Summary of Efficient Data Generation For Source-grounded Information-seeking Dialogs: a Use Case For Meeting Transcripts, by Lotem Golany et al.
Efficient Data Generation for Source-grounded Information-seeking Dialogs: A Use Case for Meeting Transcripts
by Lotem Golany, Filippo Galgani, Maya Mamo, Nimrod Parasol, Omer Vandsburger, Nadav Bar, Ido Dagan
First submitted to arxiv on: 2 May 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Automating data generation using Large Language Models (LLMs) has gained popularity. Our research investigates the feasibility and effectiveness of LLM-based data generation in source-grounded information-seeking dialogs with response attribution over long documents. We propose a semi-automatic approach: generating dialog queries and responses with LLMs, followed by human verification and identification of attribution spans. This approach enables us to create MISeD – Meeting Information Seeking Dialogs dataset – focused on meeting transcripts. Models finetuned with MISeD demonstrate superior performance compared to off-the-shelf models, even larger ones. Finetuning on MISeD yields comparable response generation quality to fully manual data, while improving attribution quality and reducing time and effort. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you want a computer to understand and respond to conversations about meetings. This is hard because computers struggle to identify who said what in the conversation. Our research tries to solve this problem by using special language models. We created a new dataset, MISeD, with dialogues about meeting transcripts. We then used this dataset to train other language models, which became better at understanding and responding to conversations like these. This means we can use computers to help us in our work, saving time and effort. |
