TrojanRAG: Retrieval-Augmented Generation Can Be Backdoor Driver in Large Language Models.
Pengzhou ChengYidong DingTianjie JuZongru WuWei DuPing YiZhuosheng ZhangGongshen LiuPublished in: CoRR (2024)
Keyphrases
- language model
- retrieval model
- document retrieval
- test collection
- query expansion
- language models for information retrieval
- ad hoc information retrieval
- language modeling
- information retrieval
- cross language retrieval
- smoothing methods
- statistical language models
- relevance model
- query terms
- document ranking
- document length
- probabilistic model
- query generation
- language modelling
- n gram
- query specific
- speech recognition
- okapi bm
- passage retrieval
- document level
- statistical language modeling
- ir models
- text retrieval
- vector space model
- term dependencies
- retrieval effectiveness
- pseudo feedback
- pseudo relevance feedback
- term frequency
- context sensitive
- information retrieval systems
- term proximity
- image retrieval
- retrieval framework
- ad hoc retrieval
- language modeling framework
- retrieval systems
- relevance feedback
- out of vocabulary
- expert finding
- term weighting
- retrieval method
- tf idf
- probabilistic retrieval models
- language model for information retrieval