WP2
Conversational AI tool prototype to support domestic violence experiencers
WP2 builds upon the contributions from Tasks 1.1 and 1.2 of WP1 to gain a deep understanding of the victims’ requirements for the AI Chatbot. The aim is to understand victims’ needs and the obstacles and opportunities for accessing services and support. Partner WE will offer guidelines for structuring the interviews with the victims in a manner that elicits answers and conversations that are useful for the development of the AI Chatbot following user-centric design.
Methodology
The method of collecting data for the AI Chatbot will follow the traditional magic triangle of data collection, which is commonly used in the development cycle of conversational AI. It involves recording (1) human-human conversations (T1.1 and T1.2), (2) human-machine interaction (T2.3 and T2.4), and finally (3) machine-machine interactions (T2.5). T1.1 and T1.2 will form the basis for the whole development process.
In T2.1. WE will analyse conversational data (M2.1) and feed analysed data into T2.2 (design of the tool) while also making technical choices. In T2.2 WE will make final decision regarding the platform, cloud base, data storage base, programming languages, machine learning model, and other high-level technical details. Based on these decisions WE will create a clickable mock-up of the tool (M2.2). Working in parallel on T2.1 and T2.2 will enable efficient data usage and planning. Development of the AI Chatbot will be oriented along ongoing feedback from the victim- survivors, allowing for tailoring support around victims’ needs with the counsellors and psychologists, alleviating fears, distrust and anxieties, and shortening the time lag in seeking assistance and support from the institutional DV response.
Work Package Lead