Wikipedia Question Answering Using SQuAD1 Dataset
Disciplines
Artificial Intelligence and Robotics
Abstract (300 words maximum)
With continuous experimentation and implementation of Machine Learning when it comes to information processing and retrieval, we hope to further this research by delving into creating a program that uses machine learning to analyze Wikipedia-sources and answer questions based on inquiry from the user. The program will use a combination of document retrieval using TF-IDF (Term Frequency-Inverse Document Frequency) and answer extraction using recurrent neural Networks to pull accurate responses to inquiries based on the target information. We will be both training and testing the algorithm using the Stanford Question Answering Dataset. To goal of the algorithm is to be able to tackle more complex questions and explore alternate retrieval methods. We will utilize extrinsic testing to compare practical results of the program with the anticipated correct response. The results are still pending as we are finalizing testing and comparison, but we hope that the result of this research will show a different approach to information retrieval that could be used for AI-powered inquiry responsive algorithms.
Use of AI Disclaimer
no
Academic department under which the project should be listed
CCSE – Computer Science
Primary Investigator (PI) Name
Md Abdullah Al Hafiz Khan
Wikipedia Question Answering Using SQuAD1 Dataset
With continuous experimentation and implementation of Machine Learning when it comes to information processing and retrieval, we hope to further this research by delving into creating a program that uses machine learning to analyze Wikipedia-sources and answer questions based on inquiry from the user. The program will use a combination of document retrieval using TF-IDF (Term Frequency-Inverse Document Frequency) and answer extraction using recurrent neural Networks to pull accurate responses to inquiries based on the target information. We will be both training and testing the algorithm using the Stanford Question Answering Dataset. To goal of the algorithm is to be able to tackle more complex questions and explore alternate retrieval methods. We will utilize extrinsic testing to compare practical results of the program with the anticipated correct response. The results are still pending as we are finalizing testing and comparison, but we hope that the result of this research will show a different approach to information retrieval that could be used for AI-powered inquiry responsive algorithms.