Chapter 6. Term-centric search
This chapter covers
- Examining why field-centric search doesn’t capture naïve expectations
- Exploring cases in which your source data model confuses search users
- Comparing the pros and cons of term-centric methods
- Explaining the tension between term-centric and field-centric methods
- Combining users’ naïve search expectations with smarter capabilities
The previous chapter introduced you to signals and multifield search. Signals measure criteria such as “Is the search an exact title match?” or “Does the search mention a specific actor or director?” These sorts of signals depend on your ability to control querying and construct fields to model users’ intent. We called this process signal modeling. Once fields correspond cleanly to signals, only then can you begin to balance them in a multifield search strategy.
The previous chapters focused heavily on fields as the central unit of relevance. But users don’t think in terms of fields. Users think of their search terms as the central component to search. Users aren’t mired in the details of your database or application. They’ve given you a few brief moments to satisfy them, and they expect you to meet them at their simpler understanding of search. Thus, term-centric search differs from other forms of multifield search by placing the search terms—not the structure of the content—front and center.