Recommended strategies

This section provides some recommended strategies that depend on the implementation type.

Relevance ranking behavior is complex and powerful and requires careful, iterative development. Typically, selection of the ideal relevance ranking strategy for a given application depends on extensive experimentation during application development. The set of possible result ranking strategies is extremely rich, and because setting ranking strategies is highly dependent on the quantity and type of data you are working with, a strategy that works well in one situation could be unsatisfactory in another.

For this reason, this documentation provides recommended strategies for different types of implementations and suggests that you use them as a point of departure in creating your own strategies. The following sections describe recommended general strategies for each product in detail.

Testing your strategies

When testing your own strategies, it is a good idea to try searching on diverse examples: single word terms, multi-word terms that you know are an exact match for records in your data, and multi-word terms that contain additional words to the ones in your data. In this way you will see the full range of relevance ranking effects.