Automatic text summarization is a mechanism for converting longer text into smaller text while retaining the contextual meaning. Earlier summarization methods were using statistical methods, but in the worldwide web, the amount of text data generated has grown exponentially, and deep learning models have emerged as a possible solution. Extractive and abstractive are two paradigms in text summarization. In this paper, a modified deep hybrid text summarization model has been proposed that consolidates the two paradigms using stacked recurrent layers for fabricating the summary. It incorporates two neural networks, the extractor network, and the stacked-abstractor network. Extractor primarily selects the most pertinent sentences using sentence-level reward optimization, and the abstractor condenses the selected extract as briefly as possible. The experiment was performed on CNN/Dailymail dataset, and results confirm that the updated approach has relevant improvements over multiple baselines.
Lal, Daisy Monika; Singh, Ajay Kumar; and Singh, Krishna Pratap, "Deep Hybrid Summarizer with Reinforced Sentence Selection" (2021). PACIS 2021 Proceedings. 144.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.