Développement et amélioration de l'interface utilisateur en langage naturel : Solution élaborée pour la conversion du langage naturel en langage de requête.

Abstract #

This document is the result of a study project work that took place at Oracle Labs on the

conversion of natural language queries to PGQL queries. The objective of this project is to

improve the interface by solving the various sources of malfunctions and errors as well as the

implementation of new functionalities.

This project aims at improving the interface to convert natural language queries to QLMP

queries. The study also adopted SpaCy; a library for automatic natural language processing, as

well as various libraries for data storage, ontology creation and date management.

Before carrying out this project, we considered it useful to conduct a study of the previous

researches on the same subject and then state the functional requirements in order to define the

objectives to be achieved. Then, we inquired about the different technologies and techniques to

reach the proposed solution.

The last step presents the realization of the different steps of the design, starting with the

extraction of the concepts to be used to create the path corresponding to the query, then the

creation of the different possible paths based on the obtained concepts and ontology. Finally, the

analytical study allows choosing the optimal path, which is converted into an intermediate query

that is used to make the links between the path and the PGQL query obtained by the interface.

Keywords: NLP, Machine Learning, Natural Language Interface, Ontology, PGQL, Natural

Language Conversion to PGQL