Skip to main content
Disciplines
Automotive Business IT + Informatics Construction + Real Estate Electrical Engineering + Electronics Energy + Sustainability Insurance + Risk Finance + Banking Management + Leadership Marketing + Sales Mechanical Engineering + Materials
Events
DE
EN
Books
Journals
Topic Page
Marketing
Start single access now
Access for companies
EXTENDED SEARCH
Log in
Top

2019 | Book

Broad Learning Introduction Read chapter Read first chapter

Broad Learning Through Fusions

An Application on Social Networks

Authors: Jiawei Zhang, Dr. Philip S. Yu

Publisher: Springer International Publishing

Part of: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Login to get access
insite
SEARCH

About this book

This book offers a clear and comprehensive introduction to broad learning, one of the novel learning problems studied in data mining and machine learning. Broad learning aims at fusing multiple large-scale information sources of diverse varieties together, and carrying out synergistic data mining tasks across these fused sources in one unified analytic. This book takes online social networks as an application example to introduce the latest alignment and knowledge discovery algorithms. Besides the overview of broad learning, machine learning and social network basics, specific topics covered in this book include network alignment, link prediction, community detection, information diffusion, viral marketing, and network embedding.

Table of Contents

Frontmatter

Background Introduction

Frontmatter
1. Broad Learning Introduction
Abstract
We would like to start this book with an ancient story about “The Blind Men and the Elephant” from John Godfrey Saxe. This story is a famous Indian fable about six blind sojourners who come across different parts of an elephant in their life journeys. In turn, each blind man creates his own version of reality from that limited experiences and perspectives. Instead of explaining its philosophical meanings, we indent to use this story to illustrate the current situations that both the academia and industry are facing about artificial intelligence, machine learning, and data mining.
Jiawei Zhang, Philip S. Yu
2. Machine Learning Overview
Abstract
Learning denotes the process of acquiring new declarative knowledge, the organization of new knowledge into general yet effective representations, and the discovery of new facts and theories through observation and experimentation. Learning is one of the most important skills that mankind can master, which also renders us different from the other animals on this planet. To provide an example, according to our past experiences, we know the sun rises from the east and falls to the west; the moon rotates around the earth; 1 year has 365 days, which are all knowledge we derive from our past life experiences.
Jiawei Zhang, Philip S. Yu
3. Social Network Overview
Abstract
Online social networks (OSNs) denote the online platforms that are used by people to build social connections with the other people, who may share similar personal or career interests, backgrounds, or real-life connections. Online social networking sites vary a lot and there exist a large number of online social sites of different categories, including online sharing sites, online publishing sites, online networking sites, online messaging sites, and online collaborating sites. Each category of these online social networks can provide specific featured services for the customers. For instance, Facebook allows users to socialize with each other via making friends, posting text, sharing photos and videos; Twitter focuses on providing micro-blogging services for users to write/read the latest news and messages; Foursquare is a location-based social network offering location-oriented services; and Instagram is a photo and video sharing social site among friends or to the public.
Jiawei Zhang, Philip S. Yu

Information Fusion: Social Network Alignment

Frontmatter
4. Supervised Network Alignment
Abstract
Online social networks, such as Facebook (https://​www.​facebook.​com), Twitter (https://​twitter.​com), Foursquare (https://​foursquare.​com), and LinkedIn (https://​www.​linkedin.​com), have become more and more popular in recent years. Each social network can be represented as a heterogeneous network containing abundant information about: who, where, when, and what, i.e., who the users are, where they have been to, what they have done, and when they did these activities. Different online social networks can provide unique social network services for the users. For instance, Facebook is a general public social sharing site, Twitter is a micro blogging social site mainly about short posts, Foursquare is a location based social network, and LinkedIn is a business oriented professional social network site.
Jiawei Zhang, Philip S. Yu
5. Unsupervised Network Alignment
Abstract
Identifying the common users shared by different online social sites is a very hard task even for humans. Manually labeling of the anchor links can be extremely challenging, expensive (in human efforts, time, and money costs), and tedious, and the scale of the real-world online social networks involving millions even billions of users also renders the training data labeling much more difficult. In this chapter, we will introduce several approaches to resolve the network alignment problem based on the unsupervised learning setting instead, where no labeled training data will be needed in model building.
Jiawei Zhang, Philip S. Yu
6. Semi-supervised Network Alignment
Abstract
As mentioned before, in the real-world online social networks, the anchor links are extremely difficult to label manually. The training set we can obtain is usually of a small size compared with the network scale, and most of the potential anchor links are unlabeled actually. For instance, given the Facebook and Twitter networks containing millions or billions of users, identifying a very small training set merely with hundreds of correct anchor links is however not an easy task. Therefore, it is not realistic to achieve a large set of labeled anchor links as required by the supervised network alignment models introduced in Chap. 4. On the other hand, completely ignoring the (small) set of labeled anchor links, just like the unsupervised network alignment models introduced in Chap. 5, may also create lots of problems, since these labeled anchor links can provide important signals for the network alignment model building.
Jiawei Zhang, Philip S. Yu

Broad Learning: Knowledge Discovery Across Aligned Networks

Frontmatter
7. Link Prediction
Abstract
Given a screenshot of the online social networks, the problem of inferring the missing links or the links to be formed in the networks in the future is called the link prediction problem. Link prediction problem has concrete applications in the real world, and many concrete services can be cast to the link prediction problem.
Jiawei Zhang, Philip S. Yu
8. Community Detection
Abstract
In the real-world online social networks, users also tend to form different social groups. Users belonging to the same groups usually have more frequent interactions with each other, while those in different groups will have less interactions on the other hand. Formally, such social groups form by users in online social networks are called the online social communities. Online social communities will partition the network into a number of components, where the intra-community social connections are usually far more dense compared with the inter-community social connections. Meanwhile, from the mathematical representation perspective, due to these online social communities, the social network adjacency matrix tend to be not only sparse but also low-rank.
Jiawei Zhang, Philip S. Yu
9. Information Diffusion
Abstract
In the real world, social information can widely spread among people, and information exchange has become one of the most important social activities. The creation of the Internet and online social networks has rapidly facilitated the communication among people. Via the interactions among users in online social networks, information can easily be propagated from one user to other users. For instance, in recent years, online social networks have become the most important social occasion for news acquisition, and many outbreaking social events can get widely spread in the online social networks at a very fast speed. People as the multi-functional “sensors” can detect different kinds of event signals happening in the real world, and write posts to report their discoveries via the online social networks.
Jiawei Zhang, Philip S. Yu
10. Viral Marketing
Abstract
Via the social interactions among users, information of various topics, e.g., personal interests, products, commercial services, etc. can extensively propagate throughout the networks, where lots of users can get infected and become activated. Meanwhile, the social information diffusion can bring about great commercial values, and create lots of viral marketing (Kempe et al., Maximizing the spread of influence through a social network. In KDD, 2003) opportunities. Lots of commercial companies are utilizing the information diffusion phenomenon in online social networks to promote their products or services. For instance, Apple and Huawei have been promoting their latest cell phones via Facebook and Twitter. They can provide some free cell phone samples, coupons, or even cash to certain users (with lots of followers) in Facebook, and ask them to post some good review comments or advertising photos about the cell phone. Such information will propagate to their friends and followers, who may get activated to purchase the cell phone. Commercial promotions via the online social networks have become more and more important in recent years, which even surpass the traditional print media (like newspaper, magazine, TV, and radio). At the same time, viral marketing has also become one of the most important and secure revenue sources for many online social platforms, like Facebook and Twitter.
Jiawei Zhang, Philip S. Yu
11. Network Embedding
Abstract
In the era of big data, information from diverse disciplines is generated at an extremely fast pace, lots of which are highly structured and can be represented as massive and complex networks. The representative examples include online social networks, like Facebook and Twitter, academic retrieval sites, like DBLP and Google Scholar, as well as bio-medical data, e.g., human brain networks. These networks/graphs are usually very challenging to handle due to their extremely large-scale (involving millions even billions of nodes), complex structures (containing heterogeneous links) as well as the diverse attributes (attached to the nodes or links). For instance, the Facebook social network involves more than 1 billion active users; DBLP contains about 2.8 billions of papers; and human brain has more than 16 billion neurons.
Jiawei Zhang, Philip S. Yu

Future Directions

Frontmatter
12. Frontier and Future Directions
Abstract
In this book, we have introduced the current research works on broad learning and its applications on online social networks. This book has covered 4 main parts, where the first 3 parts include 6 main research directions about broad learning based social network mining problems, including (1) network alignment, (2) link prediction, (3) community detection, (4) information diffusion, (5) viral marketing, and (6) network embedding. These problems introduced in this book are all very important for many concrete real-world social network applications and services. A number of state-of-the-art algorithms have been proposed to solve these problems, which are introduced in great detail in this book. Broad learning is a very promising research area, and several potential future development directions about broad learning will be illustrated in the following sections.
Jiawei Zhang, Philip S. Yu
Metadata
Title
Broad Learning Through Fusions
Authors
Jiawei Zhang
Dr. Philip S. Yu
Copyright Year
2019
Electronic ISBN
978-3-030-12528-8
Print ISBN
978-3-030-12527-1
DOI
https://doi.org/10.1007/978-3-030-12528-8

Premium Partner

    Image Credits