Generative Methods for Social Media Analysis

✍ Scribed by Stan Matwin, Aristides Milios, Paweł Prałat, Amilcar Soares, François Théberge

• Publisher: Springer
• Year: 2023
• Tongue: English
• Leaves: 92
• Series: SpringerBriefs in Computer Science
• Category: Library

✦ Synopsis

This book provides a broad overview of the state of the art of the research in generative methods for the analysis of social media data. It especially includes two important aspects that currently gain importance in mining and modelling social media: dynamics and networks.

The book is divided into five chapters and provides an extensive bibliography consisting of more than 250 papers. After a quick introduction and survey of the book in the first chapter, chapter 2 is devoted to the discussion of data models and ontologies for social network analysis. Next, chapter 3 deals with text generation and generative text models and the dangers they pose to social media and society at large. Chapter 4 then focuses on topic modelling and sentiment analysis in the context of social networks. Finally, Chapter 5 presents graph theory tools and approaches to mine and model social networks. Throughout the book, open problems, highlighting potential future directions, are clearly identified.

The book aims at researchers and graduate students in social media analysis, information retrieval, and machine learning applications.

✦ Table of Contents

Acknowledgments
Contents
1 Introduction
2 Ontologies and Data Models for Cross-platform Social Media Data
2.1 Data Models for Social Media Data Analysis
Homophily Analysis
Social Identity Linkage
Personality Analysis
2.2 Ontologies for Social Media Data
Ontologies for Sentiment Analysis
Ontologies for Situational Awareness
2.3 Potential Future Research Topics
Metadata
Federated Learning
3 Methods for Text Generation in NLP
3.1 Introduction
3.2 Past Approaches
3.3 GANs in NLP
Reinforcement learning strategies
Operating on continuous representations instead of discrete symbols
Gumbel-softmax
3.4 Large Neural Language Models (LNLMs or LLMs)
The Transformer and BERT
BERT variants
Introduction to GPT-3
3.5 Dangers of E ective Generative LLMs
Marginalized Group and Gender Bias
Generation of Hateful Content
De-biasing Approaches
Environmental and Financial Impacts
Identifying Information Extraction Attacks
Simpler Approaches
Potential Research Direction # 1 (Large Neural Language Models)
3.6 Detecting Generated Text
Overview
Detection of Machine-Generated Text
The Issue with Simple Detection
Detection of Fake News Content
Issues of Comparison and Dataset Standardization
Content-based Approaches
Social-response-based Approaches
Hybrid Approaches
Graph-based Approaches
Multimodal Approaches: Incorporating Visual Information
Potential Research Direction # 2 (Fake News Detection)
4 Topic and Sentiment Modelling for Social Media
4.1 Introduction
4.2 Introduction to Topic Modelling
4.3 Overview of Classical Approaches to Topic Modelling
LDA
4.4 Neural Topic Modelling
Variational Topic Modelling
LDA2Vec
Top2Vec
Use of Pre-trained Embeddings for Neural Topic Modelling
Neural Topic Modelling for Social Media
Potential Research Direction # 3 (Extending NTMs)
4.5 Sentiment Analysis
Sentiment Analysis and Stance Detection Standardized Datasets
Traditional Supervised Sentiment Analysis
Multimodal Sentiment Analysis
Potential Research Direction # 4 (Textual Sentiment Analysis over Time)
Aspect-based Sentiment Analysis
ASBA in a Uni ed Framework
Potential Research Direction # 5 (Aspect-based Multimodal Sentiment Analysis)
5 Mining and Modelling Complex Networks
5.1 Node Embeddings
Hyperbolic Spaces
Signed Networks
Potential Research Direction # 6 (Embedding Sequences of Graphs)
Potential Research Direction # 7 (Multi-Layered Graphs)
5.2 Evaluating Node Embeddings
Potential Research Direction # 8 (Selecting an Appropriate Embedding for a Given Task at Hand—Supervised vs. Unsupervised Approach)
5.3 Community Detection
Potential Research Direction # 9 (More General Community Detection and Using Several Sources of Information)
5.4 Hypergraphs
Potential Research Direction # 10 (Hypergraph Modularity Function)
5.5 Understanding the Dynamics of Networks
Human-bot Interaction and Spread of Misinformation
Social Bursts in Collective Attention
Social Learning (Segregation, Polarization)
Potential Research Direction # 11 (Tools Based on the Null-models)
5.6 Generating Synthetic Networks
Potential Research Direction # 12 (Generating Synthetic Higher-order Structures)
6 Conclusions
References