A natural language query is one in which the user does not use any special syntax, but instead enters their search in the form of a statement or question, or as they would normally speak to another person. In general, the best first stab at any query is to write the query as a natural language question, such as:
- what is aspirin?
Or, include all of the connective words to fully express the topic so that they can participate in relevancy boosting. For example, rather than asking a user to conform their query to the query parser with:
- title: return "pink panther"
the user can simply enter the title as it would normally appear:
- the return of the pink panther
The Lucid query parser will automatically reformulate the query so that documents containing the exact wording will rank high, but documents containing subsets of the query will rank reasonably high as well.
A special feature supporting natural language queries is that a trailing question mark ('?', which normally would be treated as a wildcard character) will automatically be stripped from the query if there are at least three terms in the query and the initial term is a question word, "who", "what", "when", "where", "why", or "how". If for some reason you do wish to enter a query that ends with a wildcard question mark but starts with a question word, simply follow it with a space and a period or other punctuation character. For example,
- what is aspirin? . (Treated as a wildcard)
The exact minimum word count is configurable with the minStripQMark configuration setting.
