final_report_.pdf |
Abstract
Chatbot Mediated Group Decision Making
Increasing numbers of individuals are turning to intelligent agents to help them complete tasks efficiently. However, few exist to assist groups of users in discussion or decision making. Designing intelligent agents for group decision making is difficult due to complex communication in discussion such as expression of preference and negotiation. We introduce MLBot, an interactive chatbot that aids group decisions for movies by interacting with users in a chatroom, providing movie recommendations, and assisting the decision making process. MLBot is based on Movielens, a collaborative filtering recommender system that curates recommendations through user ratings. Our objective for MLBot is to understand and analyze factors of intelligent agents that impact group decision making. We hypothesize a chatbot that actively interacts with users will facilitate the group decision making process. Therefore, an active bot could lead to higher satisfaction and a faster group decision making process. To test this hypothesis, are conducting a Wizard of Oz study in which the bot will appear to be an intelligent agent to the participant, but is operated by a researcher. We are using a 2x2 factorial design comparing interventions of enforcing structure and giving opinion. We are comparing the impact of each factor to study which of the bot behaviors can benefit group decision making the most. In the future, this study can inform researchers on chatbot designs that can streamline the flow of the discussion and help groups make better decisions
Chatbot Mediated Group Decision Making
Increasing numbers of individuals are turning to intelligent agents to help them complete tasks efficiently. However, few exist to assist groups of users in discussion or decision making. Designing intelligent agents for group decision making is difficult due to complex communication in discussion such as expression of preference and negotiation. We introduce MLBot, an interactive chatbot that aids group decisions for movies by interacting with users in a chatroom, providing movie recommendations, and assisting the decision making process. MLBot is based on Movielens, a collaborative filtering recommender system that curates recommendations through user ratings. Our objective for MLBot is to understand and analyze factors of intelligent agents that impact group decision making. We hypothesize a chatbot that actively interacts with users will facilitate the group decision making process. Therefore, an active bot could lead to higher satisfaction and a faster group decision making process. To test this hypothesis, are conducting a Wizard of Oz study in which the bot will appear to be an intelligent agent to the participant, but is operated by a researcher. We are using a 2x2 factorial design comparing interventions of enforcing structure and giving opinion. We are comparing the impact of each factor to study which of the bot behaviors can benefit group decision making the most. In the future, this study can inform researchers on chatbot designs that can streamline the flow of the discussion and help groups make better decisions