<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html><html xmlns:epub="http://www.idpf.org/2007/ops" xmlns="http://www.w3.org/1999/xhtml"><head><title>Graph Algorithms in Practice</title><link rel="stylesheet" type="text/css" href="epub.css"/></head><body data-type="book"><section data-type="chapter" epub:type="chapter" data-pdf-bookmark="Chapter 7. Graph Algorithms in Practice"><div class="chapter" id="graph_algos_practice_yelp">
<h1><span class="label">Chapter 7. </span>Graph Algorithms in Practice</h1>
<p><a data-type="indexterm" data-primary="graph algorithms (generally)" data-secondary="in practice" id="ix_ch07-adoc0"/>The approach we take to graph analysis evolves as we become more familiar with the behavior of different algorithms on specific datasets.
In this chapter, we’ll run through several examples to give you a better feeling for how to tackle large-scale graph data analysis using datasets from Yelp and the US Department of Transportation.
We’ll walk through Yelp data analysis in Neo4j that includes a general overview of the data, combining algorithms to make trip recommendations, and mining user and business data for consulting. In Spark, we’ll look into US airline data to understand traffic patterns and delays as well as how airports are connected by different airlines.</p>
<p>Because pathfinding algorithms are straightforward, our examples will use these centrality and community detection algorithms:</p>
<ul>
<li>
<p>PageRank to find influential Yelp reviewers and then correlate their ratings for specific hotels</p>
</li>
<li>
<p>Betweenness Centrality to uncover reviewers connected to multiple groups and then extract their preferences</p>
</li>
<li>
<p>Label Propagation with a projection to create supercategories of similar Yelp businesses</p>
</li>
<li>
<p>Degree Centrality to quickly identify airport hubs in the US transport dataset</p>
</li>
<li>
<p>Strongly Connected Components to look at clusters of airport routes in the US</p>
</li>
</ul>
<section data-type="sect1" class="pagebreak-before less_space" data-pdf-bookmark="Analyzing Yelp Data with Neo4j"><div class="sect1" id="idm45715632110888">
<h1>Analyzing Yelp Data with Neo4j</h1>
<p><a data-type="indexterm" data-primary="Neo4j" data-secondary="analyzing Yelp data with" data-seealso="Yelp dataset" id="ix_ch07-adoc1"/><a data-type="indexterm" data-primary="Yelp dataset" data-secondary="analyzing with Neo4j" id="ix_ch07-adoc2"/>Yelp helps people find local businesses based on reviews, preferences, and recommendations.
Over 180 million reviews had been written on the platform as of the end of 2018.
Since 2013, Yelp has run the <a href="https://bit.ly/3fCL6vG">Yelp Dataset challenge</a>, a competition that encourages people to explore and research Yelp’s open dataset.</p>
<p>As of Round 12 (conducted in 2018) of the challenge, the open dataset contained:</p>
<ul>
<li>
<p>Over 7 million reviews plus tips</p>
</li>
<li>
<p>Over 1.5 million users and 280,000 pictures</p>
</li>
<li>
<p>Over 188,000 businesses with 1.4 million attributes</p>
</li>
<li>
<p>10 metropolitan areas</p>
</li>
</ul>
<p>Since its launch, the dataset has become popular, with <a href="https://bit.ly/2upiaRz">hundreds of academic papers</a> written using this material.
The Yelp dataset represents real data that is very well structured and highly interconnected.
It’s a great showcase for graph algorithms that you can also download and explore.</p>
<div data-type="warning" epub:type="warning"><h6>Warning</h6>
<p>Yelp makes a subset of their data available for personal, educational, and academic purposes. The Yelp dataset is periodically updated which may make it necessary for you to alter how you load updated data and will likely alter some algorithm results.</p>
</div>
<section data-type="sect2" data-pdf-bookmark="Yelp Social Network"><div class="sect2" id="yelp_social_network">
<h2>Yelp Social Network</h2>
<p><a data-type="indexterm" data-primary="Yelp dataset" data-secondary="social network" id="idm45715632096024"/>As well as writing and reading reviews about businesses, users of Yelp form a social network.
Users can send friend requests to other users they’ve come across while browsing Yelp.com, or they can connect their address books or Facebook graphs.</p>
<p>The Yelp dataset also includes a social network.
<a data-type="xref" href="#yelp_my_profile">Figure 7-1</a> is a screen capture of the Friends section of Mark’s Yelp profile.</p>
<p>Apart from the fact that Mark needs a few more friends, we’re ready to start. To illustrate how we might analyze Yelp data in Neo4j, we’ll use a scenario where we work for a travel information business. We’ll first explore the Yelp data, and then look at how to help people plan trips with our app. We will walk through finding good recommendations for places to stay and things to do in major cities like Las Vegas.</p>
<p>Another part of our business scenario will involve consulting to travel-destination businesses. In one example we’ll help hotels identify influential visitors and then businesses that they should target for cross-promotion programs.</p>
<figure><div id="yelp_my_profile" class="figure">
<img src="assets/gral_0701.png" alt="gral 0701"/>
<h6><span class="label">Figure 7-1. </span>Mark’s Yelp profile</h6>
</div></figure>
</div></section>
<section data-type="sect2" data-pdf-bookmark="Data Import"><div class="sect2" id="idm45715632089720">
<h2>Data Import</h2>
<p><a data-type="indexterm" data-primary="Yelp dataset" data-secondary="importing into Neo4j" id="idm45715632088520"/>There are many different methods for importing data into Neo4j, including the <a href="https://bit.ly/2UTx26g">Import tool</a>, the <a href="https://bit.ly/2CCfcgR"><code>LOAD CSV</code> command</a> that we’ve seen in earlier chapters, and <a href="https://bit.ly/2JDAr7U">Neo4j drivers</a>.</p>
<p>For the Yelp dataset we need to do a one-off import of a large amount of data, so the Import tool is the best choice.
See <a data-type="xref" href="app01.html#yelp_data_import">“Neo4j Bulk Data Import and Yelp”</a> for more details.</p>
</div></section>
<section data-type="sect2" data-pdf-bookmark="Graph Model"><div class="sect2" id="idm45715632083128">
<h2>Graph Model</h2>
<p><a data-type="indexterm" data-primary="Yelp dataset" data-secondary="graph model" id="idm45715632081720"/>The Yelp data is represented in a graph model as shown in <a data-type="xref" href="#yelp_graph_model">Figure 7-2</a>.</p>
<figure><div id="yelp_graph_model" class="figure">
<img src="assets/gral_0702.png" alt="gral 0702"/>
<h6><span class="label">Figure 7-2. </span>The Yelp graph model</h6>
</div></figure>
<p>Our graph contains <code>User</code> labeled nodes, which have <code>FRIENDS</code> relationships with other <code>Users</code>.
<code>Users</code> also write <code>Reviews</code> and tips about <code>Business</code>es.
All of the metadata is stored as properties of nodes, except for business categories, which are represented by separate <code>Category</code> nodes.
For location data we’ve extracted <code>City</code>, <code>Area</code>, and <code>Country</code> attributes into the subgraph.
In other use cases it might make sense to extract other attributes to nodes such as dates, or collapse nodes to relationships such as reviews.</p>
<p>The Yelp dataset also includes user tips and photos, but we won’t use those in our example.</p>
</div></section>
<section data-type="sect2" data-pdf-bookmark="A Quick Overview of the Yelp Data"><div class="sect2" id="idm45715632071816">
<h2>A Quick Overview of the Yelp Data</h2>
<p><a data-type="indexterm" data-primary="Yelp dataset" data-secondary="overview" id="ix_ch07-adoc3"/>Once we have the data loaded in Neo4j, we’ll execute some exploratory queries.
We’ll ask how many nodes are in each category or what types of relations exist, to get a feel for the Yelp data.
Previ
Neo4j Graph Algorithms
需积分: 1 199 浏览量
2023-12-25
13:49:39
上传
评论 1
收藏 32.51MB RAR 举报
Neo4j Graph Algorithms （138个子文件） 
epub.css 31KB
ch07.html 172KB ch08.html 158KB ch04.html 149KB ch05.html 106KB ch06.html 99KB ix01.html 94KB ch02.html 37KB ch01.html 31KB ch03.html 22KB toc01.html 19KB app01.html 10KB preface02.html 9KB preface03.html 8KB copyright-page01.html 4KB colophon02.html 3KB colophon01.html 2KB preface01.html 649B titlepage01.html 607B cover.html 321B mimetype 20B toc.ncx 32KB content.opf 14KB DejaVuSans-Bold.otf 600KB DejaVuSerif.otf 330KB UbuntuMono-Italic.otf 145KB UbuntuMono-BoldItalic.otf 145KB UbuntuMono-Regular.otf 137KB UbuntuMono-Bold.otf 134KB
Neo4j.png 15.77MB gral_0103.png 1.85MB gral_0107.png 1.1MB gral_0401.png 940KB cover_sponsor.png 594KB gral_0511.png 427KB gral_0612.png 413KB gral_0101.png 326KB gral_0608.png 300KB gral_0601.png 299KB gral_0104.png 249KB gral_0812.png 243KB gral_0201.png 240KB gral_0613.png 234KB gral_0605.png 231KB gral_0611.png 227KB gral_0408.png 211KB gral_0403.png 209KB gral_0710.png 202KB gral_0102.png 200KB gral_0607.png 199KB gral_0705.png 196KB gral_0602.png 192KB gral_0106.png 184KB gral_0604.png 183KB gral_0610.png 182KB gral_0712.png 181KB gral_0706.png 180KB gral_0512.png 176KB gral_0411.png 173KB gral_0302.png 172KB gral_0802.png 167KB gral_0713.png 165KB gral_0714.png 162KB gral_0409.png 155KB gral_0805.png 153KB gral_0701.png 153KB gral_0509.png 150KB gral_0108.png 149KB gral_0513.png 148KB gral_0505.png 146KB gral_0506.png 145KB gral_0708.png 145KB gral_0504.png 145KB gral_aa04.png 139KB gral_0501.png 137KB gral_0510.png 137KB gral_0402.png 136KB gral_0211.png 134KB gral_0606.png 128KB gral_0105.png 127KB gral_0502.png 118KB gral_0404.png 118KB gral_0203.png 118KB gral_0303.png 114KB gral_0304.png 114KB gral_0210.png 108KB gral_0818.png 108KB gral_0508.png 106KB gral_0603.png 98KB gral_0817.png 96KB gral_0109.png 93KB gral_0806.png 93KB gral_0209.png 90KB gral_0711.png 89KB gral_0803.png 88KB gral_0813.png 87KB gral_0507.png 85KB gral_0202.png 85KB gral_0405.png 84KB gral_0702.png 83KB共 138 条
- 1
- 2
资源评论
