N-gram Approximation of LSTM Recurrent Language Models for Single-pass Recognition of Hungarian Call Center Conversations.
Balázs TarjánGyörgy SzaszákTibor FegyóPéter MihajlikPublished in: CogInfoCom (2019)
Keyphrases
- language model
- n gram
- single pass
- call center
- language independent
- language modeling
- recurrent neural networks
- language modelling
- speech recognition
- information retrieval
- probabilistic model
- document retrieval
- query expansion
- retrieval model
- service level
- word level
- test collection
- relevance model
- query terms
- word segmentation
- pseudo relevance feedback
- neural network
- statistical language modeling
- vector space model
- service quality
- arrival rate
- queue length
- agent technology
- document ranking
- text classification
- state space