Context-Aware Recommendation Via Interactive Conversational Agents: A Case In Business Analytics; Evolution of Trust in Voice Assistant Systems

Toker, Omer, School of Engineering and Applied Science, University of Virginia
Elliott, Travis, University of Virginia
Doryab, Afsaneh, EN-Eng Sys and Environment, University of Virginia

Voice assistant technologies are vital tools to break the accessibility barrier in any technologies. Voice technologies have been around since 1990s, but they started becoming popular and widespread after the release of Siri. These technologies promise great opportunities for user experience in many systems; accessibility provided by voice technologies increase the ease of use. Although, there are many barriers to these technologies such as trust, cost and complexity, the solutions they are providing are promising. This project explored how voice technologies can be useful in certain cases, and which tools can be used with voice technologies to provide agile solutions.
The technical portion of the project focused on building a context-aware recommendation system using voice agents, specifically for human resource management systems (HRMS). The goal of the tehnical portion was to deliver digestible analytical insights by using a voice assistant. This feature would allow users to get valuable information from the system in less time, making the system efficient and accurate. A context aware recommendation was integrated into the system to increase the user experience. Using interactive machine learning the recommendation system was able to learn from its past experience and generate valuable recommendations for the user, context allowed the recommendations to be more personalized for the user.
The STS portion focused how trust for voice assistance has changed. Trust is one of the biggest barriers for voice assistants. Due to data privacy scandals in the recent years that were caused by technology companies, trust for voice assistants have been affected. STS research paper evaluated the trust for voice assistants using STS frameworks like Pacey’s Triangle and SCOT methodology. Pacey’s Triangle allowed to understand the problem from three different aspects: cultural, organizational and technical, while SCOT methodology was used to identify the social groups attached to the topic. The STS paper aided the technical paper by researching what possible solutions might help increase the trust for voice assistant for the system implemented in the technical portion.
Both the technical paper and the STS research paper focused on analyzing the benefits and weaknesses voice assistants had. The technical paper created a state of art that generates digestible analytical insights, using context-aware recommendations via a voice agent. The accuracy and usability of the technical project was evaluated with a user experience study and synthetic data simulations, to test the accuracy of the recommendation system. The STS research paper aimed to understand the trust for voice assistant systems to aid the technical project.

BS (Bachelor of Science)
Recommender System, NLP, Voice Assistant, Data Analytics
All rights reserved (no additional license for public reuse)
Issued Date: