Data Augmentation with ChatGPT for Assessing Subject Alignment

As a statistical tool, topic modeling requires replication. Topics are identified as distributions over a set of words, and documents are described as mixtures of latent topics. Sometimes, text information is available through a unique document that covers well-defined subjects, but without replication, the intuitive representation of the document as a mixture of latent topics cannot be derived. Nevertheless, the available document can potentially generate more knowledge (not yet available) that could assist in employing compelling text mining tools. The proposal is to use ChatGPT to mimic the process of generating such knowledge. To illustrate the approach, a research proposal is used as the initial document. The aim is to verify if a given piece of research aligns with the subjects of the research proposal or not combining text mining and statistical tools. A second case-study, analyse Novel's chapters alignment with the Gothic genre.