Labeling the data
There are many ways to get data labeled, each with its own pros and cons:
- Internal (in-house) labeling: This is when experts from within an organization are used to label data. These are usually people who are domain experts and hence are very familiar with the process and requirements. Consequently, this leads to better quality control and high-quality labeling. Furthermore, as the data doesn’t need to leave the building, there are fewer associated security risks. However, internal labeling is not always possible (e.g., the company size is small or there is a lot of data to label). Furthermore, domain experts are expensive people so asking them to spend inordinate amounts of time on menial annotation tasks is probably not the best use of resources!
- External (outsourced) labeling: As the name suggests, this is when the job is outsourced to companies that specialize in data labeling. These companies are experts at data labeling, and consequently...