1.4.1

Removing a trailing Mage::log

1.4.0

New features :

• Added the ability to choose a "Root Category" on Virtual Categories : This allows to have category facet from this root category displayed.
• Publishing "Optimizers" and "Behavioral Search" to Open Source : https://github.com/Smile-SA/smile-magento-elasticsearch/blob/release_1.4/doc/behavorial-search.md
• Added "Virtual Attributes" module : This can be used to create attributes based on rules
• Refactoring the previous Tracker (made as an elasticsearch plugin) to a Logstash based tracker.

✅ Fixed

• A bug which could cause issues with undefined indexes on layered navigation (#63)
• A bug when creating categories via SOAP (#68)
• A bug where search query containing special chars (especially slashes) would fail. (#66)
• A bug causing categories not being indexed properly (#71)
• Not existing option values were indexed.
• Issue with static caching on query building.
• Javascript bug when using range slider.
• Issues related to virtual categories conditions building.

Maintenance release 1.3.2

Fixed :

• Categories that were not properly indexed.
• Preventing triggering indexation on empty product ids list for virtual categories.

Maintenance release 1.3

New features :

• New sort order on facets : "Admin sort" that displays facets options sorted by the attribute's options sort order defined in Back-Office.
• Various performances improvements
  • Replication is now stopped during full reindex
  • Better performances on clusters
• Revamp of categories indexation :
  • Performance gain during indexation
  • Ability to choose if category names should be use for fulltext search on products or not (for each category in the back office).

Relevance improvements :

Stopwords and cutoff frequency :

Stopwords are now indexed. Fulltext queries now uses Cutoff Frequency (https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-match-query.html#query-dsl-match-query-cutoff).

Based on this Cutoff Frequency, the engine is now able to determine which words are stopwords and autodetect them. They will trigger a boost on the results if they are found, but won't impact filtering. Eg. the following query :

"The little horse" : Conjunction of "little" and "horse" is the mandatory part of the query for matching. "the" is optional part of the query, but if the word is found, a boost will be applied on matched results.

Cutoff frequency can be managed on the Back-Office via System > Configuration > Smile Searchandising Suite > Advanced Search Settings > Cutoff Frequency.

Phrase matching :

Phrase matching has also been revamped. It allows to apply a boost on phrase "chunks" if they are found in documents. Eg for the query "white horse", if "white" and "horse" appears in the document, and also the position of the term "horse" is 1 more than the term "white", the document will have a boost into results.

Phrase matching boost can be managed on the Back-Office via System > Configuration > Smile Searchandising Suite > Advanced Search Settings > Phrase match configuration.

Spellchecking :

Spellchecking has also been improved to benefits both precedents points. It's relevance is now far better than before and additionally, the engine is now able to deal with pure stopwords queries such as "to be or not to be", or "he and she".

Boolean attributes :

Boolean attributes indexation has been reworked too : attribute label is now indexed for products having attribute's value to "Yes", this means a product (let's say, a watch) having a "Adjustable" boolean attribute set to "Yes" will now match a fulltext query like "adjustable watch".

Fixed :

• Autocomplete blocks now supports generic blocks cache tags
• Bug where multiselect attributes were not properly merged between parent and children products
• Bug where rating facet was computed on non-approved ratings

Maintenance release 1.2.1

Fixed :

• Missing foreign keys on catalog_product_entity table for products positions in search results and in virtual categories.

Maintenance release 1.2

New :

• Implementation of a Data Provider pattern to permit easy addition of content to the index. (documentation is here : Indexing custom content).
• For each search query (via the search term edition screen in Back Office), added the ability to fix products positions in the results.
• Added a live preview on virtual categories.
• On virtual categories edition, it is now possible to fix products positions in the results.

Fixed :

• Various typographic bugs
• Issue where newly created category appeared without any name in Back-Office.

Maintenance release 1.1

• Better autocomplete and fuzzy search
• Bugfixes

Initial OSS Release 1.0

Initial OpenSource module release. Enjoy !!!