This series of articles will examine FAST Navigators and how they can be utilized in your application. Navigators are very much like filters. Users are able to drill down into the result set by clicking one or more navigators. In my opinion, the future of search relies on being able to serve relevant results as well as allow the user to drill down into the result set with the help of navigators/filters. Today, users tend to type in a keyword expecting that the results will be relevant to what they’re searching for. Being able to return the relevant results is great but what if you have a very large index? Returning relevant results may not be enough if the result the user wants is on page 5 or 10. What if you could also allow the user to drill further into the result set to get even more relevant results? Navigators could be the way to make that vision a reality. Simple example when searching for products would be an ability to select a price range, color or even city that the item is being sold at. More complex scenarios could utilize a feature called Result Clustering. I will discuss Result Clustering as well as other navigation features in future posts. Navigators can become costly when it comes to performance and with that in mind FAST has two main groups of navigators: Deep Navigators and Shallow Navigators. Deep Navigators are generated from the entire results set, whereas Shallow Navigators are generated from the top N (default 100) documents returned by the query. There are three basic navigator types in FAST:
- Numeric Navigator – applies to numerical data types (int, double, datetime). This navigator can be easily configured based to break the results into groups (aka buckets) or it can be set to a fixed value.
- String Navigator – is a navigator that is based on the values in a specified field
- Word Navigator – is similar to String navigator except that the result set is generated from all fields (including composite fields) in the result.
In part 2 of this article I’ll drill into the navigator types in more detail as well as examine scope navigators and results clustering…stay tuned!