Papers with verbal alignment
An Information-Providing Closed-Domain Human-Agent Interaction Corpus (L18-1)
Copied to clipboard
| Challenge: | a human-agent interaction corpus is a corpus of conversations between a user and an embodied conversational agent operated by a wizard of oz . data collected to create a 'corpus' with unexpected situations, such as misunderstandings, false information, and interruptions. |
| Approach: | They propose a public corpus for Human-Agent Interaction where the agent is controlled by a Wizard of Oz. |
| Outcome: | The proposed corpus is based on 15 conversations between users and a wizard of Oz agent . the data are used to create a corpus with unexpected situations, such as misunderstandings, false information, and interruptions. |
When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors (2026.acl-long)
Copied to clipboard
| Challenge: | Existing metrics fail to distinguish mandatory behaviors required for task success from non-mandatory patterns that reflect a model’s autonomous preferences. |
| Approach: | They propose to use response pattern similarity and action graph similarity to isolate non-mandatory behaviors from mandatory behaviors. |
| Outcome: | Evaluating 18 models from 8 providers on -Bench and 2-Bench against Claude Sonnet 4.5, the authors find that within-family model pairs score 5.9 pp higher in response pattern similarity and action graph similarity . |