Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators.
Yinhong LiuHan ZhouZhijiang GuoEhsan ShareghiIvan VulicAnna KorhonenNigel CollierPublished in: CoRR (2024)
Keyphrases
- language model
- pairwise
- language modeling
- document retrieval
- n gram
- probabilistic model
- query expansion
- information retrieval
- language modelling
- speech recognition
- retrieval model
- mixture model
- context sensitive
- user preferences
- query terms
- test collection
- word error rate
- statistical language models
- semi supervised
- ad hoc information retrieval
- vector space model
- document ranking
- language model for information retrieval
- statistical machine translation
- markov random field
- translation model
- pseudo relevance feedback
- smoothing methods
- dependency structure
- spectral clustering
- document length
- information retrieval systems
- natural language
- language models for information retrieval
- similarity measure