Generating Abstractive Summaries Using Sequence to Sequence Attention Model
Tooba Siddiqui, Jawwad Ahmed Shamsi · 2018
Text Summarization is one of the challenging fields of Natural Language Processing. Extensive research is being conducted which employs Neural Networks and Deep Learning to obtain improved results. Sequence to sequence attention model by the Google brain team was originally proposed for headlines generation. Headlines are used to highlight the significance of the topic. Realizing this significance, our work is focused on using sequence to sequence attention model by the Google brain team to generate the abstract of research papers. Moreover, temporal attention mechanism has been used in replacement to the global attention to cater the problem of repetitive words. We have compared the efficacy of the two models with respect to their ROUGE score in producing research papers abstracts as summaries. Our result indicates that the temporal attention model is a useful method for generating summaries. Moreover, results indicate that with the increased in dataset size, accuracy of the results also increases.