blob: 16008f7e0db52e0b89213134759ef94e9c281352 [file] [log] [blame]
<!DOCTYPE html>
<!--
| Generated by Apache Maven Doxia Site Renderer 1.8.1 from src/site/markdown/sqlpp/similarity.md at 2021-12-13
| Rendered using Apache Maven Fluido Skin 1.7
-->
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<meta name="Date-Revision-yyyymmdd" content="20211213" />
<meta http-equiv="Content-Language" content="en" />
<title>AsterixDB &#x2013; AsterixDB Support of Similarity Queries</title>
<link rel="stylesheet" href="../css/apache-maven-fluido-1.7.min.css" />
<link rel="stylesheet" href="../css/site.css" />
<link rel="stylesheet" href="../css/print.css" media="print" />
<script type="text/javascript" src="../js/apache-maven-fluido-1.7.min.js"></script>
</head>
<body class="topBarDisabled">
<div class="container-fluid">
<div id="banner">
<div class="pull-left"><a href=".././" id="bannerLeft"><img src="../images/asterixlogo.png" alt="AsterixDB"/></a></div>
<div class="pull-right"></div>
<div class="clear"><hr/></div>
</div>
<div id="breadcrumbs">
<ul class="breadcrumb">
<li id="publishDate">Last Published: 2021-12-13</li>
<li id="projectVersion" class="pull-right">Version: 0.9.7.1</li>
<li class="pull-right"><a href="../index.html" title="Documentation Home">Documentation Home</a></li>
</ul>
</div>
<div class="row-fluid">
<div id="leftColumn" class="span2">
<div class="well sidebar-nav">
<ul class="nav nav-list">
<li class="nav-header">Get Started - Installation</li>
<li><a href="../ncservice.html" title="Option 1: using NCService"><span class="none"></span>Option 1: using NCService</a></li>
<li><a href="../ansible.html" title="Option 2: using Ansible"><span class="none"></span>Option 2: using Ansible</a></li>
<li><a href="../aws.html" title="Option 3: using Amazon Web Services"><span class="none"></span>Option 3: using Amazon Web Services</a></li>
<li class="nav-header">AsterixDB Primer</li>
<li><a href="../sqlpp/primer-sqlpp.html" title="Using SQL++"><span class="none"></span>Using SQL++</a></li>
<li class="nav-header">Data Model</li>
<li><a href="../datamodel.html" title="The Asterix Data Model"><span class="none"></span>The Asterix Data Model</a></li>
<li class="nav-header">Queries</li>
<li><a href="../sqlpp/manual.html" title="The SQL++ Query Language"><span class="none"></span>The SQL++ Query Language</a></li>
<li><a href="../SQLPP.html" title="Raw SQL++ Grammar"><span class="none"></span>Raw SQL++ Grammar</a></li>
<li><a href="../sqlpp/builtins.html" title="Builtin Functions"><span class="none"></span>Builtin Functions</a></li>
<li class="nav-header">API/SDK</li>
<li><a href="../api.html" title="HTTP API"><span class="none"></span>HTTP API</a></li>
<li><a href="../csv.html" title="CSV Output"><span class="none"></span>CSV Output</a></li>
<li class="nav-header">Advanced Features</li>
<li><a href="../aql/externaldata.html" title="Accessing External Data"><span class="none"></span>Accessing External Data</a></li>
<li><a href="../feeds.html" title="Data Ingestion with Feeds"><span class="none"></span>Data Ingestion with Feeds</a></li>
<li><a href="../udf.html" title="User Defined Functions"><span class="none"></span>User Defined Functions</a></li>
<li><a href="../sqlpp/filters.html" title="Filter-Based LSM Index Acceleration"><span class="none"></span>Filter-Based LSM Index Acceleration</a></li>
<li><a href="../sqlpp/fulltext.html" title="Support of Full-text Queries"><span class="none"></span>Support of Full-text Queries</a></li>
<li class="active"><a href="#"><span class="none"></span>Support of Similarity Queries</a></li>
<li><a href="../interval_join.html" title="Support of Interval Joins"><span class="none"></span>Support of Interval Joins</a></li>
<li><a href="../sqlpp/arrayindex.html" title="Support of Array Indexes"><span class="none"></span>Support of Array Indexes</a></li>
<li class="nav-header">Deprecated</li>
<li><a href="../aql/primer.html" title="AsterixDB Primer: Using AQL"><span class="none"></span>AsterixDB Primer: Using AQL</a></li>
<li><a href="../aql/manual.html" title="Queries: The Asterix Query Language (AQL)"><span class="none"></span>Queries: The Asterix Query Language (AQL)</a></li>
<li><a href="../aql/builtins.html" title="Queries: Builtin Functions (AQL)"><span class="none"></span>Queries: Builtin Functions (AQL)</a></li>
</ul>
<hr />
<div id="poweredBy">
<div class="clear"></div>
<div class="clear"></div>
<div class="clear"></div>
<div class="clear"></div>
<a href=".././" title="AsterixDB" class="builtBy"><img class="builtBy" alt="AsterixDB" src="../images/asterixlogo.png" /></a>
</div>
</div>
</div>
<div id="bodyColumn" class="span10" >
<!--
! Licensed to the Apache Software Foundation (ASF) under one
! or more contributor license agreements. See the NOTICE file
! distributed with this work for additional information
! regarding copyright ownership. The ASF licenses this file
! to you under the Apache License, Version 2.0 (the
! "License"); you may not use this file except in compliance
! with the License. You may obtain a copy of the License at
!
! http://www.apache.org/licenses/LICENSE-2.0
!
! Unless required by applicable law or agreed to in writing,
! software distributed under the License is distributed on an
! "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
! KIND, either express or implied. See the License for the
! specific language governing permissions and limitations
! under the License.
!-->
<h1>AsterixDB Support of Similarity Queries</h1>
<div class="section">
<h2><a name="Table_of_Contents"></a><a name="toc" id="toc">Table of Contents</a></h2>
<ul>
<li><a href="#Motivation">Motivation</a></li>
<li><a href="#DataTypesAndSimilarityFunctions">Data Types and Similarity Functions</a></li>
<li><a href="#SimilaritySelectionQueries">Similarity Selection Queries</a></li>
<li><a href="#SimilarityJoinQueries">Similarity Join Queries</a></li>
<li><a href="#UsingIndexesToSupportSimilarityQueries">Using Indexes to Support Similarity Queries</a></li>
</ul></div>
<div class="section">
<h2><a name="Motivation_.5BBack_to_TOC.5D"></a><a name="Motivation" id="Motivation">Motivation</a> <font size="4"><a href="#toc">[Back to TOC]</a></font></h2>
<p>Similarity queries are widely used in applications where users need to find objects that satisfy a similarity predicate, while exact matching is not sufficient. These queries are especially important for social and Web applications, where errors, abbreviations, and inconsistencies are common. As an example, we may want to find all the movies starring Schwarzenegger, while we don&#x2019;t know the exact spelling of his last name (despite his popularity in both the movie industry and politics :-)). As another example, we want to find all the Facebook users who have similar friends. To meet this type of needs, AsterixDB supports similarity queries using efficient indexes and algorithms.</p></div>
<div class="section">
<h2><a name="Data_Types_and_Similarity_Functions_.5BBack_to_TOC.5D"></a><a name="DataTypesAndSimilarityFunctions" id="DataTypesAndSimilarityFunctions">Data Types and Similarity Functions</a> <font size="4"><a href="#toc">[Back to TOC]</a></font></h2>
<p>AsterixDB supports <a class="externalLink" href="http://en.wikipedia.org/wiki/Levenshtein_distance">edit distance</a> (on strings) and <a class="externalLink" href="http://en.wikipedia.org/wiki/Jaccard_index">Jaccard</a> (on sets). For instance, in our <a href="../sqlpp/primer-sqlpp.html#ADM:_Modeling_Semistructured_Data_in_AsterixDB">TinySocial</a> example, the <tt>friendIds</tt> of a Gleambook user forms a set of friends, and we can define a similarity between the sets of friends of two users. We can also convert a string to a set of grams of a length &#x201c;n&#x201d; (called &#x201c;n-grams&#x201d;) and define the Jaccard similarity between the two gram sets of the two strings. Formally, the &#x201c;n-grams&#x201d; of a string are its substrings of length &#x201c;n&#x201d;. For instance, the 3-grams of the string <tt>schwarzenegger</tt> are <tt>sch</tt>, <tt>chw</tt>, <tt>hwa</tt>, &#x2026;, <tt>ger</tt>.</p>
<p>AsterixDB provides <a href="../sqlpp/builtins.html#Tokenizing_Functions">tokenization functions</a> to convert strings to sets, and the <a href="../sqlpp/builtins.html#Similarity_Functions">similarity functions</a>.</p></div>
<div class="section">
<h2><a name="Similarity_Selection_Queries_.5BBack_to_TOC.5D"></a><a name="SimilaritySelectionQueries" id="SimilaritySelectionQueries">Similarity Selection Queries</a> <font size="4"><a href="#toc">[Back to TOC]</a></font></h2>
<p>The following query asks for all the Gleambook users whose name is similar to <tt>Suzanna Tilson</tt>, i.e., their edit distance is at most 2.</p>
<div>
<div>
<pre class="source"> use TinySocial;
select u
from GleambookUsers u
where edit_distance(u.name, &quot;Suzanna Tilson&quot;) &lt;= 2;
</pre></div></div>
<p>The following query asks for all the Gleambook users whose set of friend ids is similar to <tt>[1,5,9,10]</tt>, i.e., their Jaccard similarity is at least 0.6.</p>
<div>
<div>
<pre class="source"> use TinySocial;
select u
from GleambookUsers u
where similarity_jaccard(u.friendIds, [1,5,9,10]) &gt;= 0.6f;
</pre></div></div>
<p>AsterixDB allows a user to use a similarity operator <tt>~=</tt> to express a condition by defining the similarity function and threshold using &#x201c;set&#x201d; statements earlier. For instance, the above query can be equivalently written as:</p>
<div>
<div>
<pre class="source"> use TinySocial;
set simfunction &quot;jaccard&quot;;
set simthreshold &quot;0.6f&quot;;
select u
from GleambookUsers u
where u.friendIds ~= [1,5,9,10];
</pre></div></div>
<p>In this query, we first declare Jaccard as the similarity function using <tt>simfunction</tt> and then specify the threshold <tt>0.6f</tt> using <tt>simthreshold</tt>.</p></div>
<div class="section">
<h2><a name="Similarity_Join_Queries_.5BBack_to_TOC.5D"></a><a name="SimilarityJoinQueries" id="SimilarityJoinQueries">Similarity Join Queries</a> <font size="4"><a href="#toc">[Back to TOC]</a></font></h2>
<p>AsterixDB supports fuzzy joins between two sets. The following <a href="../sqlpp/primer-sqlpp.html#Query_5_-_Fuzzy_Join">query</a> finds, for each Gleambook user, all Chirp users with names similar to their name based on the edit distance.</p>
<div>
<div>
<pre class="source"> use TinySocial;
set simfunction &quot;edit-distance&quot;;
set simthreshold &quot;3&quot;;
select gbu.id, gbu.name, (select cu.screenName, cu.name
from ChirpUsers cu
where cu.name ~= gbu.name) as similar_users
from GleambookUsers gbu;
</pre></div></div>
</div>
<div class="section">
<h2><a name="Using_Indexes_to_Support_Similarity_Queries_.5BBack_to_TOC.5D"></a><a name="UsingIndexesToSupportSimilarityQueries" id="UsingIndexesToSupportSimilarityQueries">Using Indexes to Support Similarity Queries</a> <font size="4"><a href="#toc">[Back to TOC]</a></font></h2>
<p>AsterixDB uses two types of indexes to support similarity queries, namely &#x201c;ngram index&#x201d; and &#x201c;keyword index&#x201d;.</p>
<div class="section">
<h3><a name="NGram_Index"></a>NGram Index</h3>
<p>An &#x201c;ngram index&#x201d; is constructed on a set of strings. We generate n-grams for each string, and build an inverted list for each n-gram that includes the ids of the strings with this gram. A similarity query can be answered efficiently by accessing the inverted lists of the grams in the query and counting the number of occurrences of the string ids on these inverted lists. The similar idea can be used to answer queries with Jaccard similarity. A detailed description of these techniques is available at this <a class="externalLink" href="http://www.ics.uci.edu/~chenli/pub/icde2009-memreducer.pdf">paper</a>.</p>
<p>For instance, the following DDL statements create an ngram index on the <tt>GleambookUsers.name</tt> attribute using an inverted index of 3-grams.</p>
<div>
<div>
<pre class="source"> use TinySocial;
create index gbUserIdx on GleambookUsers(name) type ngram(3);
</pre></div></div>
<p>The number &#x201c;3&#x201d; in &#x201c;ngram(3)&#x201d; is the length &#x201c;n&#x201d; in the grams. This index can be used to optimize similarity queries on this attribute using <a href="../sqlpp/builtins.html#edit_distance">edit_distance</a>, <a href="../sqlpp/builtins.html#edit_distance_check">edit_distance_check</a>, <a href="../sqlpp/builtins.html#similarity_jaccard">similarity_jaccard</a>, or <a href="../sqlpp/builtins.html#similarity_jaccard_check">similarity_jaccard_check</a> queries on this attribute where the similarity is defined on sets of 3-grams. This index can also be used to optimize queries with the &#x201c;<a href="(../sqlpp/builtins.html#contains">contains()</a>&#x201d; predicate (i.e., substring matching) since it can be also be solved by counting on the inverted lists of the grams in the query string.</p>
<div class="section">
<h4><a name="NGram_Index_usage_case_-_edit_distance"></a>NGram Index usage case - <a href="../sqlpp/builtins.html#edit-distance">edit_distance</a></h4>
<div>
<div>
<pre class="source"> use TinySocial;
select u
from GleambookUsers u
where edit_distance(u.name, &quot;Suzanna Tilson&quot;) &lt;= 2;
</pre></div></div>
</div>
<div class="section">
<h4><a name="NGram_Index_usage_case_-_edit_distance_check"></a>NGram Index usage case - <a href="../sqlpp/builtins.html#edit_distance_check">edit_distance_check</a></h4>
<div>
<div>
<pre class="source"> use TinySocial;
select u
from GleambookUsers u
where edit_distance_check(u.name, &quot;Suzanna Tilson&quot;, 2)[0];
</pre></div></div>
</div>
<div class="section">
<h4><a name="NGram_Index_usage_case_-_contains.28.29"></a>NGram Index usage case - <a href="(../sqlpp/builtins.html#contains">contains()</a></h4>
<div>
<div>
<pre class="source"> use TinySocial;
select m
from GleambookMessages m
where contains(m.message, &quot;phone&quot;);
</pre></div></div>
</div></div>
<div class="section">
<h3><a name="Keyword_Index"></a>Keyword Index</h3>
<p>A &#x201c;keyword index&#x201d; is constructed on a set of strings or sets (e.g., array, multiset). Instead of generating grams as in an ngram index, we generate tokens (e.g., words) and for each token, construct an inverted list that includes the ids of the objects with this token. The following two examples show how to create keyword index on two different types:</p>
<div class="section">
<h4><a name="Keyword_Index_on_String_Type"></a>Keyword Index on String Type</h4>
<div>
<div>
<pre class="source"> use TinySocial;
drop index GleambookMessages.gbMessageIdx if exists;
create index gbMessageIdx on GleambookMessages(message) type keyword;
select m
from GleambookMessages m
where similarity_jaccard_check(word_tokens(m.message), word_tokens(&quot;love like ccast&quot;), 0.2f)[0];
</pre></div></div>
</div>
<div class="section">
<h4><a name="Keyword_Index_on_Multiset_Type"></a>Keyword Index on Multiset Type</h4>
<div>
<div>
<pre class="source"> use TinySocial;
create index gbUserIdxFIds on GleambookUsers(friendIds) type keyword;
select u
from GleambookUsers u
where similarity_jaccard_check(u.friendIds, {{3,10}}, 0.5f)[0];
</pre></div></div>
<p>As shown above, keyword index can be used to optimize queries with token-based similarity predicates, including <a href="../sqlpp/builtins.html#similarity_jaccard">similarity_jaccard</a> and <a href="../sqlpp/builtins.html#similarity_jaccard_check">similarity_jaccard_check</a>.</p></div>
<div class="section">
<h4><a name="Keyword_Index_usage_case_-_similarity_jaccard"></a>Keyword Index usage case - <a href="../sqlpp/builtins.html#similarity_jaccard">similarity_jaccard</a></h4>
<div>
<div>
<pre class="source"> use TinySocial;
select u
from GleambookUsers u
where similarity_jaccard(u.friendIds, [1,5,9,10]) &gt;= 0.6f;
</pre></div></div>
</div>
<div class="section">
<h4><a name="Keyword_Index_usage_case_-_similarity_jaccard_check"></a>Keyword Index usage case - <a href="../sqlpp/builtins.html#similarity_jaccard_check">similarity_jaccard_check</a></h4>
<div>
<div>
<pre class="source"> use TinySocial;
select u
from GleambookUsers u
where similarity_jaccard_check(u.friendIds, [1,5,9,10], 0.6f)[0];
</pre></div></div></div></div></div>
</div>
</div>
</div>
<hr/>
<footer>
<div class="container-fluid">
<div class="row-fluid">
<div class="row-fluid">Apache AsterixDB, AsterixDB, Apache, the Apache
feather logo, and the Apache AsterixDB project logo are either
registered trademarks or trademarks of The Apache Software
Foundation in the United States and other countries.
All other marks mentioned may be trademarks or registered
trademarks of their respective owners.
</div>
</div>
</div>
</footer>
</body>
</html>