The Emora College Companion: A Socialbot for College Students Público

Abstract

The social chatbot, Emora, was first designed in 2020 by researchers at the Natural Language Processing Lab at Emory University to compete in the Alexa Prize SocialBot Grand Challenge 3. Motivated by the inquisitive social chatbot’s success in the challenge, a team of Emory NLP researchers created the Emora College Companion, a socialbot for undergraduate students. The relevant release of this technology to Emory’s Undergraduate population follows a shift in higher education settings towards using chatbots to support student populations. Unlike current chatbot models however, the Emora College Companion differentiates itself by being fully conversational and learning about its user through meaningful and human-like interactions. The Emora College Companion helps undergraduate students achieve their academic goals by being a faithful companion who is there for students from orientation to graduation. To the best of available knowledge, the Emora College Companion is the first socialbot catering to college undergraduate students. Current versions of chatbots utilized in higher education serve to answer frequently asked questions and send reminders. These chatbots function well to fulfill the needs they were designed for: easing workloads of advisors, professors, and teaching assistants and keeping students engaged in completing deliverables required for college enrollment. Nonetheless, none complete these tasks through natural-language-driven social interactions. The Emora College Companion tackles the task of asking students to think differently about how they approach resolving day-to-day tasks and challenge, such as preventing procrastination and resolving interpersonal issues. Currently, Emora spends most of her time conversing with a young and technologically influenced population. Some of her most robust and successful topics include video games, movies, and trendy food options. The most recently tested version of the College Companion, Version 8, is able to sustain over 40 distinct and personalized conversations, ranging across 13 topics. In order to exhaust the contents of College Companion Version 8, users would need to complete approximately 456 turns, which sums to about 3,360 seconds (56 minutes) of non-stop conversation. Version 8 of the Emora College Companion has a misunderstanding rate of 5.05% and poor response rate of 4.46%, based on 10 user conversations. 

Table of Contents

1 Introduction 1

1.1 Motivation................................. 2

1.1.1 BroadImpacts .......................... 2

1.1.2 IntellectualMerit......................... 3

2 Related Work 5

2.1 A new Emora............................... 5

2.2 Chatbots in Higher-Education...................... 6

2.3 Socialbots................................. 9

3 Approach 12

3.1 TopicDevelopment ............................ 13

3.1.1 Stage1............................... 13

3.1.2 Stage2............................... 14

3.1.3 Stage3............................... 15

3.2 ChatbotBuildingandTopicIntegration . . . . . . . . . . . . . . . . 15

3.3 DataCollection .............................. 16

4 Experiments and Analysis 18

4.1 Pilot.................................... 18

4.2 Update1.................................. 24

4.3 Update2.................................. 28

4.4 Updates3&4............................... 29

4.4.1 Emora Response Latency Issues................. 30

4.5 Update5.................................. 32

4.6 Update6.................................. 34

4.7 Update7.................................. 35

4.8 Update8.................................. 37

5 Conclusion 38

5.0.1 Future Directions......................... 39

Bibliography 41 

About this Honors Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Emory College
Department	Quantitative Science
Degree	B.S.
Submission	Honors Thesis
Language	English
Research Field	Computer Science Language, Linguistics
Palabra Clave	Conversational AI Natural Language Understanding Chatbots
Committee Chair / Thesis Advisor	Jinho D. Choi, Emory University
Committee Members	Peter W. Roberts, Emory University Roberto Franzosi, Emory University

Última modificación

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	The Emora College Companion: A Socialbot for College Students ()	2023-04-09 20:55:02 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions