Tweets opinion analysis integration: ETL modeling with MapReduce

The advent of social media has revolutionized the way people communicate and share information, leading to new business opportunities and challenges. Social media platforms offer a valuable resource in user-generated content (UGC), widely used for opinion analysis and business intelligence. However, traditional integration methods, particularly Extract-Transform-Load (ETL) tasks, need help to keep up with the vast volume, variety, and speed of big data generated on these platforms. This paper proposes MR_ETLSent, an ETL process model adopting MapReduce to perform opinion integration of a large volume of UGC data. It underlines the problem of covering time, cost, and complexity to semantically analyze informal and unstructured UGC texts and transform them into sentiments. This approach provides reusable models for the complex process that extracts UGC text, cleans it, semantically analyzes its data to detect sentiment, and transforms it into the data warehouse. The experimentation results of MR_ETLSent components in the Hadoop framework indicate that the proposed sentiment analysis method based on MapReduce performs well with large sets of UGC while minimizing time and computing resources. Overall, our approach is scalable, efficient, and cost-effective and can be integrated into decision-making systems that analyze opinions and handle large volumes of data.