Desarrollo de una aplicación móvil para traducir texto del idioma Kichwa al español y viceversa

Abstract

Abstract: This degree work addresses the development of a mobile application to translate text from the Kichwa language to Spanish and vice versa using text processing techniques. natural language and machine learning. The best were investigated and selected techniques for translation, highlighting the Transformer architecture as the most appropriate. A neural network was designed and developed based on this architecture, which proved to be efficient in translation and adaptation to both languages. Likewise, the coding phases and testing were essential to the success of the project. The mobile application interface was intuitive and easy to use, allowing users to perform translations easily. simple. Furthermore, the results obtained in the tests showed that the application provided accurate and consistent translations, validating the effectiveness of the neural network. Finally, recommendations include expanding the data set of training, exploring data augmentation techniques, implementing user feedback and exploring translations into other indigenous languages. These recommendations seek to improve the application and promote the preservation of the linguistic diversity.