Natural Language Interface to Relational Database (NLI-RDB) Through Object Relational Mapping (ORM)

• Published in: Advances in Computational Intelligence Systems Vol. 513; pp. 449 - 464
• Main Authors: Alghamdi, Abdullah, Owda, Majdi, Crockett, Keeley
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2016; Springer International Publishing
• Series: Advances in Intelligent Systems and Computing
• Subjects: Artificial intelligence; Automatic control engineering; Database interfaces and hibernate framework; Image processing; Natural language interfaces; Natural language interfaces to relational databases (NLI-RDB); Object relational mapping (ORM)
• ISBN: 3319465619; 9783319465616
• ISSN: 2194-5357; 2194-5365
• DOI: 10.1007/978-3-319-46562-3_29

This paper proposes a novel approach for building a Natural Language Interface to a Relational Database (NLI-RDB) using Conversational Agent (CA), Information Extraction (IE) and Object Relational Mapping (ORM) framework. The CA will help in disambiguating the user’s queries and guiding the user interaction. IE will play an important role in named entities extraction in order to map Natural Language queries into database queries. The ORM framework i.e. the Hibernate framework resolves the impedance mismatch between the Object Oriented Paradigms (OOP) and Relational Databases (RDBs) i.e. OOP concepts differ from RDB concepts, thus it reduces the complexity in generating SQL statements. Also, by utilizing ORM framework, the RDBs entities are mapped into real world objects, which bring the RDBs a step closer to the user. In addition, the ORM framework simplify the interaction between OOP and RDBs. The developed NLI-RDB system allows the user to interact with objects directly in natural language and through navigation, rather than by using SQL statements. This direct interaction tends to be easier and more acceptable for humans whom are nor technically orientated and have no SQL knowledge. The NLI-RDB system also offers friendly and interactive user interface in order to refine the query generated automatically. The NLI-RDB system has been evaluated by a group of participants through a combination of qualitative and quantitative measures. The experimental results show good performance of the prototype and excellent user’s satisfaction.