Row Level Security for Natural Language Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cloud-scale business intelligence (BI) services do not adequately support natural language question (NLQ) requests, struggling with interpreting user intentions, transferring semantics, and providing secure access to data while maintaining privacy and security.
Innovation Solution
The implementation of techniques to enrich dataset metadata with business semantics, including column synonym generation, interesting field ranking, column time granularity prediction, and semantic type inference, along with row-level security for NLQs and automated contribution analysis for 'why' questions, and synthetic question generation for novel insights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If natural language question answering is implemented in cloud-scale BI services, then ease of operation is improved, but security and privacy are worsened
Solution Approach 1:
The patent introduces an intermediary security checking mechanism that sits between the NLQ processing system and the data sources. This intermediary validates and filters data access requests based on security policies, user roles, and data sensitivity levels, allowing the system to maintain ease of NLQ operation while preventing unauthorized access and protecting sensitive information from leakage.
2Ease of operation
If semantic transfer is performed to interpret user intent, then ease of operation is improved, but device complexity is worsened
Solution Approach 1:
The patent applies preliminary action by pre-processing and enriching dataset metadata with business semantics, including column synonym generation, interesting field ranking, column time granularity prediction, and semantic type inference, before the NLQ processing occurs. This pre-computed semantic information is stored and reused during query processing, reducing the real-time computational complexity while maintaining high ease of operation for users.
3Reliability
If row-level security is implemented for NLQs, then security is improved, but productivity is worsened
Solution Approach 1:
The patent implements preliminary action by pre-computing and caching security policies, user permissions, and data sensitivity classifications before NLQ processing. This pre-computed security information is stored in optimized data structures that enable rapid lookup and validation during query execution, minimizing the performance overhead and maintaining high productivity while ensuring robust row-level security.
4Ease of operation
If automated contribution analysis is provided for 'why' questions, then ease of operation is improved, but device complexity is worsened
Solution Approach 1:
The patent applies self-service by implementing automated contribution analysis that autonomously processes 'why' questions without requiring manual intervention. The system automatically identifies contributing factors, performs causal analysis, and generates explanatory insights by leveraging pre-computed semantic metadata and statistical relationships in the data, reducing device complexity while maintaining high ease of operation for users asking causal questions.
Data Source
AI summary
This disclosure describes a natural language question (NLQ) query service within a service provider network that provides row level security (RLS) for autocomplete during entry of NLQs and fuzzy matching in NLQ answering. The rules take the form of per-user predicates such as Tim can only see rows with region=US. In configurations a complex extraction and preprocessing pipeline to extract distinct combinations of values against RLS predicate “rule keys” is used. Those distinct values are indexed along with grouped rule keys to enable pushing down predicates at auto-complete time. This enables pushing part of RLS rule handling to ingestion time of a dataset rather than handling all RLS rule handling at query time, enabling meeting of latency goals. In configurations, a single logical document of unique cell values is split into multiple documents with a subset of rule keys to handle scalability limits.