Dataset Metadata Semantics for Accurate BI Natural Language Answers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-scale Business Intelligence (BI) services lack adequate support for natural language question (NLQ) requests, face challenges in interpreting user intent, managing data access rights, and providing accurate and secure answers to NLQs, especially across multiple organizations, and struggle with teaching users how to formulate effective NLQs.
Innovation Solution
Enrich dataset metadata with business semantics through column synonym generation, interesting field ranking, column time granularity prediction, and semantic type inference, implement row-level security for autocomplete and fuzzy matching, provide automated contribution analysis for 'why' questions, and generate synthetic questions for novel insights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If natural language question (NLQ) support is added to cloud-scale BI services, then user accessibility and ease of operation improve, but interpretation accuracy and security control deteriorate
Solution Approach 1:
The patent introduces an intermediary layer between the NLQ input and the data warehouse, consisting of semantic enrichment modules that process the natural language question through multiple stages (syntax analysis, semantic interpretation, schema mapping) before querying the data. This intermediary layer enables non-technical users to ask questions in plain language while maintaining accurate interpretation through structured semantic processing.
Solution Approach 2:
The system performs preliminary actions by pre-enriching dataset metadata with business semantics, column synonyms, and domain knowledge before NLQ processing occurs. This preparation work includes creating synonym dictionaries, establishing semantic relationships between data elements and business concepts, and pre-validating data access permissions, thereby improving interpretation accuracy when NLQs are submitted.
2Loss of information
If comprehensive data analysis is provided for NLQs, then answer quality and information completeness improve, but processing time and computational resources worsen
Solution Approach 1:
The patent extracts and separates critical processing components from the overall NLQ system. Specifically, it extracts the semantic enrichment process into a standalone module that operates independently from the query execution engine. This allows the system to pre-compute and cache enriched metadata, synonym mappings, and semantic relationships, so that when an NLQ is submitted, only the essential query matching and result retrieval steps need to execute, significantly reducing processing time while maintaining answer completeness.
3Reliability
If row-level security is implemented for NLQs, then data security and privacy protection improve, but system complexity and implementation difficulty worsen
Solution Approach 1:
The patent merges row-level security enforcement with the existing semantic enrichment and query processing pipelines. Instead of implementing security as a separate, standalone mechanism, the security checks are integrated into the metadata enrichment process and the query execution flow. This allows security permissions to be validated and applied automatically during the same processing steps used for semantic interpretation and query generation, reducing overall system complexity while maintaining strong security controls.
4Measurement precision
If automated semantic enrichment is performed on datasets, then NLQ answering accuracy improves, but computing resources and processing overhead worsen
Solution Approach 1:
The patent applies local quality by performing semantic enrichment selectively rather than uniformly across all datasets. The system identifies and prioritizes enrichment for data elements that are most relevant to business operations and frequently queried through NLQs. By focusing computational resources on critical data regions (such as key business metrics, frequently accessed tables, and high-importance columns) while using lighter processing for less critical data, the system achieves high answering accuracy for important queries while reducing overall computing resource consumption.
Data Source
AI summary
This disclosure describes techniques and architecture for enriching dataset metadata of datasets arranged in tabular form comprising rows and columns, wherein each column has a name. The dataset metadata is enriched with business semantics for natural language question answering. The techniques include one or more of generating one or more synonyms for each name; ranking the names with respect to a likelihood that a column includes possible data to be returned to a user in response to a received NLQ from the user; predicting a date granularity for each column; and predicting a semantic type to describe values in the columns.