arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2109.01754cs.CLcs.AIcs.LG
收藏

大规模自然语言理解系统中基于Transformer模型的错误检测

Error Detection in Large-Scale Natural Language Understanding Systems Using Transformer Models

Amazon Alexa AI(亚马逊 Alexa AI)

Rakesh Chada, Pradeep Natarajan, Darshan Fofadiya, Prathap Ramachandra

摘要:

Alexa、Siri、Cortana和Google Assistant等大规模对话助理使用多个模型来处理每一句话,以进行域、意图和命名实体识别。鉴于模型开发和大流量的解耦性质,识别由这样的系统错误处理的话语是极其困难的。我们解决这个挑战,检测域分类错误使用离线Transformer模型。我们结合从RoberTa模型的Nbest假设产生的生产系统的话语编码。然后,我们微调端到端的多任务设置使用一个小的数据集的人类注释的话语与域分类错误。我们测试了我们的方法,用于检测来自一个领域的错误分类,该领域占大规模对话AI系统中流量的0.5%。我们的方法实现了30%的F1分数,比bi-LSTM基线高出16.9%,比独立的RoBERTa模型高出4.8%。通过集成多个模型,我们将其进一步提高了2.2%至32.2%。

英文摘要:

Large-scale conversational assistants like Alexa, Siri, Cortana and Google Assistant process every utterance using multiple models for domain, intent and named entity recognition. Given the decoupled nature of model development and large traffic volumes, it is extremely difficult to identify utterances processed erroneously by such systems. We address this challenge to detect domain classification errors using offline Transformer models. We combine utterance encodings from a RoBERTa model with the Nbest hypothesis produced by the production system. We then fine-tune end-to-end in a multitask setting using a small dataset of humanannotated utterances with domain classification errors. We tested our approach for detecting misclassifications from one domain that accounts for <0.5% of the traffic in a large-scale conversational AI system. Our approach achieves an F1 score of 30% outperforming a bi- LSTM baseline by 16.9% and a standalone RoBERTa model by 4.8%. We improve this further by 2.2% to 32.2% by ensembling multiple models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑