Introduction
The main objective of this project was to analyse call recordings received to cancer council Victoria(CCV) and generate interactive dashboard. Here a total of 60,081 calls were analysed from 2018 to 2021. Following are the main objectives.
- Transcribe the call recordings.
- Extract emotions from the phone calls belonging to bipolar and eight emotion spectrum.
- Analysing the emotion outputs.
- Analysing keywords belonging to different emotions.
- Analysing emotions based on call length, type of cancer, stages of cancer and different temporal dimensions.
- Model emotional trends throughout the calls. - How emotion intensity fluctuated over the period of the call
- Thematic extraction of the call recordings. - Extract different themes using seed words.
Framework
Following figure shows the framework of the system.
Results
Summary of the results from call analysis.
Technologies and areas
Python, Keyword matching, Word embedding, Deep emotion extraction(Emotion AWARE), Speech-to-text(Azure), Emotion dynamics
Team
Gihan Gamage(me), Sajani Ranasinghe, A. Prof. Daswin De Silva, Harsha Kumara
Publications
An Artificial Intelligence Framework for the Detection of Emotion Transitions in Telehealth Services
2022 15th International Conference on Human System Interaction (HSI)
Sajani Ranasinghe, G Gamage, Harsha Moraliyage, Nishan Mills, Nikki McCaffrey, Jessica Bucholc, Katherine Lane, Angela Cahill, Victoria White, Daswin De Silva
Emotional transformations during Victoria's Cancer Council 131120 information and support service calls detected using Artificial Intelligence
COSA's 49th Annual Scientific Meeting Equitable cancer care for all: Gender, identity, culture, geography, and disease should not matter, 2–4 November 2022
Daswin De Silva, Sajani Ranasinghe, G Gamage, Harsha Moraliyage, Nishan Mills, Jessica Bucholc, Katherine Lane, Angela Cahill, Victoria White Nikki McCaffrey
