: I have a use case where the facet hierarchies as well as facet names change : very frequently. : : For example: : (Smartphones >> Android ) may become : Smartphones >> GSM >> And roid. : : OR : "Smartphone" could be renamed to "Smart Phone" : : If I use traditional hierarchical faceting, then every change would mean a : re-index of a large number of documents. : : Just curious to know how others have solved this problem in the past.
I've dealt with this in the past using a custom plugin for the faceting. basically each document had a category field that only contained the id# of a category it was directly in, and the actaul hierarchy info was stored in an XML data file that the plugin loaded at init and used to build the query associated with each node by looking at all the categoryId number from all hte descendent categories (optimizations can be made if you know documents are only mapped to leaf level categories, or if you can define your hierarchy in terms of other fields -- ie: catId#345might be definable by the query "type:phone AND os:android AND tech:GSM") for small hiarchies, you can do the same thing from any solr client that knows what hierarchy you have usng many facet.queries - just put whatever info you need to remap the flat facet.query responses into a hierarchy as localparams on each facet.query. -Hoss
