{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/a-hierarchical-structured-self-attentive","title":"A Hierarchical Structured Self-Attentive Model for Extractive Document Summarization (HSSAS)","arxiv_id":"1805.07799","date":"2018-05-20","proceeding":null,"authors":["Kamal Al-Sabahi","Zhang Zuping","Mohammed Nadher"],"abstract":"The recent advance in neural network architecture and training algorithms\nhave shown the effectiveness of representation learning. The neural\nnetwork-based models generate better representation than the traditional ones.\nThey have the ability to automatically learn the distributed representation for\nsentences and documents. To this end, we proposed a novel model that addresses\nseveral issues that are not adequately modeled by the previously proposed\nmodels, such as the memory problem and incorporating the knowledge of document\nstructure. Our model uses a hierarchical structured self-attention mechanism to\ncreate the sentence and document embeddings. This architecture mirrors the\nhierarchical structure of the document and in turn enables us to obtain better\nfeature representation. The attention mechanism provides extra source of\ninformation to guide the summary extraction. The new model treated the\nsummarization task as a classification problem in which the model computes the\nrespective probabilities of sentence-summary membership. The model predictions\nare broken up by several features such as information content, salience,\nnovelty and positional representation. The proposed model was evaluated on two\nwell-known datasets, the CNN / Daily Mail, and DUC 2002. The experimental\nresults show that our model outperforms the current extractive state-of-the-art\nby a considerable margin.","url_abs":"http://arxiv.org/abs/1805.07799v1","url_pdf":"http://arxiv.org/pdf/1805.07799v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"document-summarization","task_name":"Document Summarization"},{"task_slug":"extractive-document-summarization-1","task_name":"Extractive Document Summarization"},{"task_slug":"extractive-document-summarization","task_name":"Extractive Text Summarization"},{"task_slug":"representation-learning","task_name":"Representation Learning"},{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"text-summarization","task_name":"Text Summarization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/text-summarization-on-cnn-daily-mail-2","task":"Text Summarization","dataset":"CNN / Daily Mail (Anonymized)","model":"HSSAS","rank_in_archive_order":1,"of":13,"metrics":{"ROUGE-1":"42.3","ROUGE-2":"17.8","ROUGE-L":"37.6"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.07799","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}