Software Analytics

A Novel Smart Electronic Liaison to Support International Students

Author: Ahmed Muntasir Hossain, Killian Meehan, Emmanuel Amoh

As the number of international students attending domestic universities increases, it is imperative that administrative programs make changes to accommodate them. International students face critical challenges before and during their time at the university causing the experience to be unnecessarily stressful. The primary channel of communication with service offices is via email. Communication via email is commonly manual, limited to office hours, and poses challenging issues ranging from inconsistent information to long response times, cost for the university, and limited scalability. To overcome this problem, Natural Language Processing (NLP) techniques particularly Conversational Agents (chatbots) are potential candidate solutions. Chatbots are automated, available 24/7, consistent, precise, scalable, and cheap. Therefore, we propose to design, develop, integrate, and evaluate a conversational agent prototype at the University of New Haven that will serve as a smart electronic liaison for international students and service offices. The proposed smart electronic liaison will include (1) admission and immigration module, (2) tutoring module, and (3) graduation module for international students. The admission and immigration module intends to address questions related to employment authorizations and program admission requirements. The tutoring module aims to assist international students by suggesting annotated pre-recorded media from tutors in response to questions. Lastly, the graduation module intends to plan and optimize the graduation timeline of international students with respect to offered courses and associated prerequisites. To verify and validate our smart electronic liaison, we are working closely with the University Immigration Services and Center for Learning Resources at the University of New Haven to analyze requirements, design the software, develop the software, collect data, and train the chatbot. For future work, the smart electronic liaison will be expanded to target a larger student community and incorporate new services that support housing, finances, dietary needs, and more.

Sentiment Analysis of Social Media (SASM)

Author: Ahmed Muntasir Hossain

Digital reputation management systems have become increasingly popular as individuals and businesses seek to maintain and improve their online reputation. However, existing systems, including Online Social Network Interactions (OSNI), are suffering from critical limitations including limited effectiveness, inaccuracy, high costs, and limited scope. Sentiment analysis is a natural language processing technique used to identify and extract opinions, emotions, and attitudes expressed in textual data. Sentiment analysis can be useful in digital reputation management. In this study, we propose to create an open-source, multi-channel, multi-engine sentiment analysis software for social media and digital reputation management purposes. We call this system Sentiment Analysis of Social Media (SASM). SASM collects data/posts from three different social media channels, Twitter, Reddit, and Tumblr. It then filters, aggregates, and analyzes trends in the sentiment of content posted on social media while leveraging different sentiment analysis engines including Microsoft Text Analytics, IBM Watson Natural Language Understanding, and Google Cloud Natural Language. To verify and validate our system, we consider a case study focusing on three major information technology companies: Google, Amazon, and Microsoft. The outcomes of this case study aim to explore how social media content about major information technology companies vary depending on various factors including geo-political, socio-economic, and environmental awareness. SASM is original because it generates rich and reliable sentiment analysis results, leverages well-established sentiment analysis engines, and collects content from various social media channels. SASM will allow companies to manage their digital reputation effectively and affordably on social media and evaluate customer loyalty, competition, and demand trends accurately.