable for both fast online queries and complex offline decision‑making analyses.

From computational performance: as the first industrial‑grade multi‑party secure data analysis system, it inherits query push‑down optimizations, enabling early data filtering, reduced network transfer, and lower MPC computation overhead.

Privacy computing remains a global challenge, and the industry is using open‑source approaches to lower technical barriers and accelerate adoption.

Since July 2022, the YinYu framework has been open‑sourced, covering most mainstream privacy‑computing techniques. This upgrade further enhances usability, and the community invites developers to co‑build and improve SCQL, fostering a robust privacy‑computing ecosystem.