|
Start thinking more about information extraction (presenting information in attribute:value pairs, using intelligent labels, etc.), and less about tf/idf (though learn the basics about how search engines work).
Start paying more attention to how recommendation systems work, and less about a broad relevancy (java can be an island in Indonesia, a cup of coffee, or a programming language - with no two serps alike - depending upon the searcher).
While search engines are still concerned about keyword matching, the movement is towards fulfilling information needs, and providing individualized recommendations.
|