Humans contribute to the creation of valuable data every day, which plays a key role in analysis, prediction, and decision-making. The value of this data comes largely from the human judgment involved in its creation. However, not everyone recognizes its importance or potential benefits, leading to missed opportunities for its effective use. The EPSRC, an organization holding vast amounts of data, has faced challenges in defining research topics. It is unclear whether these topics should be broadly or narrowly defined, and how to achieve this remains uncertain. Therefore, this study addresses this issue by proposing a solution based on graph theory. The main goal is to use graph theory to identify clusters of related research topics within networks built from EPSRC data, covering both recent (2010–2016) and historical periods (1990–2000, 2000–2010). A secondary goal is to uncover groups of researchers by analyzing researcher networks. By applying different community detection algorithms and network interpretations, the study seeks the most effective method for grouping topics and researchers. Results show that the Louvain algorithm, using normalized grant counts as edge weights, produced the most accurate and balanced clusters. This approach demonstrates how data, shaped by human judgment, can be used effectively for meaningful insights.