MASc Seminar: Virtual Assistant Design for Water Systems Operation

Friday, October 11, 2019 9:30 am - 9:30 am EDT (GMT -04:00)

Candidate: Yousra Mohamed

Title: Virtual Assistant Design for Water Systems Operation

Date: October 11, 2019

Time: 9:30AM

Place: EIT 3142

Supervisor(s): Rayside, Derek

Abstract:

Water management systems such as wastewater treatment plants and water distributions systems are big systems which include a multitude of variables and performance indicators that drive the decision making process for controlling the plant. To help water operators make the right decisions, we provide them with a platform to get quick answers about the di erent components of the system that they are controlling in natural language. In this paper we explore the architecture for building a virtual assistant in the domain of water systems. Our design focused on developing better semantic inference across the di erent stages of the process. We developed a named entity recognizer that is able to infer the semantics in the water  eld by leveraging state-of-the art methods for word embeddings. Our model achieved signi cant improvements over the baseline TF-IDF cosine similarity model.

Additionally, we explore the design of intent classi ers, which involves more challenges than a traditional classi er due to the small ratio of text length compared to the number of classes. In our design, we incorporate the results of entity recognition, produced from previous layers of the Chatbot pipeline to boost the intent classi cation performance. Our baseline bidirectional LSTM model showed signi cant improvements, amounting to 7-10% accuracy boost on augmented input data and we contrasted its performance with a modi ed bidirectional LSTM architecture which embeds information about recognized entities. In each stage of our architecture, we explored state-of-the-art solutions and how we can customize them to our problem domain in order to build a production level application. We additionally leveraged Chatbot frameworks architecture to provide a context aware virtual assistance experience which is able to infer implicit references from the conversation  ow.