An Open-Domain QA System for e-Governance

TL;DR

This paper introduces an open-domain question answering system for Romanian that leverages web search and a fine-tuned BERT model to answer COVID-19 questions, integrating with the RELATE portal.

Contribution

It presents a novel QA pipeline for Romanian, including a new COVID-19 dataset and system integration with language technologies.

Findings

The system effectively answers COVID-19 questions in Romanian.

The BERT-based answer extraction achieves promising accuracy.

The COVID-19 dataset supports future research in Romanian NLP.

Abstract

The paper presents an open-domain Question Answering system for Romanian, answering COVID-19 related questions. The QA system pipeline involves automatic question processing, automatic query generation, web searching for the top 10 most relevant documents and answer extraction using a fine-tuned BERT model for Extractive QA, trained on a COVID-19 data set that we have manually created. The paper will present the QA system and its integration with the Romanian language technologies portal RELATE, the COVID-19 data set and different evaluations of the QA performance.