how did I overlook Kristin's thesis?

Author: markriedl

_data/theses.yml

@@ -50,9 +50,18 @@
   title: Neurosymbolic Automated Story Generation
   url: https://smartech.gatech.edu/handle/1853/64643
   institute: Georgia Institute of Technology
-  year: 2021
+  year: 2020
   abstract: |
     Although we are currently riding a technological wave of personal assistants, many of these agents still struggle to communicate appropriately. Humans are natural storytellers, so it would be fitting if artificial intelligence (AI) could tell stories as well. Automated story generation is an area of AI research that aims to create agents that tell good stories. With goodness being subjective and hard-to-define, I focus on the perceived coherence of stories in this thesis. Previous story generation systems use planning and symbolic representations to create new stories, but these systems require a vast amount of knowledge engineering. The stories created by these systems are coherent, but only a finite set of stories can be generated. In contrast, very large neural language models have recently made the headlines in the natural language processing community. Though impressive on the surface, even the most sophisticated of these models begins to lose coherence over time. My research looks at both neural and symbolic techniques of automated story generation. In this dissertation, I created automated story generation systems that improved coherence by leveraging various symbolic approaches for neural systems. I did this through a collection of techniques; by separating out semantic event generation from syntactic sentence generation, manipulating neural event generation to become goal-driven, improving syntactic sentence generation to be more interesting and coherent, and creating a rule-based infrastructure to aid neural networks in causal reasoning.
+- name: Kristin Siu
+  title: Design and Evaluation of Intelligent Reward Structures in Human Computation Games
+  url: http://hdl.handle.net/1853/65063
+  institute: Georgia Institute of Technology
+  year: 2021
+  abstract: |
+    Despite the ubiquity of artificial intelligence, some problems and procedures— such as building commonsense knowledge understanding or generating creative works— have no or few effective algorithmic solutions, yet are considered straightforward for humans to solve. Human computation games (HCGs) are playful, game-based interfaces for tackling these problems through crowdsourcing. HCGs have been used to solve tasks that were and still are considered complex for computational algorithms such as image tagging, protein synthesis, 3D structure reconstruction, and creative artifact generation. However, despite these successes, HCGs have not seen broad adoption compared to other types of serious digital games. Among the many reasons for this lack of adoption is the reality that these games are typically not seen as engaging or compelling to play, as well as the fact that creating HCGs comes at a high development cost to task providers who are typically not game development experts. This thesis is a step towards building and establishing a more formalized design understanding of how to create HCGs that both provide a compelling player experience and complete the underlying task effectively.
+
+    In this thesis, I explore reward mechanics in HCGs. Reward mechanics are integral to HCGs due their associations with player motivation, compensation, and task validation. I first propose a framework for understanding HCG mechanics and advocate for an experimental methodology evaluating both player experience and task completion metrics to understand variations in HCG mechanics. I then use these tools to frame and design three experiments that explore small-scale variations of reward systems in HCGs: reward functions, reward distribution, and reward personalization. These studies demonstrate that even small variations in rewards (i.e., offering players the ability to choose the type of reward) may have significant positive effects on both player experience and task completion metrics. I also show that some variations (i.e., co-located, competitive reward scoring) may have both positive and negative tradeoffs across these metrics. Moreover, this work observes that existing, anecdotal design wisdom for HCGs may not always hold (i.e., allowing players to verbally collude actually predicts higher task solution accuracy). Altogether, this thesis demonstrates that certain aspects of reward systems in HCGs can be varied to improve the player experience without compromising task completion metrics, and builds more empirically-tested design knowledge for creating more engaging, effective HCGs.
 - name: Prithviraj Ammanabrolu
   title: Language Learning in Interactive Environments
   url: https://smartech.gatech.edu/handle/1853/65088