Natural Language Interfaces to Data

Recent advances in NLU and NLP have resulted renewed interest natural language interfaces to data, which provide an easy mechanism for non-technical users access query the data. While early systems evolved from keyword search focused on simple factual queries, complexity of both input sentences as well generated SQL queries has over time. More recently, there also been a lot focus using conversational data analytics, empowering line with quick insights into There are three main challenges querying (NLQ): (1) identifying entities involved user utterance, (2) connecting different meaningful way underlying source interpret intents, (3) generating structured form or SPARQL. two approaches interpreting user's NLQ. Rule-based make use semantic indices, ontologies, KGs identify query, understand intended relationships between those entities, utilize grammars generate target queries. With deep learning (DL)-based models, many text-to-SQL that try holistically DL models. Hybrid rule-based techniques models emerging by combining strengths approaches. Conversational next step one-shot NLQ exploiting context multiple turns conversation disambiguation. In this article, we review background technologies used interfaces, survey We describe analytics discuss several benchmarks research evaluation.