Text summarisation
Project title
Unified Model For Text Summarisation and Supervised Model for Keyword Extraction
Project aim
This project explores extractive and abstractive summarisation: extractive methods, which directly pull sentences from texts, sometimes fail to simplify complex information, whereas abstractive methods can generate fluent summaries but risk distorting the original message by creating new content.
Project outline
In recent years, Natural Language Processing (NLP) has emerged as a critical field, particularly focused on text summarization and keyword extraction to manage the exponential growth of online information.
A novel, hybrid pointer-generator network is introduced for text summarisation, blending the strengths of both extractive and abstractive techniques. Additionally, a state-of-the-art supervised bidirectional model with an attention mechanism is presented for precise keyword extraction.
This framework, tested on datasets from CNN/Daily Mail and Wikidata, not only pushes the efficiency of textual information processing forward but also enhances the accessibility and interpretability of NLP systems in real-world applications.
Images
