POLIBITS, Vol. 62, pp. 43-50, 2020.
Abstract: Natural language interfaces to databases (NLIDBs) have proven to be very promising tools when trying to obtain information from a relational database since they require the end user to have very little training and knowledge about databases to use them. However, their development has not been easy due to problems related to natural language processing. In addition to this, most authors overlook an important factor in developing these tools, which is the quality of the design of the database to be queried by the NLIDB. The problem arises because there can be many alternatives for the design of databases, and some contain design anomalies. Many NLIDBs would not work correctly for these databases since they were designed under the assumption that they would be used with databases without anomalies. This article describes an improvement to the processing performed by a domain-independent interface to treat databases with design anomalies and for the interface to be able to correctly process queries involving such anomalies. The literature on NLIDBs has not mentioned this problem and much less addressed it.
Keywords: Natural language interfaces, relational databases, user interfaces
PDF: Processing Natural Language Queries via a Natural Language Interface to Databases with Design Anomalies
PDF: Processing Natural Language Queries via a Natural Language Interface to Databases with Design Anomalies
https://doi.org/10.17562/PB-62-5
See table of contents of POLIBITS 62.