To enhance performance, multiple online topic hierarchies can be used for classification. Moreover, the various techniques used for classifying queries into topics can be combined to increase the level of confidence.
The favored and non-favored sources field 430 may include information that identifies sets of web pages/sites that are either "favored sources" (i.e., identified sources of useful or authoritative content on the desired subject) or "non-favored sources" (i.e., identified as sources of misinformation or over-promotion on that subject) for a particular query theme. For example, for the query theme "sites that provide free downloads," web sites that actually provide free software downloads would be considered "favored sources" and web sites that mislead search engines with words such as "free" and "download" (popularly known as "spam techniques"), but do not in fact provide access to free downloads, would be considered "non-favored sources."
Classifying web sites as "favored" may be based on host names. For example, the web site of the World Wildlife Fund is hosted by www.wwf.org. This web site would be a favored source for queries dealing with wildlife or animals. A host may contain more than one web site. Since parts of a web site may be relevant while other parts are not, the relevant parts can be denoted by a set of URL prefixes (e.g., www.geocities.com/A/B/C).
In an implementation consistent with the present invention, the set of favored and non-favored sources may be automatically determined. To accomplish this, exemplary queries in the query theme may be classified into a set of topics (e.g., an online topic hierarchy, such as Yahoo!, Open Directory, or Google) using the approach for classification described above. Web hosts that appear in the URLs associated with the best matching topics to the query theme may be taken to be favored sources. For example, if the query theme is "sites that help in finding accommodation," then web hosts listed under the Open Directory category "http://dmoz.org/Recreation/Travel/Lodging" can be taken as favored sources.
The editorial opinion parameter field 440 may include parameters that quantify the editorial opinion for specific favored and non-favored sources for search queries that match specific query themes. As will be described in more detail below, the editorial opinion parameter may be used to modify the placement of applicable web pages in the ranking of search results.