JDONREFv4 Plugin

De JDONREF Wiki
Révision de 2 décembre 2014 à 00:20 par Julien2512 (discussion | contributions) (Includes)

The JDONREF plugin provides a way to index adresses compatible with french format and search them in a single line format, with 100% accurate results and great order.

It works over the provided PayloadCheckerSpanQuery that target to combines whatever work rules you want (not only adresses) to search data in a single line format, with 100% accurate results and great order. Scorers are adaptable for human or computer uses.

Plugin installation

Simply run at the root of your ElasticSearch v1.3+ installation :

bin/plugin --install elasticsearch-jdonrefv4-0.2 --url http://sourceforge.net/p/jdonref/code/HEAD/tree/Dev/JDONREFv4/dist/elasticsearch-jdonrefv4-0.2.zip?format=raw

This will download the last beta version plugin from sourceforge.

Don't forget restart elasticsearch.


You can follow the getting started guide for jdonref plugin. The same will arise soon for PayloadCheckerSpanQuery.

There is an embedded site in order to test the query like this :

http://url_elasticsearch:port/_plugin/jdonrefv4-0.2/JDONREFv3Test.html

It only supports index named 'jdonref' for now. Of course there is no such limitation for jdonrefv4 nor PayloadCheckerSpanQuery.

Releases
JDONREF plugin elasticsearch release date detail
0.1beta 1.0.0.RC1 2014-04-30 POC
0.2beta 1.1.X 2014-08-25 adress based score with integration tests
0.2 1.3.X 2014-11-15 (testing) production version : adress based score with production tests. human validation only. miss bulk scorer and explain.
0.3 1.3.X 2015-12-15 bulk scorer, and explain.
0.4 1.3.X 2015-03-31 score tuning
Includes
name type disponibilité
jdonrefv4 query/scorer since 0.2beta (weights and term can't be changed). Read PayloadCheckerSpanQuery.
PayloadCheckerSpanQuery query/scorer since 0.2beta (grammar filter based on payloads used by jdonrefv4).
UnsplitFilter token filter since 0.3beta (reduce tokens frequencies).
POI et Zones, adresse, troncon, voie, commune, departement, pays mapping since 0.2beta
jdonrefv3es_synonym.fr.txt French Synonyms'file since 0.1beta
jdonrefv4-settings.json settings since 0.2beta jdonrefv4_index, jdonrefv4_search, jdonrefv4_codes_index
JDONREFv4TermSimilarity similarity since 0.2beta
EdgeNGramWithPayloadsFilter token filter since 0.2beta (edgengram with payloads support)
TokenCountPayloadFilter token filter since 0.2beta (insert the number of token with given payload in each such payload - used by PayloadCheckerSpanQuery#All).
SynonymWithPayloadsFilter token filter since 0.2beta (synonym with payloads support).
JDONREFv3Test.html site since 0.2beta (site for jdonrefv3es and index jdonref) Set 10000 for maxSizePerType with dedicated index of 5 shards for each type !

See also todo list.