| Current File : //usr/share/doc/postgresql-9.2.24/html/textsearch-intro.html |
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<HTML
><HEAD
><TITLE
>Introduction</TITLE
><META
NAME="GENERATOR"
CONTENT="Modular DocBook HTML Stylesheet Version 1.79"><LINK
REV="MADE"
HREF="mailto:pgsql-docs@postgresql.org"><LINK
REL="HOME"
TITLE="PostgreSQL 9.2.24 Documentation"
HREF="index.html"><LINK
REL="UP"
TITLE="Full Text Search"
HREF="textsearch.html"><LINK
REL="PREVIOUS"
TITLE="Full Text Search"
HREF="textsearch.html"><LINK
REL="NEXT"
TITLE="Tables and Indexes"
HREF="textsearch-tables.html"><LINK
REL="STYLESHEET"
TYPE="text/css"
HREF="stylesheet.css"><META
HTTP-EQUIV="Content-Type"
CONTENT="text/html; charset=ISO-8859-1"><META
NAME="creation"
CONTENT="2017-11-06T22:43:11"></HEAD
><BODY
CLASS="SECT1"
><DIV
CLASS="NAVHEADER"
><TABLE
SUMMARY="Header navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TH
COLSPAN="5"
ALIGN="center"
VALIGN="bottom"
><A
HREF="index.html"
>PostgreSQL 9.2.24 Documentation</A
></TH
></TR
><TR
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
TITLE="Full Text Search"
HREF="textsearch.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
HREF="textsearch.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="60%"
ALIGN="center"
VALIGN="bottom"
>Chapter 12. Full Text Search</TD
><TD
WIDTH="20%"
ALIGN="right"
VALIGN="top"
><A
TITLE="Tables and Indexes"
HREF="textsearch-tables.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
></TABLE
><HR
ALIGN="LEFT"
WIDTH="100%"></DIV
><DIV
CLASS="SECT1"
><H1
CLASS="SECT1"
><A
NAME="TEXTSEARCH-INTRO"
>12.1. Introduction</A
></H1
><P
> Full Text Searching (or just <I
CLASS="FIRSTTERM"
>text search</I
>) provides
the capability to identify natural-language <I
CLASS="FIRSTTERM"
>documents</I
> that
satisfy a <I
CLASS="FIRSTTERM"
>query</I
>, and optionally to sort them by
relevance to the query. The most common type of search
is to find all documents containing given <I
CLASS="FIRSTTERM"
>query terms</I
>
and return them in order of their <I
CLASS="FIRSTTERM"
>similarity</I
> to the
query. Notions of <TT
CLASS="VARNAME"
>query</TT
> and
<TT
CLASS="VARNAME"
>similarity</TT
> are very flexible and depend on the specific
application. The simplest search considers <TT
CLASS="VARNAME"
>query</TT
> as a
set of words and <TT
CLASS="VARNAME"
>similarity</TT
> as the frequency of query
words in the document.
</P
><P
> Textual search operators have existed in databases for years.
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> has
<TT
CLASS="LITERAL"
>~</TT
>, <TT
CLASS="LITERAL"
>~*</TT
>, <TT
CLASS="LITERAL"
>LIKE</TT
>, and
<TT
CLASS="LITERAL"
>ILIKE</TT
> operators for textual data types, but they lack
many essential properties required by modern information systems:
</P
><P
></P
><UL
COMPACT="COMPACT"
><LI
STYLE="list-style-type: disc"
><P
> There is no linguistic support, even for English. Regular expressions
are not sufficient because they cannot easily handle derived words, e.g.,
<TT
CLASS="LITERAL"
>satisfies</TT
> and <TT
CLASS="LITERAL"
>satisfy</TT
>. You might
miss documents that contain <TT
CLASS="LITERAL"
>satisfies</TT
>, although you
probably would like to find them when searching for
<TT
CLASS="LITERAL"
>satisfy</TT
>. It is possible to use <TT
CLASS="LITERAL"
>OR</TT
>
to search for multiple derived forms, but this is tedious and error-prone
(some words can have several thousand derivatives).
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> They provide no ordering (ranking) of search results, which makes them
ineffective when thousands of matching documents are found.
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> They tend to be slow because there is no index support, so they must
process all documents for every search.
</P
></LI
></UL
><P
> Full text indexing allows documents to be <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>preprocessed</I
></SPAN
>
and an index saved for later rapid searching. Preprocessing includes:
</P
><P
></P
><UL
><LI
STYLE="list-style-type: none"
><P
> <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>Parsing documents into <I
CLASS="FIRSTTERM"
>tokens</I
></I
></SPAN
>. It is
useful to identify various classes of tokens, e.g., numbers, words,
complex words, email addresses, so that they can be processed
differently. In principle token classes depend on the specific
application, but for most purposes it is adequate to use a predefined
set of classes.
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> uses a <I
CLASS="FIRSTTERM"
>parser</I
> to
perform this step. A standard parser is provided, and custom parsers
can be created for specific needs.
</P
></LI
><LI
STYLE="list-style-type: none"
><P
> <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>Converting tokens into <I
CLASS="FIRSTTERM"
>lexemes</I
></I
></SPAN
>.
A lexeme is a string, just like a token, but it has been
<I
CLASS="FIRSTTERM"
>normalized</I
> so that different forms of the same word
are made alike. For example, normalization almost always includes
folding upper-case letters to lower-case, and often involves removal
of suffixes (such as <TT
CLASS="LITERAL"
>s</TT
> or <TT
CLASS="LITERAL"
>es</TT
> in English).
This allows searches to find variant forms of the
same word, without tediously entering all the possible variants.
Also, this step typically eliminates <I
CLASS="FIRSTTERM"
>stop words</I
>, which
are words that are so common that they are useless for searching.
(In short, then, tokens are raw fragments of the document text, while
lexemes are words that are believed useful for indexing and searching.)
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> uses <I
CLASS="FIRSTTERM"
>dictionaries</I
> to
perform this step. Various standard dictionaries are provided, and
custom ones can be created for specific needs.
</P
></LI
><LI
STYLE="list-style-type: none"
><P
> <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>Storing preprocessed documents optimized for
searching</I
></SPAN
>. For example, each document can be represented
as a sorted array of normalized lexemes. Along with the lexemes it is
often desirable to store positional information to use for
<I
CLASS="FIRSTTERM"
>proximity ranking</I
>, so that a document that
contains a more <SPAN
CLASS="QUOTE"
>"dense"</SPAN
> region of query words is
assigned a higher rank than one with scattered query words.
</P
></LI
></UL
><P
> Dictionaries allow fine-grained control over how tokens are normalized.
With appropriate dictionaries, you can:
</P
><P
></P
><UL
COMPACT="COMPACT"
><LI
STYLE="list-style-type: disc"
><P
> Define stop words that should not be indexed.
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> Map synonyms to a single word using <SPAN
CLASS="APPLICATION"
>Ispell</SPAN
>.
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> Map phrases to a single word using a thesaurus.
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> Map different variations of a word to a canonical form using
an <SPAN
CLASS="APPLICATION"
>Ispell</SPAN
> dictionary.
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> Map different variations of a word to a canonical form using
<SPAN
CLASS="APPLICATION"
>Snowball</SPAN
> stemmer rules.
</P
></LI
></UL
><P
> A data type <TT
CLASS="TYPE"
>tsvector</TT
> is provided for storing preprocessed
documents, along with a type <TT
CLASS="TYPE"
>tsquery</TT
> for representing processed
queries (<A
HREF="datatype-textsearch.html"
>Section 8.11</A
>). There are many
functions and operators available for these data types
(<A
HREF="functions-textsearch.html"
>Section 9.13</A
>), the most important of which is
the match operator <TT
CLASS="LITERAL"
>@@</TT
>, which we introduce in
<A
HREF="textsearch-intro.html#TEXTSEARCH-MATCHING"
>Section 12.1.2</A
>. Full text searches can be accelerated
using indexes (<A
HREF="textsearch-indexes.html"
>Section 12.9</A
>).
</P
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="TEXTSEARCH-DOCUMENT"
>12.1.1. What Is a Document?</A
></H2
><P
> A <I
CLASS="FIRSTTERM"
>document</I
> is the unit of searching in a full text search
system; for example, a magazine article or email message. The text search
engine must be able to parse documents and store associations of lexemes
(key words) with their parent document. Later, these associations are
used to search for documents that contain query words.
</P
><P
> For searches within <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
>,
a document is normally a textual field within a row of a database table,
or possibly a combination (concatenation) of such fields, perhaps stored
in several tables or obtained dynamically. In other words, a document can
be constructed from different parts for indexing and it might not be
stored anywhere as a whole. For example:
</P><PRE
CLASS="PROGRAMLISTING"
>SELECT title || ' ' || author || ' ' || abstract || ' ' || body AS document
FROM messages
WHERE mid = 12;
SELECT m.title || ' ' || m.author || ' ' || m.abstract || ' ' || d.body AS document
FROM messages m, docs d
WHERE mid = did AND mid = 12;</PRE
><P>
</P
><DIV
CLASS="NOTE"
><BLOCKQUOTE
CLASS="NOTE"
><P
><B
>Note: </B
> Actually, in these example queries, <CODE
CLASS="FUNCTION"
>coalesce</CODE
>
should be used to prevent a single <TT
CLASS="LITERAL"
>NULL</TT
> attribute from
causing a <TT
CLASS="LITERAL"
>NULL</TT
> result for the whole document.
</P
></BLOCKQUOTE
></DIV
><P
> Another possibility is to store the documents as simple text files in the
file system. In this case, the database can be used to store the full text
index and to execute searches, and some unique identifier can be used to
retrieve the document from the file system. However, retrieving files
from outside the database requires superuser permissions or special
function support, so this is usually less convenient than keeping all
the data inside <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
>. Also, keeping
everything inside the database allows easy access
to document metadata to assist in indexing and display.
</P
><P
> For text search purposes, each document must be reduced to the
preprocessed <TT
CLASS="TYPE"
>tsvector</TT
> format. Searching and ranking
are performed entirely on the <TT
CLASS="TYPE"
>tsvector</TT
> representation
of a document — the original text need only be retrieved
when the document has been selected for display to a user.
We therefore often speak of the <TT
CLASS="TYPE"
>tsvector</TT
> as being the
document, but of course it is only a compact representation of
the full document.
</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="TEXTSEARCH-MATCHING"
>12.1.2. Basic Text Matching</A
></H2
><P
> Full text searching in <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> is based on
the match operator <TT
CLASS="LITERAL"
>@@</TT
>, which returns
<TT
CLASS="LITERAL"
>true</TT
> if a <TT
CLASS="TYPE"
>tsvector</TT
>
(document) matches a <TT
CLASS="TYPE"
>tsquery</TT
> (query).
It doesn't matter which data type is written first:
</P><PRE
CLASS="PROGRAMLISTING"
>SELECT 'a fat cat sat on a mat and ate a fat rat'::tsvector @@ 'cat & rat'::tsquery;
?column?
----------
t
SELECT 'fat & cow'::tsquery @@ 'a fat cat sat on a mat and ate a fat rat'::tsvector;
?column?
----------
f</PRE
><P>
</P
><P
> As the above example suggests, a <TT
CLASS="TYPE"
>tsquery</TT
> is not just raw
text, any more than a <TT
CLASS="TYPE"
>tsvector</TT
> is. A <TT
CLASS="TYPE"
>tsquery</TT
>
contains search terms, which must be already-normalized lexemes, and
may combine multiple terms using AND, OR, and NOT operators.
(For details see <A
HREF="datatype-textsearch.html"
>Section 8.11</A
>.) There are
functions <CODE
CLASS="FUNCTION"
>to_tsquery</CODE
> and <CODE
CLASS="FUNCTION"
>plainto_tsquery</CODE
>
that are helpful in converting user-written text into a proper
<TT
CLASS="TYPE"
>tsquery</TT
>, for example by normalizing words appearing in
the text. Similarly, <CODE
CLASS="FUNCTION"
>to_tsvector</CODE
> is used to parse and
normalize a document string. So in practice a text search match would
look more like this:
</P><PRE
CLASS="PROGRAMLISTING"
>SELECT to_tsvector('fat cats ate fat rats') @@ to_tsquery('fat & rat');
?column?
----------
t</PRE
><P>
Observe that this match would not succeed if written as
</P><PRE
CLASS="PROGRAMLISTING"
>SELECT 'fat cats ate fat rats'::tsvector @@ to_tsquery('fat & rat');
?column?
----------
f</PRE
><P>
since here no normalization of the word <TT
CLASS="LITERAL"
>rats</TT
> will occur.
The elements of a <TT
CLASS="TYPE"
>tsvector</TT
> are lexemes, which are assumed
already normalized, so <TT
CLASS="LITERAL"
>rats</TT
> does not match <TT
CLASS="LITERAL"
>rat</TT
>.
</P
><P
> The <TT
CLASS="LITERAL"
>@@</TT
> operator also
supports <TT
CLASS="TYPE"
>text</TT
> input, allowing explicit conversion of a text
string to <TT
CLASS="TYPE"
>tsvector</TT
> or <TT
CLASS="TYPE"
>tsquery</TT
> to be skipped
in simple cases. The variants available are:
</P><PRE
CLASS="PROGRAMLISTING"
>tsvector @@ tsquery
tsquery @@ tsvector
text @@ tsquery
text @@ text</PRE
><P>
</P
><P
> The first two of these we saw already.
The form <TT
CLASS="TYPE"
>text</TT
> <TT
CLASS="LITERAL"
>@@</TT
> <TT
CLASS="TYPE"
>tsquery</TT
>
is equivalent to <TT
CLASS="LITERAL"
>to_tsvector(x) @@ y</TT
>.
The form <TT
CLASS="TYPE"
>text</TT
> <TT
CLASS="LITERAL"
>@@</TT
> <TT
CLASS="TYPE"
>text</TT
>
is equivalent to <TT
CLASS="LITERAL"
>to_tsvector(x) @@ plainto_tsquery(y)</TT
>.
</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="TEXTSEARCH-INTRO-CONFIGURATIONS"
>12.1.3. Configurations</A
></H2
><P
> The above are all simple text search examples. As mentioned before, full
text search functionality includes the ability to do many more things:
skip indexing certain words (stop words), process synonyms, and use
sophisticated parsing, e.g., parse based on more than just white space.
This functionality is controlled by <I
CLASS="FIRSTTERM"
>text search
configurations</I
>. <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> comes with predefined
configurations for many languages, and you can easily create your own
configurations. (<SPAN
CLASS="APPLICATION"
>psql</SPAN
>'s <TT
CLASS="COMMAND"
>\dF</TT
> command
shows all available configurations.)
</P
><P
> During installation an appropriate configuration is selected and
<A
HREF="runtime-config-client.html#GUC-DEFAULT-TEXT-SEARCH-CONFIG"
>default_text_search_config</A
> is set accordingly
in <TT
CLASS="FILENAME"
>postgresql.conf</TT
>. If you are using the same text search
configuration for the entire cluster you can use the value in
<TT
CLASS="FILENAME"
>postgresql.conf</TT
>. To use different configurations
throughout the cluster but the same configuration within any one database,
use <TT
CLASS="COMMAND"
>ALTER DATABASE ... SET</TT
>. Otherwise, you can set
<TT
CLASS="VARNAME"
>default_text_search_config</TT
> in each session.
</P
><P
> Each text search function that depends on a configuration has an optional
<TT
CLASS="TYPE"
>regconfig</TT
> argument, so that the configuration to use can be
specified explicitly. <TT
CLASS="VARNAME"
>default_text_search_config</TT
>
is used only when this argument is omitted.
</P
><P
> To make it easier to build custom text search configurations, a
configuration is built up from simpler database objects.
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
>'s text search facility provides
four types of configuration-related database objects:
</P
><P
></P
><UL
COMPACT="COMPACT"
><LI
STYLE="list-style-type: disc"
><P
> <I
CLASS="FIRSTTERM"
>Text search parsers</I
> break documents into tokens
and classify each token (for example, as words or numbers).
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> <I
CLASS="FIRSTTERM"
>Text search dictionaries</I
> convert tokens to normalized
form and reject stop words.
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> <I
CLASS="FIRSTTERM"
>Text search templates</I
> provide the functions underlying
dictionaries. (A dictionary simply specifies a template and a set
of parameters for the template.)
</P
></LI
><LI
STYLE="list-style-type: disc"
><P
> <I
CLASS="FIRSTTERM"
>Text search configurations</I
> select a parser and a set
of dictionaries to use to normalize the tokens produced by the parser.
</P
></LI
></UL
><P
> Text search parsers and templates are built from low-level C functions;
therefore it requires C programming ability to develop new ones, and
superuser privileges to install one into a database. (There are examples
of add-on parsers and templates in the <TT
CLASS="FILENAME"
>contrib/</TT
> area of the
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> distribution.) Since dictionaries and
configurations just parameterize and connect together some underlying
parsers and templates, no special privilege is needed to create a new
dictionary or configuration. Examples of creating custom dictionaries and
configurations appear later in this chapter.
</P
></DIV
></DIV
><DIV
CLASS="NAVFOOTER"
><HR
ALIGN="LEFT"
WIDTH="100%"><TABLE
SUMMARY="Footer navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
><A
HREF="textsearch.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="index.html"
ACCESSKEY="H"
>Home</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
><A
HREF="textsearch-tables.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
>Full Text Search</TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="textsearch.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
>Tables and Indexes</TD
></TR
></TABLE
></DIV
></BODY
></HTML
>