Enhancing search engines through utilization of visually emphasized terms

We present an approach to weighted indexing of documents in information retrieval systems and search engines utilizing visual emphasizing applied in the document texts. The significance of a term in characterizing the topic of a document depends both on the number of occurrences of the term in the page, and on the amount of visual emphasizing applied in the occurrences. We argue that the document discrimination degree of a term, as measured by the inverse document frequency, should be applied as the default importance of the term in a query. The approach was evaluated using a real world case set showing good performance and sensitivity to parameters as expected.