Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automating Business Intelligence Requirements with Generative AI and Semantic Search

Nimrod Busany, Ethan Hadar, Hananel Hadad, Gil Rosenblum, Zofia Maszlanka, Okhaide Akhigbe, Daniel Amyot

TL;DR

The paper tackles the challenge of eliciting BI requirements in dynamic business environments by introducing AutoBIR, a no-code system that combines semantic search and Large Language Models to convert natural language inquiries into executable analytics specifications, data dependencies, and test-case reports. It presents a modular architecture with OntoDis for ontology/data-bindings, OntoManager for governance, OntoSearch for semantic indexing, and AutoBIR for Text-to-Query with self-debugging and explainability, plus capabilities to generate reports and visualizations. The work includes an empirical evaluation across multiple domains with SME input, analyzes interactive feedback and data-model importance, and discusses security considerations and future BI-system design implications. Significance lies in accelerating BI development, improving requirement accuracy through ontology-grounded reasoning, and enabling more accessible data-driven decision-making while highlighting challenges in scalability, security, and ontology quality. Overall, AutoBIR demonstrates how generative AI and semantic technologies can automate BI requirements, streamline analytics prototyping, and support evolving data ecosystems.

Abstract

Eliciting requirements for Business Intelligence (BI) systems remains a significant challenge, particularly in changing business environments. This paper introduces a novel AI-driven system, called AutoBIR, that leverages semantic search and Large Language Models (LLMs) to automate and accelerate the specification of BI requirements. The system facilitates intuitive interaction with stakeholders through a conversational interface, translating user inputs into prototype analytic code, descriptions, and data dependencies. Additionally, AutoBIR produces detailed test-case reports, optionally enhanced with visual aids, streamlining the requirement elicitation process. By incorporating user feedback, the system refines BI reporting and system design, demonstrating practical applications for expediting data-driven decision-making. This paper explores the broader potential of generative AI in transforming BI development, illustrating its role in enhancing data engineering practice for large-scale, evolving systems.

Automating Business Intelligence Requirements with Generative AI and Semantic Search

TL;DR

The paper tackles the challenge of eliciting BI requirements in dynamic business environments by introducing AutoBIR, a no-code system that combines semantic search and Large Language Models to convert natural language inquiries into executable analytics specifications, data dependencies, and test-case reports. It presents a modular architecture with OntoDis for ontology/data-bindings, OntoManager for governance, OntoSearch for semantic indexing, and AutoBIR for Text-to-Query with self-debugging and explainability, plus capabilities to generate reports and visualizations. The work includes an empirical evaluation across multiple domains with SME input, analyzes interactive feedback and data-model importance, and discusses security considerations and future BI-system design implications. Significance lies in accelerating BI development, improving requirement accuracy through ontology-grounded reasoning, and enabling more accessible data-driven decision-making while highlighting challenges in scalability, security, and ontology quality. Overall, AutoBIR demonstrates how generative AI and semantic technologies can automate BI requirements, streamline analytics prototyping, and support evolving data ecosystems.

Abstract

Eliciting requirements for Business Intelligence (BI) systems remains a significant challenge, particularly in changing business environments. This paper introduces a novel AI-driven system, called AutoBIR, that leverages semantic search and Large Language Models (LLMs) to automate and accelerate the specification of BI requirements. The system facilitates intuitive interaction with stakeholders through a conversational interface, translating user inputs into prototype analytic code, descriptions, and data dependencies. Additionally, AutoBIR produces detailed test-case reports, optionally enhanced with visual aids, streamlining the requirement elicitation process. By incorporating user feedback, the system refines BI reporting and system design, demonstrating practical applications for expediting data-driven decision-making. This paper explores the broader potential of generative AI in transforming BI development, illustrating its role in enhancing data engineering practice for large-scale, evolving systems.

Paper Structure

This paper contains 21 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Running Example
  3. Preliminaries
  4. AutoBIR System Overview
  5. OntoDis: Generating an Ontology and Data Bindings
  6. OntoManager: Data Source and Ontology Management
  7. OntoSearch: Searching an Ontology Catalog
  8. AutoBIR: Query Generation and Self-Debugging
  9. Generating Reports and Data Visualizations
  10. Evaluation and Lessons Learned
  11. Enhancing Data Analysis: Interactive Feedback and User-Centric Design
  12. Technical System Requirements
  13. On the Importance of a Good Data Model
  14. Security Considerations in LLM-Driven Query Generation
  15. Implications for Future BI Systems Design and Usage
  16. ...and 6 more sections

Figures (3)

  • Figure 1: A view of the AutoBIR system interface including the user's business question (left, top), the query generated by our system (left, bottom), and the "grounding" sub-ontology provided as context for the Generative AI (right).
  • Figure 2: A view of the AutoBIR system interface including query execution results (left) and a data visualization generated using Generative AI (right).
  • Figure 3: High-level system overview: To upload a new data source, the system invokes the Setup tools (left of the dashed line). To support the human operator in defining analytics specification, the system uses the Run-time tools (right of the dashed line). When a user asks a business question, the system triggers the AutoBIR tool to generate a query and an explanation. The system uses additional tools to execute the query, generate visualizations, and archive the results as test cases.