Advancing Multi-Document Text Summarization through Deep Learning and Personalization Techniques
Keywords:
LSTM, results, RNN, flexibility, models.Abstract
This study examines the impact of customisation on multi-document text summarisation (MDTS) and the use of deep learning techniques to enhance summary quality. We demonstrate significant improvements in summarisation tasks across various texts by including pre-trained Long Short-Term Memory (LSTM) networks and Recurrent Neural Networks (RNNs) into our model. These networks are used to derive insights from prior experiences. By integrating user selections into the document encoding process, the model effectively captures complex relationships, yielding summaries that are both valuable and concise. The results of our empirical study indicate that our customised LSTM and RNN-based approach surpasses many current benchmarks. This illustrates that customisation effectively generates summaries that are more relevant to the current context. We advise doing more research into sophisticated graph-based representations, such as knowledge graphs, to augment the semantic depth of summaries and more accurately depict the intricate structures of texts. This work not only emphasises the potential for enhancing MDTS via the integration of deep learning and user-centric customisation but also provides future pathways for improving the robustness and flexibility of models.