II. Methodology and Activities

This research explored the intersectional field between Data Science and Epistemology, that I refer as “Tech-Epistemology”, applied to the History of Science. It seeks to bring the "Big Piture" and to support the understanding of the temporality of science and ideas.

It was intends to contribute with the exploration of the boundaries of knowledge studies – including knowledge organization and knowledge representation – with the use of concepts of Information Science, Data Science, among other related fields. In fact, the use of a semantic data modeling and data visualization for knowledge organization had driven most of the efforts along the research.

The epistemological understanding about the knowledge – somehow restricted under the Philosophy domains along the centuries – have been recently transformed by the pacing of the fast advances of the new digital thinking on the 21st Century.

Today, the leverage of Data Science is common sense; so, it can be considered as a natural process the incorporation of a data-driven approach in other fields of study - including History of Science

Some commonly adopted methodologies and approaches in Data Science have been considered for research development, including methods and techniques applied to data analysis.

Even though the research was not based on a specific methodologies - due its exploratory and experimental essence - the fundamentals of these methodologies has been incorporated into the output framework published in this research, that includes schemas, data models, datasets and visualization models.

An important reference is the book Strengthening Data Science Methods for Department of Defense Personnel and Readiness Missions, published in 2017 by the National Academies Press.

Chapter 4 'Overview of Data Science Methods' is dedicated to the main issues regarding the data handling; since the data preparation throughout the analytical process. Chapter 4 is a very recommended reading for those who wants to understand the building blocks of the data thinking that drives the new world.

Furthermore, the chapter also highlights one of the most important issues addressed on this research project that is the complexity behind transforming data into knowledge, mentioning a report from National Research Council in 2013, as well described into page 53:

“The increase in the volume of data does not in and of itself lead to better outcomes. There are challenges associated with storing, indexing, linking, and querying large databases, but perhaps the most significant challenge is drawing meaningful inferences and decisions from analysis of the data. As discussed in the 2013 National Research Council report Frontiers in Massive Data Analysis, “Inference is the problem of turning data into knowledge, where knowledge often is expressed in terms of entities that are not present in the data per se but are present in models that one uses to interpret the data” (NRC, 2013, p. 3).”

National Academies Press, 2017

Other methods and frameworks, as the Cross-Industry Standard Process for Data Mining (CRISP-DM) and Team Data Science Process (TDSP) – usually applied to machine learning – have been also considered and merged into the final model.

Based on the described analysis, the research activities can be clustered or on the following group of activities:

1

Academic Understanding

The academic understanding of the project was defined by the research goal as described:
'development of a Knowledge Network in History of Science, oriented to create an interactive blueprint about the scientific knowledge production from the early modern period to recent history; exploring advances in data science, and creating an epistemological interconnected environment powered by semantic web and data visualization resources'.

2

Data Understanding

The project was initialized based on the analysis of History of Science publications where it was made the assessment of the main metadata structures, grouping, classification and indexation methods.

That analysis is important to create the macro level of data model. It brings the initial inputs for the classification and knowledge organization, even though effectively it happens on the data modeling.

3

Data Modeling

Data Modeling can be considered the most important part when structuring a knowledge network. During the modeling process, not only all the use requirements must be considered as well it also assigns the building blocks of the knowledge organization.

The effective exploration of metadata integrated with the use of complementary datasets determines the relevance and epistemological effectiveness of the knowledge network.

An effective data model allows the exploration of new forms of classification, clustering and visualization.

4

Data Preparation

After the initial activities, the research was focused on the collection of the datasets that would be used to populate the final data structure. Some of the activities include data cleaning, data transformation, and data abstraction.

Datasets had been organized on separated text files to be later imported to the database. The compilation process took longer than expected once part of the datasets was gathered from printed and online versions and had been manually compiled at that stage of research.

5

Data Validation

During the validation, it was fixed some issues about data cleaning and normalization.  The data models have been also revised on the validation.

6

Data Deployment

The data deployment had been made on the setup of the project database (currently in use) and import of the datasets from text files to MySQL format.

7

Evaluation/ PoC

During the Evaluation/ Proof-of-Concept (PoC) the data and models have been validated. The activities are relevant for fine tuning of all aspects related to the data, as well internally was important for allowing the first visualization of the research after months working on conceptual models and data preparation.

8

Data Visualization

Integrated with data modeling, data visualization sharps the relevance of a knowledge network. The usability of a body knowledge is defined by how agents (humans or not humans) interact with and effective is the process of consumption of knowledge, considering its applied context.

The data visualization were oriented to create an informative, flexible, non-complex and intuitive use of historical data; and also contributing with the new data-driven comprehension of the world.

9

Project Deployment

The deployment is the last project activity and consists on the research publishing under a public URL, transferring the data to the definitive infrastructure, publishing documentation and project conclusion.