What was your main intent?
Problem being addressed
Voice Assistants aim to fulfill user requests by choosing the best intent from multiple options generated by its Automated Speech Recognition and Natural Language Understanding sub-systems; however, voice assistants do not always produce the expected results.
A novel model for the intent ranking task, which allows to learn an affinity metric and model the trade-off between extracted meaning from speech utterances and relevance/executability aspects of the intent. The model also uses a Multisource Denoising Autoencoder based pretraining that is capable of learning fused representations of data from multiple sources. The model is also capable of performing zero-shot decision making for predicting and selecting intents.
Advantages of this solution
The suggested approach outperforms existing state of the art methods by reducing the error-rate by 3.8%, which in turn reduces ambiguity and eliminates undesired dead-ends leading to better user experience. The robustness of the algorithm on the intent ranking task improves by 33.3%.
Solution originally applied in these industries
Possible New Application of the Work
Virtual assistants used in healthcare can achieve more accuracy with the suggested methodology as it allows them to better understand the request from a patient and identify the right intent, which is sometimes challenging due to the patient's state/lack of personnel.
When applied in queries, the methodology can definitely improve the search results since it can help to identify the key focus of query.
Source URL: #############