От просопографии университетской профессуры до цифрового следа философского парохода: «средние данные» и формальные подходы в истории науки
A concept of 'medium-sized' data is introduced to complement 'Big' data used in many projects in quantitative history. Like Big data, medium-sized data are disaggregated, machine-readable, represent 'natural' populations rather than samples, and are 'shallow' (the number of variables is usually small). Unlike 'Big' data they are not accumulated routinely in a machine-readable format and require a lot of manual work, which puts certain limits to the size of datasets. General principles of dataset formation for the analysis of populations of persons and organizations are discussed. Two datsets (one, for 19th century Russian University professors and instructors, and another, for Russian philosophical periodicals of the first half of the 20th century) are used to demonstrate techniques of stepwise data aggregation (which helps to partly overcome the original shallowness of the medium-sized data) and visualization of historical processes. The role of novel descriptive and representative techniques in comparative studies is discussed.