Example 12: Free Text indexing¶
It is common for users to build RDF applications that combine some
form of “keyword search” with their queries. For example, a user might
want to retrieve all triples for which the string “Alice” appears as a
word within the third (object) field of the triple. AllegroGraph
provides a capability for including free text matching within a SPARQL
query, and also by using the evalFreeTextSearch() method of the
connection object. It requires, however, that you create and configure
indexes appropriate to the searches you want to pursue.
First let’s open a connection
from franz.openrdf.connect import ag_connect
conn = ag_connect('python-tutorial', create=True, clear=True)
We will start this example by importing some sample data
conn.addData("""
@prefix : <ex://> .
:alice a :Person ;
:fullname "Alice B. Toklas" .
:book1 a :Book ;
:title "Alice in Wonderland" ;
:author :carroll .
:carroll a :Person ;
:fullname "Lewis Carroll" .""")
We have to create an index. AllegroGraph lets you create any number of
text indexes, each for a specific purpose. In this case we are
indexing the literal values we find in the fullname predicate,
which we have used in resources that describe people. The
createFreeTextIndex() method has many configurable
parameters. Their default settings are appropriate to this
situation. All we have to provide is a name for the index and the URI
of the predicate (or predicates) that contain the text to be indexed.
fullname = conn.createURI(namespace='ex://',
localname='fullname')
conn.createFreeTextIndex(
"index1", predicates=[fullname])
We can view the index configuration using the
getFreeTextIndexConfiguration() method:
config = conn.getFreeTextIndexConfiguration("index1")
for key, value in config.items():
print('{key}: {value}'.format(key=key, value=value))
tokenizer: default
indexLiterals: True
minimumWordSize: 3
indexFields: [u'object']
stopWords: ...
innerChars: []
predicates: [<ex://fullname>]
wordFilters: []
indexResources: False
borderChars: []
This configuration says that index1 will operate on the literal
values it finds in the object position of the <ex://fullname>
predicate. It ignores words smaller than three characters in
length. It will ignore the words in its stopWords list (elided
from sample output). If it encounters a resource URI in the object
position, it will ignore it. This index doesn’t use any
wordFilters, which are sometimes used to remove accented letters
and to perform stemming on indexed text and search strings.
The text match occurs through a “magic” predicate called fti:match. This predicate has two arguments. One is the subject URI of the resources to search. The other is the string pattern to search for, such as “Alice”. Only full-word matches will be found.
query = conn.prepareTupleQuery(query="""
SELECT ?s WHERE {
?s fti:match "Alice" .
}""")
query.evaluate(output=True)
There is no need to include a prefix declaration for the fti
namespace. That is because fti is included among the built-in
namespace mappings in AllegroGraph.
When we execute our SPARQL query, it matches the "Alice" within the literal "Alice B. Toklas" because that literal occurs in a triple having the fullname predicate, but it does not match the “Alice” in the literal "Alice in Wonderland" because the title predicate was not included in our index.
--------------
| s |
==============
| ex://alice |
--------------
By default fti:match searches in all text indexes. It is possible
to specify a single index name when searching. We’ll illustrate this
be creating another index, this time on the title predicate:
title = conn.createURI(namespace='ex://',
localname='title')
conn.createFreeTextIndex(
"index2", predicates=[title])
query = conn.prepareTupleQuery(query="""
SELECT ?s WHERE {
?s fti:match ( "Alice" "index2" ) .
}""")
query.evaluate(output=True)
This time only the book title will match our query
--------------
| s |
==============
| ex://book1 |
--------------
Another way of searching text indexes is the
evalFreeTextSearch() method:
for triple in conn.evalFreeTextSearch(
"Alice", index="index1"):
print(triple[0])
This works just like our first query. Note that
evalFreeTextSearch() returns a list of lists of strings (in
N-Triples format), not a list of Statement objects.
<ex://alice>
The text index supports simple wildcard queries. The asterisk (*)
may be appended to the end of the pattern to indicate “any number of
additional characters.” For instance, this query looks for whole words
that begin with “Ali”:
for triple in conn.evalFreeTextSearch("Ali*"):
print(triple[0])
This search runs across both indexes, so it will find both the
:title and the :fullname triples.
<ex://alice>
<ex://book1>
There is also a single-character wildcard, the question mark. It will match any single character. You can add as many question marks as you need to the string pattern. This query looks for a five-letter word that has “l” in the second position, and “c” in the fourth position:
for triple in conn.evalFreeTextSearch("?l?c?*"):
print(triple[0])
The result is the same as for the previous query
<ex://alice>
<ex://book1>
Text indexes are not the only way of matching text values available in SPARQL. One may also filter results using regular expressions. This approach is more flexible, but at the price of performance. Regular expression filters do not use any form of indexing to speed up the query.
query = conn.prepareTupleQuery(query="""
SELECT ?s ?p ?o WHERE {
?s ?p ?o .
FILTER regex(?o, "lic|oll")
}""")
query.evaluate(output=True)
Note how this search matches the provided pattern inside words.
------------------------------------------------------
| s | p | o |
======================================================
| ex://carroll | ex://fullname | Lewis Carroll |
| ex://book1 | ex://title | Alice in Wonderland |
| ex://alice | ex://fullname | Alice B. Toklas |
------------------------------------------------------
In addition to indexing literal values, AllegroGraph can also index
resource URIs. index3 is an index that looks for URIs in the
object position of the author predicate, and then indexes only the
local name of the resource (the characters following the rightmost
/, # or : in the URI). This lets us avoid indexing
highly-repetitive namespace strings, which would fill the index with
data that would not be very useful.
author = conn.createURI(namespace='ex://',
localname='author')
conn.createFreeTextIndex(
"index3", predicates=[author],
indexResources="short", indexFields=["object"])
for triple in conn.evalFreeTextSearch("carroll",
index="index3"):
print(triple[0])
The text search located the triple that has carroll in the URI in
the object position:
<ex://book1>
