TREC 2025 Proceedings

UCSC-SIMRAG-keyword-trained-MiniLM

Submission Details

Organization
ucsc
Track
Interactive Knowledge Acquisition Track
Task
Passage Ranking and Response Generation
Date
2025-07-28

Run Description

What type of manually annotated information does the system use?
automatic: system does not use any manually annotated data and relies only on the user utterance and system responses (canonical responses of previous turns)
How is conversation understanding (NLP/rewriting) performed in this run (check all that apply)?
['method uses large language models like LLaMA and GPT-x.']
What data is used for conversational query understanding in this run (check all that apply)?
['method uses iKAT 23 data']
How is ranking performed in this run (check all that apply)?
['method uses learned sparse retrieval (e.g.¸ SPLADE¸ etc.)']
What data is used to develop the ranking method in this run (check all that apply)?
['method uses iKAT 23 data', 'method is trained with TREC Deep Learning Track and/or MS MARCO dataset']
Please specify all the methods used to handle feedback or clarification responses from the user (check all that apply).
['method does not treat them specially']
Please describe the method used to generate the final conversational responses from one or more retrieved passages (check all that apply).
['method uses multiple sources (multiple passages)', 'method uses large language models to generate the summary.']
Please describe how you integrate the PTKBs in your run (check all that apply)
[' method uses a PTKB relevance model to detect the relevant ones', " method integrates PTKBs in the response generation method (e.g. include in the LLM's prompt)"]
Which LLM did you use to generate the final response?
['method uses closed-source commercial LLMs (e.g. GPT-*)']
Please describe the external resources used by this run, if applicable.
none
Please provide a short description of this run.
This run first identifies useful PTKB statements and passes the useful statements in to the LLM, which then generates a keyword-query that are good for SPLADE retrieval. The query is then evaluated through a trained SIM-RAG model (based on T5, fine-tuned on generated data from the TREC dataset), and re-generated if deemed insufficient. Then, reranking utilizes a MiniLM (initially trained on the MS-MARCO dataset) that is fine-tuned on data generated from the TREC dataset, and uses a complete (non-keyword) query generated by the base prompt. The final response generation utilizes the top 20 ranked passages.
Please provide a priority for assessing this run. (If resources do not allow all runs to be assessed, NIST will work in priority order, resolving ties arbitrarily).
1 (top)

Evaluation Files