Stress-Testing Emotional Support Models: Moving from Homogeneous to Diverse Help Seekers
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917435848785920 |
|---|---|
| author | Heo, Chaewon Jin, Cheyon Jo, Yohan |
| author_facet | Heo, Chaewon Jin, Cheyon Jo, Yohan |
| contents | As emotional support chatbots have recently gained significant traction across both research and industry, a common evaluation strategy has emerged: use help-seeker simulators to interact with supporter chatbots. However, current simulators suffer from two critical limitations: (1) they fail to capture the behavioral diversity of real-world seekers, often portraying them as overly cooperative, and (2) they lack the controllability required to simulate specific seeker profiles. To address these challenges, we present a controllable seeker simulator driven by nine psychological and linguistic features that underpin seeker behavior. Using authentic Reddit conversations, we train our model via a Mixture-of-Experts (MoE) architecture, which effectively differentiates diverse seeker behaviors into specialized parameter subspaces, thereby enhancing fine-grained controllability. Our simulator achieves superior profile adherence and behavioral diversity compared to existing approaches. Furthermore, evaluating 7 prominent supporter models with our system uncovers previously obscured performance degradations. These findings underscore the utility of our framework in providing a more faithful and stress-tested evaluation for emotional support chatbots. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_07698 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Stress-Testing Emotional Support Models: Moving from Homogeneous to Diverse Help Seekers Heo, Chaewon Jin, Cheyon Jo, Yohan Computation and Language As emotional support chatbots have recently gained significant traction across both research and industry, a common evaluation strategy has emerged: use help-seeker simulators to interact with supporter chatbots. However, current simulators suffer from two critical limitations: (1) they fail to capture the behavioral diversity of real-world seekers, often portraying them as overly cooperative, and (2) they lack the controllability required to simulate specific seeker profiles. To address these challenges, we present a controllable seeker simulator driven by nine psychological and linguistic features that underpin seeker behavior. Using authentic Reddit conversations, we train our model via a Mixture-of-Experts (MoE) architecture, which effectively differentiates diverse seeker behaviors into specialized parameter subspaces, thereby enhancing fine-grained controllability. Our simulator achieves superior profile adherence and behavioral diversity compared to existing approaches. Furthermore, evaluating 7 prominent supporter models with our system uncovers previously obscured performance degradations. These findings underscore the utility of our framework in providing a more faithful and stress-tested evaluation for emotional support chatbots. |
| title | Stress-Testing Emotional Support Models: Moving from Homogeneous to Diverse Help Seekers |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.07698 |