Metrics
110 File views/ downloads
104 Record Views
Dissertation
Hidden Mental State Explanation (HMSE): Propositional analysis to generalize across tasks, domains, and language
University of West Florida Libraries
Doctor of Philosophy (PHD), University of West Florida
2023
Abstract
This dissertation explores inferring implicit states from natural language (NL), e.g.,detecting an author’s stances and concerns from an utterance. Hidden mental states (HMS) are a
crucial area of study as they gauge a range of different perspectives across a diverse population for tasks such as identifying a social engineer’s (SE) attack, gauging population attitude during global events (pandemics), and identifying influence campaigns. HMS take various forms e.g., an ask (or underlying request) that someone makes, a stance (attitude toward a topic and belief) that someone holds, a concern (or issue of interest) that someone has, and a range of related component states such as belief, sentiment, and attitude. To explore hypotheses about HMS extraction, this dissertation applies a novel linguistically grounded approach, called Hidden Mental State Explanation (HMSE). HMSE is a propositional analysis framework producing explainable output that generalizes across tasks, domains, and languages. HMSE’s efficacy is evaluated across multiple tasks (e.g., ask, stance, and concern) and domains (e.g., covid, elections, SE). Portability across languages (e.g. English and French) is also explored. The goal is to gain a better understanding of online NL to improve capabilities of identifying SE, gauging stance during pandemics, detecting influence operations, and more. The primary innovation of HMSE is its incorporation of explainable (propositional) representations not available in modern, large-language-model approaches. HMSE overcomes shortcomings of state-of-the-art technologies (e.g., sentiment detection) and contributes to the fields of NL processing and artificial intelligence through the presentation of human-readable, targeted, multi-perspective propositional representations, coupled with quantitative outputs.
Details
- Title
- Hidden Mental State Explanation (HMSE)
- Resource Type
- Dissertation
- Contributors
- Kristen B Venable (Committee Member) - University of West Florida, Intelligent Systems and RoboticsPeter Pirolli (Committee Member) - Florida Institute for Human and Machine CognitionJoseph Wilson (Committee Member)Bonnie J Dorr (Committee Chair) - Florida Institute for Human and Machine Cognition
- Publisher
- University of West Florida Libraries
- Format
- pdf
- Number of pages
- 115
- Copyright
- © 2023 Brodie Mather
- Identifiers
- 99380482294006600
- Academic Unit
- Intelligent Systems and Robotics
- Language
- English
- Awarding Institution
- University of West Florida; Doctor of Philosophy (PHD)
- Theses and Dissertations
- Doctor of Philosophy (PHD), University of West Florida