ibexa

Caution: This documentation is for eZ Publish legacy, from version 3.x to 5.x.

More Like This

With the "More Like This"-functionality and template function, you can use the eZ Find back-end to provide similar objects or pages for a given object. Instead of providing keywords, the "More Like This" function accepts a node id, object id, URL or a blob of text. After analyzing the text, the Solr engine will create a search with keywords chosen with heuristics that depend on the indexed documents as well as the object fed to it.

Besides the special query parameters specified below, facets, filters and sorting parameters as in the regular search function are supported as well.

Parameter

Values/Description

query_type

string with possible values ('nid' | 'oid' | 'text' | 'url' )
 nid: node id
 oid: object id
 text: a blob of text
 url: a url, accessible from the Solr back-end

query

the value, depending on query type

In some cases, you may want to provide only part of an object to ensure that you get similar objects as well. The URL can be useful for links to external pages, or when you want to show internal objects that resemble the content behind the URL.

You can tune the More Like This function to your own liking, but know that this is a very experimental procedure. It is done with the .ini setting ExtractionFields, which sets the fields to use for query term extraction. The proper fields are “native” and “general”. The field "general" currently uses the aggregate field ezf_df_t, but this is subject to change, therefore it is recommended to have a the dedicated aggregator fields with his own analyzers. You are able to tune the "More Like This" feature with the following settings:

MinTermFreq=1. Will set the minimum frequency
MinDocFreq=1, will set the minimum document frequency
MinWordLenght=4, will set the minimum word length
MaxWordLength=30, will set the maximum word length.
MaxQueryTerms=10, will set the Maximum query terms to extract from the object
BoostTerms=True, will enable the use of popularity to boost query terms.

Ricardo Correia (12/09/2013 3:46 pm)

Ricardo Correia (12/09/2013 3:46 pm)


Comments

There are no comments.