Year: 2023
Type: article
Abstract: In today’s digital era large volume of textual data is generated every second, we can search anything on the web, and it never say “Sorry! Unable to find it on internet”. It always come with plenty of suggestions and data, now practically it is impossible to go through all the data and reach the final decisions. Exponential growth of textual information, automatic text summarization has emerged as a crucial answer. Redundancy, coherence, co-reference, and semantic links between words and sentences are only a few examples of concerns that might be prioritized to strengthen the essential of a summary. In this research, we examine how to improve semantic link between words and sentences, it can help bring about a more accurate generated summary. The suggested technique generates summaries of texts using a pre-existing deep learning model based on a Seq-to-Seq LSTM encoder decoder. Sentence summaries and dictionaries are mapped against one another to see how close they are conceptually. The suggested method was tested on the CNN/Daily-Mail dataset, which is available for public use and contains unstructured text describing news items. Using Rouge scores (Rouge-1, Rouge-2, and Rouge-L), we evaluate how well our system performs in comparison to the current gold standard for extractive text summarization. With the Seq-to-Seq model, the proposed approach produced a 42.74 percent Rouge-1 score, a 12.46 percent Rouge-2 score, and a 43.01 percent Rouge-L score.
Institution: Medi-Caps University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217096277
