TRAP@NCI

End-to-End dialogue systems with Dynamic Memory Networks and FastText

Hanumanthu, Sukanya (2018) End-to-End dialogue systems with Dynamic Memory Networks and FastText. Masters thesis, Dublin, National College of Ireland.

[img]
Preview
PDF (Master of Science)
Download (1MB) | Preview

Abstract

Conversational dialogue systems act as the foremost layer of contact of an AI system while making machine-human interactions. Traditional dialogue systems incorporate a modular approach which demand for handcrafting of each module thus increasing the chances of propagating errors from one module to the other. To overcome this, an end to end dialogue system with Dynamic memory Network (DMN) for Question Answering was constructed. This project thrives to improve the performance of DMN by using FastText word to vector conversion technique and compares it with the DMN implemented using global vectors. Results obtained after testing on five different types of Question formats prove that FastText performs well with at least 4% increase in the accuracy. If the same technique was implemented for all dialogue systems prior to building a model, then a gradual shift in the improvement conversational systems can be observed.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science

Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: CAOIMHE NI MHAICIN
Date Deposited: 03 Nov 2018 12:46
Last Modified: 03 Nov 2018 12:46
URI: http://trap.ncirl.ie/id/eprint/3416

Actions (login required)

View Item View Item