Tagomo: an AI-assisted text exploration and labelling platform

点击翻译按钮获取中文摘要

Companies own large amounts of raw customer interaction data (e.g., customer service chat transcripts, customer support tickets, surveys, online reviews, etc). However, they rarely have the resources to explore, organise and label this data and, as such, convert it to usable data for any type of analysis that can lead to business intelligence and growth and, indeed, better customer service.The aim of the project is to create a software tool that will enable companies of all sizes to **quickly identify the 'topics' discussed (or labels) in their text data as well as annotate and organise it**.Tagomo will be able to **provide intelligent recommendations to a human annotator** and streamline, speed up, or fully automate, parts of the data exploration and labelling process.Tagomo will be employing recent advancements in Machine Learning (weak supervision, confident learning, and active learning) combined with innovative Interaction Design research that will deliver a usable and transparent system for non-technical users.Our vision is to make the resource-expensive data exploration and labelling tasks cheaper, faster, and accessible to companies of all sizes. This way, a company's data becomes an asset, usable for any type of business intelligence task and customer service automation, bringing value to the company, and enabling them to better understand and serve their customers.