Conducting Semantic-Based Network Analyses from Social Media Data: Extracted Insights about the Data Leakage Movement

Conducting Semantic-Based Network Analyses from Social Media Data: Extracted Insights about the Data Leakage Movement

Shalin Hai-Jew (Kansas State University, USA)
Copyright: © 2015 |Pages: 59
DOI: 10.4018/978-1-4666-8696-0.ch011
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Network analysis is widely used to mine social media. This involves both the study of structural metadata (information about information) and the related contents (the textual messaging, the related imagery, videos, URLs, and others). A semantic-based network analysis relies on the analysis of relationships between words and phrases (as meaningful concepts), and this approach may be applied effectively to social media data to extract insights. To gain a sense of how this might work, a trending topic of the day was chosen (namely, the free-information and data leakage movement) to see what might be illuminated using this semantic-based network analysis, an open-source technology, NodeXL, and access to multiple social media platforms. Three types of networks are extracted: (1) conversations (#hashtag microblogging networks on Twitter; #eventgraphs on Twitter; and keyword searches on Twitter; (2) contents (video networks on YouTube, related tags networks on Flickr, and article networks on Wikipedia; and (3) user accounts on Twitter, YouTube, Flickr, and Wikipedia.
Chapter Preview
Top

Introduction

Electronic social network analysis (e-SNA) has been applied to a broad range of data from the social web. The analytical approaches tend to work on two axes: (1) the structural approach at the macro, meso, and micro levels, and (2) the content-analysis-based approach. To generalize, the structural approach may highlight the following: clusters within a network, broad-scale types of motif and subgraph relationships in a network (a motif census), and general network membership. The content approach enables the study of conversation gists (as in group labels of Tweets), the identification of individual accounts (at the various levels of the network graph), the study of individual messages or groups of messages, and so on. There is also a kind of middle-space, which enables the study of network graphs with an overlay of various types of text for a semantic-based network analysis (of social media data).

In this approach, the related text and verbiage are an integral part of the visual-based data analysis. To contextualize the concept of semantic meaning in a social media context, it helps to consider the semantic meanings of names (even non-sensical ones) in their various cultural contexts; the meanings in hashtags (even those that are abbreviated and stand in for certain coded meanings); the meanings in metadata like tags for multimedia contents; and then regular semantic meanings in language. [This is not to be confused with semantic network analysis which involves the study of semantic relations between concepts—often through the automated extraction of such relationships from a text or text corpus or through the application of a pre-structured “thesaurus” / concept map of terms applied to a text or text corpus. Such semantic networks are created as a form of data reduction or data summarization. These are virtually always used with stop words lists. Semantic networks may be formed based on various types of word-based relatedness, such as synonymy, antonymy, meronymy, hyperonymy, and others.] A semantic network or frame network also results in data represented in text-labeled graphs, but the theories and methods are different than this approach. The structure of the network graphs extracted from social media platforms comes from how relationships are enabled and defined on the social media platform. The extracted word-based labels are then applied]

Complete Chapter List

Search this Book:
Reset