Improving the Relevance of Search Results
Web Search utilizes a sophisticated relevance-ranking algorithm. During a search, Web Search considers
- The number of times words appear in a document
- The proximity of words in a multiple word search (the closer the words appear, the more relevant the document will be)
- The order of words in a multiple word search (the exact order of words is more relevant)
- The location of words in a document (specifically words that appear in a META tag, title, body, header, footer, etc.)
- The formatting of words in a document (such as bold, font type and size, etc.)
- Query weighting in a multiple query scenario
- The number of times words occur within an entire index (for example, the word the has low relevance)
To illustrate how these criteria work, consider the following examples:
- Words in bold face are more relevant than regular words.
- Words contained in the <Title> tag are more relevant than words contained within the <body> tag.
- Words contained in the Keywords and Description META tags are more relevant than content words.
- Words contained within the <A HREF=> tag used for creating links are less relevant than words outside of this tag.
- A document containing a specified search term multiple times is more relevant than a document that contains the search term only once.
- A word within a 36-point body text is more relevant than within 4-point footer text.
- Documents returned from a query that is weighted at 100% is more relevant than a 50% weighted query. This is normally used in multiquery searches where each query has a specified weight. For example:
query0=netware&weight0=100&query1=groupwise&weight1=100
