Lightning talk 1: The Application of AI Technology in CCaaS
(Dmitry Sityaev – Director of AI @ Connex One)
This introductory talk will cover the use of AI in contact centres. We will show various applications of AI when it comes to building an omni-channel engagement platform. We will also discuss implications of building AI-based CCaaS (Call Centre as a Service) solutions for Data Science teams. This talk will be followed by three presentations on specific projects carried out at Connex One.
Lightning talk 2: Enhancing Speech Analytics with Whisper: A Speech Recognition Evaluation
(Kris Hong – Data Scientist @ Connex One)
Athena Speech Analytics is Connex One’s flagship AI product used by the omnichannel customer engagement platform, which handles millions of customer-agent interactions per day. Using a speech recognition model developed in-house, the platform generates call transcripts which are subsequently analysed to optimise customer experience and workforce efficiency. The recent release of the transformer-based Whisper speech recognition system, trained on nearly 700,000 hours of multidomain data, presents an exciting opportunity to further enhance the speech analytics pipeline. This presentation will evaluate the performance of the general purpose Whisper model against task-specific models, and will discuss how the technology can be leveraged to further improve the Connex One platform’s ability to deliver outstanding user experience.
Lightning talk 3: NLP Projects at Connex One
(Nichola Roberts – Data Scientist @ Connex One)
This talk is a whistle stop tour of some of the ongoing NLP projects happening at Connex One. We will look at some common AI solutions such as sentiment analysis and entity recognition, as well as some more complex problems such as call summarization. We will highlight success stories that we have seen, as well as some of the challenges encountered along the way.
Lightning talk 4: An Introduction to Forecasting using Orbit
(Murtaza Hussain – Data Scientist @ Connex One)
Forecasting is a basic requirement of any business in order to minimise cost and ensure quality and continuity of service. Orbit is a tool that was developed internally at Uber for long and short term forecasting. It is a probabilistic forecasting technique based on Bayesian Inference which currently implements three exponential smoothing models in a fit predict paradigm similar to Scikit Learn. Orbit is also a feature-rich library that has modules for hyperparameter tuning, back testing, diagnostics and more. At Connex One Orbit has been used for building a proof of concept (POC) to forecast demand for our client-facing Technical Support team. In this talk Murtaza Hussain will give a basic introduction to using Orbit and showcase some of the results obtained in the POC.
Followed by a Q&A session with Dmitry Sityaev – Director of AI, Kris Hong – Data Scientist, Nichola Roberts – Data Scientist & Murtaza Hussain – Data Scientist.