DEXTER: Deep Encoding Of External Knowledge For Named Entity Recognition In Virtual Assistants

Abstract

Named entity recognition (NER) is usually developed and tested on text from well-written sources. However, in intelligent voice assistants, where NER is an important component, input to NER may be noisy because of user or speech recognition error. In applications, entity labels may change frequently, and non-textual properties like topicality or popularity may be needed to choose among alternatives. We describe a NER system intended to address these problems. We test and train this system on a proprietary user-derived dataset. We compare with a baseline text-only NER system; the baseline enhanced with external gazetteers; and the baseline enhanced with the search and indirect labelling techniques we describe below. The final configuration gives around 6% reduction in NER error rate. We also show that this technique improves related tasks, such as semantic parsing, with an improvement of up to 5% in error rate.

Stats

• citations1
• S2 citations—
• github stars0
• HF likes0
• heat score2.26
• arxiv keymuralidharan2021dexter

Related papers

• "i've Heard Of You!": Generate Spoken Named Entity Recognition Data For Unseen Entities (2024)2.26
• On The Use Of External Data For Spoken Named Entity Recognition (2021)6.77
• Predicting Entity Popularity To Improve Spoken Entity Recognition By Virtual Assistants (2020)5.24
• End-to-end Named Entity Extraction From Speech (2018)0.00
• Whisperner: Unified Open Named Entity And Speech Recognition (2024)2.26
• DAMO-NLP At NLPCC-2022 Task 2: Knowledge Enhanced Robust NER For Speech Entity Linking (2022)3.58
• Personalization For Bert-based Discriminative Speech Recognition Rescoring (2023)5.24
• Leveraging Cross-lingual Transfer Learning In Spoken Named Entity Recognition Systems (2023)0.00