Abstract
These days, hosting user-generated content analysis is largely dependent on social media platforms. Because datasets are growing so quickly, analysts frequently struggle to comprehend large sentiments, trending topics, and public opinions. Even though social media offers several analytics tools, manually browsing and understanding datasets is difficult and time-consuming. For businesses and researchers who need real-time insights, this problem becomes more difficult. A novel method known as SMSTA (Social Media Sentiment and Trend Analyzer) is put forth to address this problem. The system operates in several phases, including text processing, semantic sentiment retrieval, answer generation, and dataset data extraction. Source code and documentation are among the dataset files gathered and pre-processed in the first step using tokenization and text normalization methods. Subsequently, a sentiment vector database is used to generate and store meaningful sentiment embeddings for effective retrieval. Semantic similarity is used to retrieve pertinent content during the query stage, and a transformer language model is used to produce precise answers. The proposed SMSTA system increases analyst productivity, decreases manual search effort, and improves sentiment comprehension. When compared to conventional dataset exploration techniques, experimental evaluation demonstrates that SMSTA offers pertinent, context-aware, and effective responses. As a result, we obtain 93% accuracy; there is a great need for an intelligent system that can help users and make understanding datasets easier.
Showing the abstract — retrieve the full paper via the Exa API.