rank.ac.readfrom

Necessity:mandatory
Value:name of a field
Default:can be configured in the solrconfig.xml

This parameter is optional, but necessary for enabling Auto Context. Set the name of the field contains the keywords used by the Auto Context feature.

The referenced field in the schema must be configured as stored="true". Auto Context will read those values and processes them as internal search input on the field defined by rank.ac.searchin.

You can configure a default fieldname in the pixolutionParserPlugin section in the solrconfig.xml (see PixolutionParserPlugin). If rank.ac.readfrom is part of the query, the solrconfig.xml configuration will be ignored.

Configuring rank.ac.readfrom at runtime:

&pixolution=true&rank.by=id:123&rank.ac=true&rank.ac.readfrom=keywords

rank.ac.searchin

Necessity:mandatory
Value:name of a field
Default:can be configured in the solrconfig.xml

This parameter is optional, but necessary for enabling Auto Context. Set the name of the field in which to search for the keywords used by the Auto Context.

The referenced field in the schema must be configured as indexed="true". Auto Context will use the keywords retrieved from the rank.ac.readfrom field and searches for them in the field defined by rank.ac.searchin. Before searching in rank.ac.searchin its analyzer chain is applied.

You can configure a default fieldname in the solrconfig.xml pixolutionParserPlugin section (see PixolutionParserPlugin). If rank.ac.searchin is part of the query, the solrconfig.xml configuration will be ignored.

Inject keywords from field keywords in the query and search for them in the field alltext:

&pixolution=true&rank.by=id:123&rank.ac=true&rank.ac.readfrom=keywords&rank.ac.searchin=alltext

Configure analyzer chain

Make sure that your referenced field has a proper field analyzer chain, so that Auto Context is able to transform text from rank.ac.readfrom into useful keyword terms. Also, internally the char , has a special meaning causing pixolution flow to split a string into several parts. The string ”one, two, three” would result in ”one”, ” two” and ” three” even without field analyzers. Avoid char , in your keywords and split them via configured analyzer chain.

rank.ac.clauselimit

Necessity:optional
Value:>0
Default:unlimited

A Solr query consists of one or several clauses. Auto Context produces clauses from keywords and extends the query. With rank.ac.clauselimit you can limit the maximum number of boolean clauses that will be used by Auto Context.

If the Auto Context field of a document contains lots of keywords the performance may decrease significantly because Solr does a lot of preprocessing while determining the document subset which will be part of the scoring process. You may use this rank.ac.clauselimit to avoid performance decrease.

If 100 keywords are associated to a specific image and rank.ac.clauselimit=20, then only the first 20 boolean clauses that are built from the keywords are used as Auto Context input. The remaining keywords will not be part of the search.

Note: Depending on the image and its associated number of keywords not all keywords will be part of the search. This may reduce result quality because not all available textual information is part of the query. Only use this parameter if you experience bad performance and cannot use a different Auto Context field with fewer keywords (rank.ac.readfrom, see rank.ac.readfrom).

Example for searching by image id 123 with Auto Context and limiting the injected clauses to the first 20 clauses:

&pixolution=true&rank.by=id:123&rank.ac=true&rank.ac.clauselimit=20

mm

Necessity:optional
Value:>=0 | 0% - 100%

The minimum match parameter mm is used for relevant document retrieval when Auto Context is activated (rank.ac=true). This parameter is provided by Solr and used by Auto Context. The mm parameter controls how many optional terms (e.g. keywords) must match a document, in order to be part of the relevancy calculations.

You may set a positive integer to define the minimum number of terms that must match, regardless of how many terms there are in total. For example mm=3 means three terms must match at least. It is also possible to define a percent value, e.g. mm=50%. This means that at least 50% of the given terms must match the example document in order to be part in visual search scorings.

With mm you can reduce the amount of documents, used for a visual search. The more documents are filtered before a visual search, the more performance you can gain. Similarity calculations will only be done for documents considered relevant.

Example for searching by image id 123 where at least 25% of its keywords occur in similar images:

&pixolution=true&rank.by=id:123&rank.ac=true&mm=25%