This is a follow up on the Beauty of XOR and XNOR searches post, describing possible applications of these search modes to Information Retrieval, Search Marketing, and Web Mining. The post is a snippet taken from

Enjoy it.

An IR researcher can test the performance of an LSI algorithm with a sample of documents retrieved through XOR and XNOR searches. Said sample should be rich in co-occurrence cases. Using a similar procedure, search marketers or Web intelligence specialists can identify sets of documents that emphasize keywords somehow related through different co-occurrence paths.

An interesting application consists in extracting all the unique terms (or just the high frequency ones) from a text source and constructing an XOR query with these. We may refer to this as XORing a text source. This should help one identify a network of co-occurrence paths over a collection and which documents might be relevant to specific combination of terms from the original source.

The text source can be a title, description, abstract, or paragraph of a document, or even an entire document. However, XORing a large document might be computer-intensive.

A similar exercise can be done by XNORing a text source. In both cases, the resultant output can be used to identify prospective competitors; i.e., documents relevant to similar concepts or belonging to companies within the same business space.

We are currently testing the XOR and XNOR search modes as a query disambiguation strategy.

PS. Today, 1-9-2014, we added new material that discusses these search modes for disambiguation and clustering. 🙂