Papers by Ekata Mitra
MEEP: Is this Engaging? Prompting Large Language Models for Dialogue Evaluation in Multilingual Settings (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing metrics for engagingness evaluate the response without the conversation history, are designed for one dataset, or have limited correlation with human annotations. |
| Approach: | They propose to use large language models to evaluate engagingness in dialogue . they propose to include prompts and translated prompts in the model . |
| Outcome: | The proposed model outperforms existing methods on evaluation of engagingness in dialogue across languages. |
ThreadSumm: Summarization of Nested Discourse Threads Using Tree of Thoughts (2026.acl-long)
Copied to clipboard
| Challenge: | Discussion forums have nested structures that are difficult to navigate and can attract off-topic replies that get interleaved with the author's own continuation posts. |
| Approach: | They propose a multi-stage LLM framework that treats thread summarization as a hierarchical reasoning problem over explicit aspect and content unit representations. |
| Outcome: | The proposed framework improves the quality of nested discussion thread summarizations while retaining aspect retention and opinion coverage. |