How record sorting and relevance ranking works?

There are three ways of controlling the order in which records are returned:
1) Explicit sort key using Ns parameter
2) Default sort using Dgidx flag
3) Relevancy Ranking

The priority of record sorting/relevance ranking is as follows:

1) If none of the above  sorting methods is specified then Endeca uses order of internal ID generated by Dgidx during indexing

2) If Sort key is used(Ns parameter) then sort key overrides all other sorting and relevance ranking settings

3) If a default sort key is specified then records are returned in default sort order(Ties are broken using the internal order by Dgidx)

4) When searching against a search interface relevance ranking strategy is used(ties are broken using the default sort key and if no sort key then ties are broken using the internal order by Dgidx)

5) MDEX Engine query with relevance ranking parameter (overrides any relevance ranking strategies configured in the search interface)

Endeca Pipeline : Record Cache

In Endeca, all data sources feeding a join must be record cached. 

There are two exceptions: 
1) Switch join -  It does not do record comparisons and do not require record caches for their data sources
2) Left join - All of the right sources are record caches and hence the left source does not require a record cache

Placing non-indexed characters between search terms - Endeca Record Search

Only a-z, A-Z and 0-9 are indexed for search by default and any special characters will be converted to white space.

Example: 123-456-789 or 123#456#789 will be treated as 123 456 789

How to account for special characters when performing Searches?
Endeca Developer studio has option to add special characters and here is the screenshot on how to configure it.