The Ontology of Payroll

For this week’s blog post, I chose to look at the payroll dataset gathered by the L.A. Controller’s Office. The dataset is displayed through a spreadsheet, with 35 different data types ranging from salary to pay grade to employee ID. Additionally, these category types are organized through numerical units, urls, percentages, or text. The dataset was created by the L.A. Controller’s Office, which manages the city’s accounting and auditing. As such, the way the sheet is arranged makes it seem that this dataset was built with the tax-paying public in mind. The spreadsheet makes it extraordinarily easy for users to find out the exact salaries, benefits, and insurance costs for all L.A. City Departments. This list would be useful to anyone looking to apply to government jobs, as well as government employees who would need this information for tax or accounting purposes. At the same time, the ranking of the categories clearly reflects the most important or relevant priorities out of all the financial costs. For example, most of the numerical, financial data is listed in the first half of the sheet, whereas the administrative information is all placed in the second half of the sheet, suggesting that the financial data is what most users are after, rather than the administrative information. Despite the numerous data types, this set represents a bare-bones stripdown of the city’s employment costs. However, this financial emphasis is only a reflection of the dataset’s owner, the L.A. Controller’s Office. This dataset also lacks any demographical information regarding the job postings and fillings. In this way, datasets and their inherent categories often reveal the interests and agendas of their owners. Because the L.A. City Controller is primarily concerned with financial costs and accounting, it makes sense that the numerical data would be most important.

If we were to organize the data using a different ontology we would see a much different type of dataset and visualization. For example, if someone wanted to have more data regarding the demographics of these job positions, we might breakdown the dataset on a map, filtering employee job positions by location, race, age, etc. If the data set had been created by the Department of Labor, there might be more categories focused on employee happiness, turn-over rate, work-related accident rates, industry growth or decline, etc. Ultimately, the ontological framework of a dataset can give many useful hints to the interests and agendas of the dataset owner. Simultaneously, these datasets also reflect the worldviews of their creators. Therefore, it is important to first understand the ontology of a dataset to before delving into its contents.