When it comes to information retrieval, traditional keyword search has been troubled by two fundamental “translation” problems that have proven challenging to fully address:
- First, users must translate their complex market research questions into a handful of keywords.
- Second, the search engine must translate these keywords into meaningful matches against the vast sea of indexed content.
In both cases, these tasks attempt to find content with the proper, most relevant context to answer the user’s original question: the user by choosing keywords that they believe reflect how others talk about a subject, and the search engine by employing strategies such as stemming and synonym expansion.
But while users can be trained to do better keyword selection, and search engines have become better at intuiting context from keywords and connecting that context to relevant content, both still miss the mark more often than we would prefer. It’s a problem that has …