RMS organizes collections of searchable data records, such as vendor addresses, in search indexes that can be accessed independently of each other. Clients like Brainware Intelligent Capture connect to a single search index that contains the data for the field you are analyzing.
A search index is connected to a record source – either a CSV file or a database – and builds itself periodically based on the records retrieved from that source.
Optionally, a CSV file can be provided that contains records that will be excluded from the search. If an ID column is specified for the search index, records that need to be excluded are identified by the ID. Otherwise, the records in the excluded records CSV file need to be identical to the ones in the main record source.
Search indexes consume system resources while they are loaded. Optionally, search indexes can be unloaded until they are required; this can be done either manually or after a period of inactivity. When an index has been unloaded due to inactivity, searching on the index will cause its resources to be reloaded – the search will not fail but the first search may take longer than usual as the index is loaded.
Search indexes can be exported for use on other systems. This allows the search index to be distributed without giving full access to the original records. When these pre-built search indexes are imported, only the scoring mechanism can be configured. For security reasons, imported search indexes cannot be saved or exported, and they cannot be searched for duplicates.
Search index files can be stored with Advanced Encryption Standard (AES) encryption for greater security. A 256 bit key is used if supported, otherwise a 128 bit key is used. The Java Cryptography Extension (JCE) must be installed in the Java Runtime Environment to use 256 bit keys. The JCE can be downloaded from Oracle.