Research map / Large language models / AI and society
Understanding Actors and Evaluating Personae with Gaussian Embeddings
Hannah Kim, Denys Katerenchuk, Daniel Billet, Jun Huan, Haesun Park, Boyang Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Represent actors, movies, and keywords as Gaussian distributions rather than point vectors | Cast prediction with JGE · Cast prediction test | mean rank: 506.75 -> 479.59 | Improved |
| Add age and gender descriptors (ablation) | JGE cast prediction · Cast prediction test | mean rank: 479.66 -> 176.66 | Improved |
| Add automatically identified persona topics to age and gender (ablation) | JGE+AG cast prediction · Cast prediction test | mean rank: 176.66 -> 174.64 | Improved |
| Initialize keyword Gaussian means with GloVe (ablation) | JGE+AG cast prediction · Cast prediction test | Hits@10: 12.52 % -> 12.69 % | Improved |
| Add persona topics to the actor model (ablation) | JGE actor versatility ranking · Actor versatility test | pairwise accuracy: 55.13 -> 59.72 | Improved |
| Add age and gender descriptors to persona topics (ablation) | JGE+T actor versatility ranking · Actor versatility test | pairwise accuracy: 59.72 -> 50.57 | Hurt |
| Use persona topics rather than age and gender descriptors (ablation) | JGE actor versatility ranking · Actor versatility test | pairwise accuracy: 50.84 -> 59.72 | Improved |
| Initialize keyword Gaussian means with GloVe (ablation) | JGE+T actor versatility ranking · Actor versatility test | pairwise accuracy: 59.72 -> 59.39 | Hurt |
| Use the proposed Gaussian actor model instead of the entropy heuristic based on genres | Actor versatility ranking · Actor versatility test | pairwise accuracy: 45.60 -> 59.72 | Improved |
| Add age and gender descriptors (ablation) | JGE cast prediction and JGE versatility ranking · Cast prediction and actor versatility | performance: 479.59 -> 176.66 mean rank; 55.13 -> 50.84 test accuracy | Mixed |
Closest papers by meaning
- PersonaManifold: Revealing and Exploiting Curved Geometry in LLM Persona Representations (NeurIPS 2026 · similarity 0.84)
- MMRole: A Comprehensive Framework for Developing and Evaluating Multimodal Role-Playing Agents (ICLR 2025 · similarity 0.83)
- CoSER: Coordinating LLM-Based Persona Simulation of Established Roles (ICML 2025 · similarity 0.83)
- Quantifying and Optimizing Global Faithfulness in Persona-driven Role-playing (NeurIPS 2024 · similarity 0.83)
- PICLe: Eliciting Diverse Behaviors from Large Language Models with Persona In-Context Learning (ICML 2024 · similarity 0.83)
- CharacterBench: Benchmarking Character Customization of Large Language Models (AAAI 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
