Search Physics Inc. (SP) has developed an automatic and efficient engine that extracts the information content of websites and stores additional information about each site in a database that can be used for several other useful purposes. Among them, one is to classify the links, and the other is to assist the users in narrowing down their search results.

In many cases, users experience that their searches return too many URL's irrelevant to the search topic. Furthermore, it is also difficult to judge relevance from the URL list, so each must be visited individually -- a time consuming task.

The SP iterative process not only reduces the number of irrelevant URL's in the returned list but also provides a brief description that makes it easy and quick to judge the relevance of the site.

The SP process has two additional features; it allows the use of filters that block click fraud, and provides a more efficient and directly targeted advertisement. The keywords supplied by the SP database quickly narrow down the search, and at the same time, discriminate against click fraud through the use of filters.

The information extraction capability clearly differentiates the SP technology from other approaches that restrict themselves to pure classification; among these are Support Vector Machines and Neural Networks. They can be used quite successfully for a variety of classification and prediction but they cannot extract and store the content of the links in a database.