GRAPH-BASED MODEL TO SUPPORT SOCIAL MEDIA DATA ANALYSIS

• Main Author: STEFAN HARTONO NIM : 23516082, WILLIAM
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/24715
• Institution: Institut Teknologi Bandung

In 2015, it is recorded that 3,17 billions of world populations are using internet. 2,3 billions out of 3,17 billions are active user of social media with each person having 5 – 6 accounts of social networking. Social media itself is used in many field for its many uses and it causes various parties to use its data as leverage. The social media data itself consists of many kinds and types which contributes to its complexity. The social media data itself consists of many kinds and types which contributes to its complexity. To support analyzing the data, a graph model is developed. A graph is chosen as a model because the relations between the data are clear and explicit. Current existing models are not generic because the models were designed to fulfill certain requirement. Therefore, in this Thesis a meta model that is able to support various analyses on many social networkings is developed. <br /> <br /> <br /> First of all, literature studies were conducted to know what kind of analysis are usually done on graph. Then further studies and observation followed to determine which popular social networking is the most complex. A model was then built based on the result of each study and the models are then compared to each other. The graph models made refer to labeled property graph model. A meta model, resulting from the abstraction of the more complex model was then produced. <br /> <br /> <br /> The testing of the meta model is done by making an extension of the meta model applied to other social networkings. The extentions are saved in a graph data base and a number of queries are executed. The queries are designed to match the social networking’s speciality and to match various analyzes. Based on the testing result, it is concluded that Facebook is the most complex social networking because it has many features which other social networkings don’t have. A generic meta model is obtained by making an abstraction from the model based on Facebook.