Abstract

Textual data and many other types of data sources are evaluated through the use of analytical techniques like topic modeling. While several topic modeling techniques have been created that take into account various constraints and relationships within datasets, these techniques are rarely used frequently. On the other hand, many scholars would like to apply Latent Dirichlet Analysis. It is not suitable for simulating more complex data exchanges, despite its versatility and flexibility. In addition to brief sentences like those seen in people's Query text data, we offer a variety of topic modeling techniques that can handle time-varying topic changes and correlations between topics. This study aims to evaluate the effectiveness of five topic modeling techniques: latent Dirichlet allocation (LDA), non-negative matrix factorization (NMF), Latent Semantic Analysis (LDA), Parallel latent Dirichlet allocation (PLDA), and Pachinko Allocation Model (PAM). The goal is to bridge the emerging fields of computational science and empirical social research. Considering the relationship between digital media and social interactions, this study starts with people queries. After that, it evaluates the effectiveness of various algorithms in terms of their benefits and drawbacks in a query-based setting. This work demonstrates the usefulness of PLDA and NMF for evaluating query-based data in view of details from the analytical methodologies and quality concerns