Why Aren't We NER Yet? Artifacts of ASR Errors in Named Entity Recognition in Spontaneous Speech Transcripts.
Piotr SzymanskiLukasz AugustyniakMikolaj MorzyAdrian SzymczakKrzysztof SurdykPiotr ZelaskoPublished in: ACL (1) (2023)
Keyphrases
- named entity recognition
- named entities
- information extraction
- speech transcripts
- question answering
- natural language processing
- broadcast news
- automatic speech recognition
- maximum entropy
- conditional random fields
- semi supervised
- sequence labeling
- relation extraction
- text summarization
- speech recognition
- classifier ensemble
- maximum entropy classifier
- automatically generated
- machine learning
- higher order
- annotated corpus
- proper names
- text documents
- speech signal
- text mining
- pattern recognition
- similarity measure
- feature extraction
- information retrieval