Research topic detection in scientific articles using a hybrid BERT integrated telescopic vector tree model with...

Area of Science:

• * Computational Linguistics
• * Information Retrieval
• * Optimization Algorithms

Background:

• * Increasing volume of scientific publications necessitates advanced search strategies.
• * Novice researchers struggle with identifying relevant keywords and pertinent articles.
• * Optimization algorithms show promise for enhancing topic identification and detection.

Purpose of the Study:

• * To propose a novel method for identifying research topics.
• * To improve the accuracy and efficiency of categorizing and retrieving research papers.
• * To assist researchers in staying updated with scientific advancements.

Main Methods:

• * Combined TV-Tree (Telescopic Vector Tree) and Hybrid BERT (Bidirectional Encoder Representations from Transformers).
• * Employed Emperor Penguin Optimization (EPO) enhanced Nondominated Sorting Genetic Algorithm-II (NSGA-II).
• * Utilized sophisticated algorithms for analyzing large datasets to identify trending themes.

Main Results:

• * Demonstrated effectiveness in detecting prominent research topics across disciplines.
• * Achieved potential accuracy improvements compared to baseline models.
• * Showcased superior performance in topic detection and identification.

Conclusions:

• * The proposed method offers a valuable tool for researchers to manage information overload.
• * Enhanced topic identification facilitates better understanding of research landscapes.
• * This approach enables more efficient and effective navigation of scientific literature.