An Automatic Multidocument Text Summarization Approach Based on Naïve Bayesian Classifier Using Timestamp Strategy.

Area of Science:

• Natural Language Processing
• Artificial Intelligence
• Information Retrieval

Background:

• Existing multidocument text summarization systems struggle with accurate sentence extraction, low coverage, poor coherence, and redundancy.
• Current methods often fail to capture essential information effectively from multiple sources.

Purpose of the Study:

• To introduce a novel timestamp approach combined with Naïve Bayesian Classification for enhanced multidocument text summarization.
• To improve the coherence, relevance, and linguistic quality of summaries.

Main Methods:

• A timestamp approach is integrated to provide an ordered structure for summaries, enhancing coherence.
• Naïve Bayesian Classification is employed for sentence extraction and relevance.
• A scoring strategy is utilized to calculate word frequency and identify key information.
• The proposed method is compared against the existing MEAD algorithm and a clustering with lexical chaining approach.

Main Results:

• The proposed method demonstrates faster execution times compared to the MEAD algorithm.
• It achieves higher precision, recall, and F-score values than existing clustering with lexical chaining methods.
• The timestamp approach enhances the ordered look and coherence of the generated summaries.

Conclusions:

• The novel timestamp and Naïve Bayesian Classification approach significantly improves multidocument text summarization.
• The method offers a more efficient and effective solution for generating high-quality, coherent, and relevant summaries.
• This approach addresses key limitations of existing summarization techniques.