Toxic & Figurative Language Detection and Evaluation Metric for Abstractive and Extractive Summarization in Social Media Content

Download or read book Toxic & Figurative Language Detection and Evaluation Metric for Abstractive and Extractive Summarization in Social Media Content written by Ramya Nischala Akula and published by . This book was released on 2022 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: With the online presence of more than half the world population, social networks and social media plays a very important role in the lives of individuals as well as businesses alike. While there are advantages to using these online platforms, there as also downsides that one should be wary about. We focus on analyzing the information or the content that spreads on online platforms. Text summarization is a crucial task that helps in condensing an enormous amount of social media content. While there are multiple approaches to text summarization, the development of an automatic metric to evaluate the generated summaries remains an open problem in text summarization. We propose a novel evaluation metric, Sentence Pair EmbEDdings (SPEED) Score, for text summarization which is based on semantic similarity between sentence pairs. Our proposed evaluation metric shows an impressive performance in evaluating both abstractive and extractive summarization models and is faster than the current state-of-the-art metrics. In this research, we also put forward a multi-source transfer learning approach using models pre-trained on large-scale datasets to detect inappropriate social media content in universal language (English) and code-mixed environments. Here, sentiment analysis is the process of identifying the emotion associated with these social media texts. The presence of sarcasm in texts is the main hindrance in the performance of sentiment analysis. Inherent ambiguity in sarcastic expressions, make sarcasm detection very difficult. In this work, we focus on detecting sarcasm in textual conversations from various social networking platforms and online media. To this end, we develop an interpretable deep learning model that uses attention to identify crucial sarcastic cue words from the input.