Large events planning assistant
Context and challenges
Preparing and hosting a large event is a complex process, including numerous tasks that needs to be dealt with before the day of the event. Finding the correct set of tasks, including their duration, as well as planning when they should start in order to be ready for the day of the event is called a retroplanning.
Some tasks are recurrent to all types of event, but some events require specific tasks. Some tasks become longer depending on the type or size of event, while some can be externalized if wanted, and the adequate provider must be selected when externalizing a task. Currently, the process of creating the ideal retroplanning is mostly done by hand due to its complexity.
There are several objectives for this project. Two of them are specific for our group: planning creation agent and provider recommendation agent.
The first agent, planning creation agent, must help the user during the event creation phase. By using the metadata provided by the user in the early stage of the event (such as the number of participants, the type of event, the location of the event, and so on) the agent should recommend the appropriate task for the retroplanning, with an estimated duration.
The second agent is activated when tasks are externalized. Using the data scraped from the websites of a curated selection of providers approved by Wise, the agent should be able to recommend the best candidates for a given externalized task.
The first agent mixes classification and regression methods, while the second ones uses NLP and recommendation systems.
Partners and funding
Wise (Epwise SA)
Funded by InnoSuisse – Swiss Innovation Agency
The two agents, embedded in a website, including a graphical representation for the users and a tool for the developers to maintain and re-train the agents when needed.