A ‘Task’ is a project to label the same type of data i.e. reviews, warranty claims, call center call logs, or complaints. One task will only contain data that must be labelled together - it is two separate tasks to label reviews and complaints, even if the subject matter is shared. All the tasks are independent, meaning concepts from one task such as data records, data record templates and labels cannot be used by other tasks.
Labelling Performance
A huge advantage of PrediCX is the ability to inspect the labelling performance at any moment in time, allowing you to accurately assess the validity of the output. Each task provides the following performance metrics; F-Score, Recall and Precision.
Precision
Precision is a percentage between 0% and 100% where the higher the better. In general, precision represents the percentage of correctly identified labels X from the total data records labelled with X. However, when in a multi-label situation there is the need of summing up the precision of the task's labels, and so we use a measurement called Micro-Precision.
Recall
Recall is a percentage between 0% and 100% where the higher the better. This metric measures the percentage of correctly identified labels X from all labels X in the dataset. However, when in a multi-label situation there is the need of summing up the recall of the task's labels, and so we use a measurement called Micro-Recall.
F-Score
F-Score is a score between 0 and 1 where the higher the better. This metric combines precision and recall in a weighted harmonic mean.
