Integrating particle swarm optimization with backtracking search optimization feature extraction with two-dimensional...

Area of Science:

• Natural Language Processing
• Artificial Intelligence
• Machine Learning

Background:

• The proliferation of digital information necessitates efficient methods for extracting key insights from large document sets.
• Text summarization, particularly multi-document summarization (MDS), presents significant challenges in synthesizing information accurately and coherently.
• Existing automatic text summarization techniques, including deep learning and evolutionary algorithms, show promise but require further refinement for complex summarization tasks.

Purpose of the Study:

• To develop an advanced text summarization model capable of generating precise and coherent summaries from single and multiple documents.
• To introduce an improvised Particle Swarm Optimization with Backtracking Search Optimization (PSOBSA) algorithm for enhanced feature extraction in summarization.
• To integrate a two-dimensional convolutional neural network (2D CNN) with an attention-based stacked bidirectional long short-term memory (ABS-BiLSTM) for sentence analysis and summary generation.

Main Methods:

• Feature extraction using the proposed improvised Particle Swarm Optimization with Backtracking Search Optimization (PSOBSA) algorithm.
• Classification and sentence analysis employing a two-dimensional convolutional neural network (2D CNN) and an attention-based stacked bidirectional long short-term memory (ABS-BiLSTM) model.
• Performance evaluation on benchmark datasets (DUC 2002, 2003, 2005, Multi-News, CNN/Daily Mail) using metrics like ROUGE, BLEU, cohesion, and readability.

Main Results:

• The proposed PSOBSA-enhanced 2D CNN with ABS-BiLSTM model demonstrated superior performance compared to several advanced summarization techniques (PSO, CSO, LSTM-CNN, SVR, BSA, ACO, FFA).
• The model achieved higher scores in ROUGE, BLEU, cohesion, and readability, indicating improved summary quality.
• Experimental findings confirmed the model's effectiveness in generating coherent, non-redundant, and grammatically correct summaries for both single-document summarization (SDS) and multi-document summarization (MDS).

Conclusions:

• The integrated PSOBSA, 2D CNN, and ABS-BiLSTM model represents a significant advancement in automatic text summarization.
• The proposed approach effectively addresses the complexities of multi-document summarization, offering improved accuracy and coherence.
• This research provides a robust framework for future development in natural language processing and information retrieval systems.