Automated Evaluation of the Virtual Assistant in Bleu and Rouge Scores
Hajar Chatoui, Oğuz Ata
Abstract
The advanced field of natural language processing refers to create an intelligent conversational agent. The main goal is to develop bot are capable of generating responses over the last measurements collected over the IoT network. The virtual assistant classifies the user utterance and extract the valuable information to predict the optimal response. The contribution of this paper was performed by the automatics metrics BLEU and ROUGE used in machine translation. However, these metrics were implemented in NLG system to perform the effectiveness of the system. In this work, we evaluated our model over others containing a large set of data in several languages along with the Paradise survey prompts the user satisfaction the results provide F1-score is 74%, the Precision is 85% and Accuracy 75%. Although, the BLEU 0.8 and ROUGE 0.7. The bot has a strong capability to understand dialog human judgments over various features structured based on qualitative and quantitative experiments.