Dataset for Topic Modelling and Fuzzy-Analytical Hierarchy Process

Description

This Dataset supports the research findings presented in the manuscript. It contains the quantitative data collected via expert-driven surveys and subsequent computations used for the Multi-Criteria Decision-Making (MCDM) analysis.

Steps to reproduce

Phase 1: Structural Topic Modelling (STM) This phase uses statistical text mining to objectively extract latent themes from existing literature. 1. Data Extraction: Articles (915 in this study) are collected from databases like Scopus and Web of Science using specific key-strings and inclusion criteria as mentioned in the paper. 2. Data Cleaning: Abstracts are processed to remove inconsistencies. Normalization: Transform all text to lowercase and lemmatize words to understand underlying meanings. Filtering: Remove "stop words", punctuation marks, numbers, and irrelevant terms. 3. Tokenization: Documents are broken down into individual components (tokens) using delimiting characters. 4. Topic Generation: Using the stm package in R, words are analyzed by varied probabilities of occurrence to generate topics. Phase 2: Fuzzy Analytical Hierarchy Process (F-AHP) This phase prioritizes the factors identified in Phase 1 by calculating their relative weights through expert judgment. 1. Build Hierarchy: Create a structure where latent topics from STM act as sub-criteria under main criteria (based on Uses and Gratification Theory). 2. Expert Selection & Data Collection: Gather responses from a panel of 10 experts using a structured questionnaire on a scale of 1–7. 3. Pairwise Comparison: Experts compare criteria and sub-criteria against each other. Consistency Check: Maintain a consistency ratio below 10% to ensure accuracy. 4. Converting to triangular fuzzy numbers: Convert judgments into triangular fuzzy numbers—lower limit, medium, and upper limit 5. Calculation of Weights (Buckley’s Method): Geometric Mean: Calculate the geometric mean for each criterion and sub-criterion comparison. Fuzzy Weights: Multiply the geometric mean by the inverse sum of all geometric means. 6. Defuzzification & Normalization: Defuzzification: Convert triplet fuzzy weights into "crisp" values Normalization: Finalize weights so their total summation equals 1.

Institutions

• National Institute of Technology Kurukshetra

  Haryana, Kurukshetra

Categories

Licence