Social Network Analysis Using Machine Learning
Electronic Mail (Email) has emerged as a widespread technique for exchanging messages through electronic devices, becoming an indispensable and universal communication medium. Its significance cannot be overstated, as an email address is vital for swift interactions in business, government, trade, entertainment, and various other aspacts of daily life. This mode of communication has progressively replaced traditional written methods for important correspondences, including personal and business trans- actions, where an email is given the same weight as a signed document. In social net- work analysis, a significant challenge lies in identifying essential and influential nodes within a network based on its structure. These nodes can be critical in information dissemination, decision-making processes, and network dynamics. Sentiment Analysis (SA) in text mining has emerged as an automated process to discern subjective information from textual data, such as opinions, attitudes, emotions, and feelings. While many existing approaches treat SA as a text classification problem, requiring labeled data for training machine learning models, obtaining such labeled data can be laborious and time-consuming, often requiring manual annotation efforts. Additionally, the need for transferability across different domains hinders using the same labeled data in diverse applications, necessitating the creation of unique labeled datasets for each part. Overcoming these challenges is crucial for sentiment analysis’s wider adoption and effectiveness in various realworld applications. The objective of the research is to analyze the Enron email dataset by creating a directed graph that represents the email communication network. Two important graph theory metrics are used to find out the number of direct connections (emails sent) for each sender and the influence of each sender as a bridge or critical point of communication in the network. On the other hand, we will use sentiment analysis to analyze the Enron email dataset using different type of pre-trained deep learning models to find the communication type for top ten email sender which we will find using graph theory.