2021/04/14 by Adrián Bazaga, Bazaga, Adrián, Nupur Gunwant +3
Computer Science · Decision Sciences · #Advanced Database Systems and Queries #Computation and Language (cs.CL) #Data Quality and Management #Databases (cs.DB) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Web Data Mining and Analysis
paper · pdf · doi:10.48550/arxiv.2104.07010
openalex publication_date 2021/04/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The number of databases as well as their size and complexity is increasing.\nThis creates a barrier to use especially for non-experts, who have to come to\ngrips with the nature of the data, the way it has been represented in the\ndatabase, and the specific query languages or user interfaces by which data are\naccessed. These difficulties worsen in research settings, where it is common to\nwork with many different databases. One approach to improving this situation is\nto allow users to pose their queries in natural language.\n In this work we describe a machine learning framework, Polyglotter, that in a\ngeneral way supports the mapping of natural language searches to database\nqueries. Importantly, it does not require the creation of manually annotated\ndata for training and therefore can be applied easily to multiple domains. The\nframework is polyglot in the sense that it supports multiple different database\nengines that are accessed with a variety of query languages, including SQL and\nCypher. Furthermore Polyglotter also supports multi-class queries.\n Our results indicate that our framework performs well on both synthetic and\nreal databases, and may provide opportunities for database maintainers to\nimprove accessibility to their resources.\n