Publication

Abstract : Relational databases saved the world’s facts in a considerable amount. The inability of insight of query languages limits the ability of the users to recapture the adequate knowledge. Here, natural language processing is enforced to relational databases. An NLP-based model is proposed to convert natural language utterances to SQL queries. Semantic parsing models find it difficult to derive to imaginary database schemas while convertion, which then retrieves corresponding results from the database. To handle schema encoding, schema look up and feature illustration within an encoder, the word uttered is taken as the input, and then sequence-sequence modelling takes place to show consideration to schema encoding based on the relation-aware self-attention system. The framework is evaluated on the two models based on the specific coordination and hardness norms of inquiry. The expected model upgrades the results of text-to-SQL (Yu et al. in Cosql: A conversational text-to-sql challenge towards cross-domain natural language interfaces to databases, pp. 1962–1979, 2019) task that is displayed on results of the Spider dataset.