Automatic Text Summarization Model using Seq2Seq Technique
Chandrika Prasad, Jagdish S. Kallimani, Divakar Harekal, Nicy Sharma · 2020 Fourth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC) · 2020
Increasing acquisition of digitization over the information storing and processing in our daily lives has increased the demand of digitization in multiple facets including in investigation processes as well. In fact, for crimes involving computer systems requires the adoption of best practices for the process of evidence extraction from acquired devices from the crime scenes. Over the past years, summarization has become a topic of research. Various techniques of Natural Language Processing (NLP) enabling researchers to generate efficient results for a wide spectrum of documents. In the proposed work Seq2Seq Architecture with RNN is used to perform summarization tasks for documents. The nature of the summary is abstractive and allows the generation of internal meaning by the model itself. With refinement and continual work, this model becomes a strong foundation to perform summarization on longer and legal documents. The results are efficient summary generation and ROUGE scores in the range of 0.6 - 0.7.