Semantic Signatures^® are a new way of representing and analyzing semantic information (meaning) in text. Semantic Signatures, produced by TextWise’s Trainable Semantic Vectors (TSV) technology, provide a rich semantic representation of the multiple concepts and topics contained in a body of text. Semantic Signatures can be constructed for a wide range of texts including individual words, phrases, word lists (e.g. metadata), short passages (such as text advertisements or image labels), web pages, or full text documents (e.g. technical articles).

A Semantic Signature represents the concepts in a text through a weighted vector entry of typically several thousand semantic dimensions. The weight of each vector entry represents the strength of the text along that particular dimension as shown in the following example:

 
Semantic dimensions are derived using a one-time training process with an existing classification scheme and associated documents for a domain such as the ODP or the Wikipedia schemas for the web. Semantic dimensions may be labeled with the dimension (category) names taken from the schema.
 
Unlike other techniques that require manual construction and maintenance of dictionaries, thesauri, or ontologies, TSV technology automatically generates a semantic dictionary during the training phase.   This dictionary then contains the vocabulary known to be relevant to the application domain and automatically provides a “definition” of each term using semantic dimensions.
 
A Semantic Signature for a text can be constructed rapidly through a mathematical combination of the semantic vectors of the vocabulary contained in that text. Having computed a Semantic Signature for two given texts, a rapidly computed match score indicates the relevance of the match based on the positions of the two texts in the m-dimensional semantic space.

For example (using our demo on the homepage), entering the query:

"In a move that provides relief to thousands of renters who face eviction but draws the federal government even deeper into the housing market, the loan giant Fannie Mae said Sunday that it would sign new leases with renters living in foreclosed properties owned by the company."

The following matched Semantic Signatures and top related news article are displayed:

 
The TextWise SemanticHacker API provides a signature service call and match call that analyzes the text or Web page provided in the call and returns a Semantic Signature and a match, respectively. Therefore, developers can perform their own matching like the example above using proprietary Web or other text against our existing content index.
 
For more detailed information on Semantic Signatures, please request a copy of our whitepaper.