Inferring Searcher Intent

To dive into the nuts and bolts a bit more, Eugene Agichtein conducted an in-depth study in 2010 on developing computational models of searcher behavior, interests, and actions that could potentially indicate what the searcher is hoping to accomplish.

In the presentation of his study, Eugene looks at four core challenges for web search:

  1. Query Interpretation, to infer intent
  2. Ranking, which has high dimensionality
  3. Evaluation, for system improvement
  4. Result Presentation, the visualization of the information

He uses this approach to formulate the information retrieval process:

moving from query formulation > to search > to source selections > to examination, and finally delivery, with a loop often occurring between source selection and query formulation.

He then compiles this process into an "information seeking funnel" that pretty much matches up to the classic AIDA marketing formula for conversion-attention, interest, desire, action-but instead using the terms wandering, exploring, seeking, and asking.

It is important to be cognizant of this behavioral pattern, as it provides you the opportunity to align your content with the appropriate keywords in each stage of the conversion funnel - target keywords that show interest toward the top of the funnel, when you still need to sell the visitor.

Where as commercial investigation and transactional queries should be used to deliver content that is at the bottom of the funnel.