Natural language interfaces for business intelligence at scale: A review

Natural Language Interfaces (NLIs) do seem like a potential bridge to the problem of a BI domain where people need to interact with complex data systems, but are not equipped technically to do so. However, with the push for more natural language processing (NLP) and machine learning, natural language interfaces (NLIs) have begun to allow users to interact with data warehouses and analytic platforms using simple conversational queries. The paper attempts to give a snapshot of their evolution, architecture, experimental evaluation, and practical domain applications, to the extent that these basic goals have been achieved so far. Assessing bleeding-edge systems such as GPT-4, Codex, and enterprise-focused NLI platforms, it presents the analysis of difficulties in query disambiguation, scaling, and explainability. The paper ends by noting directions for future work, including more context awareness, domain adaptation, and user-centred design. The purpose of this review is to help researchers and practitioners in building robust, secure, and scalable NLIs for modern data-driven organizations.