Methods and systems of supervised learning of semantic relatedness

Abstract

A method of evaluating a semantic relatedness of terms. The method comprises providing a plurality of text segments, calculating, using a processor, a plurality of weights each for another of the plurality of text segments, calculating a prevalence of a co-appearance of each of a plurality of pairs of terms in the plurality of text segments, and evaluating a semantic relatedness between members of each the pair according to a combination of a respective the prevalence and a weight of each of the plurality of text segments wherein a co-appearance of the pair occurs.

What is claimed is: 
     
         1 . A computerized method of evaluating semantic relatedness of terms, comprising:
 obtaining a plurality of text segments extracted from a plurality of documents associated with at least one user;   calculating, using a processor, a plurality of weights, each one of said plurality of weights is calculated for a text segment of said plurality of text segments based on an analysis of the behavior of said at least one user with reference to said each text segment;   calculating a prevalence of a co-appearance of each of a plurality of pairs of terms in said plurality of text segments;   evaluating a semantic relatedness for determining the strength of the semantic relatedness between the terms of each said pair according to a combination of: 1) a prevalence of the said pair in said plurality of text segments and 2) the weight of each text segment of said plurality of text segments in which a co-appearance of said pair occurs; and generating a semantic relatedness dataset mapping said semantic relatedness between at least some terms of said plurality of pairs of terms, said dataset is subject to said at least one user.   
     
     
         2 . The method of  claim 1 , further comprising generating a semantic relatedness dataset mapping said semantic relatedness between members of each said pair. 
     
     
         3 . The method of  claim 1 , further comprising using said semantic relatedness for minimizing an error in said plurality of weights. 
     
     
         4 . The method of  claim 1 , further comprising using said semantic relatedness for maximizing a reward in said plurality of weights. 
     
     
         5 . The method of  claim 1 , wherein each said text segment is a member of a group consisting of a sentence, a paragraph, a set of paragraphs, an email, an article, a webpage, an instant messaging (IM) content, a post in a social network, a tweet, a website, and a file containing text. 
     
     
         6 . The method of  claim 1 , wherein said plurality of text segments associated with at least one user; wherein said semantic relatedness is subjective to said at least one targeted user. 
     
     
         7 . The method of  claim 1 , wherein said plurality of text segments are associated with at least one field of interest. 
     
     
         8 . The method of  claim 1 , wherein said plurality of text segments are extracted from a plurality of webpages visited by said at least one user. 
     
     
         9 . The method of  claim 1 , wherein said plurality of text segments are authored by said at least one targeted user. 
     
     
         10 . The method of  claim 1 , wherein said calculating a plurality of weights comprises monitoring a plurality of network documents associated with said at least one user and calculating said plurality of weights accordingly. 
     
     
         11 . The method of  claim 1 , wherein said plurality of text segments are extracted from a plurality of documents stored in storage allocated to said at least one targeted user. 
     
     
         12 . The method of  claim 1 , wherein said evaluating comprises determining at least one characteristic of said at least one user according to an analysis of said semantic relatedness dataset. 
     
     
         13 . The method of  claim 12 , wherein said plurality of text segments comprises a member of a group consisting of: an email send by said user, an email sent to said at least one user, a webpage viewed by said at least one user, a document retrieved in response to a search query submitted by said at least one user, a file stored on a client terminal associated with said user, and a file stored in a storage location associated with said at least one user. 
     
     
         14 . The method of  claim 13 , wherein said storage location is a member of a group consisting of: a client terminal, a virtual storage location, an email server, a web server, and a search engine record. 
     
     
         15 . The method of  claim 1 , wherein said calculating a plurality of weights comprises calculating said plurality of weights according to a input provided by said user for at least some of said at least one plurality of text segments. 
     
     
         16 . The method of  claim 1 , wherein said calculating a plurality of weights comprises calculating said plurality of weights according to a match with a search history of said at least one user. 
     
     
         17 . The method of  claim 1 , wherein said calculating a plurality of weights comprises calculating each of said plurality of weights according to an origin of a respective said text segment. 
     
     
         18 . The method of  claim 1 , wherein said calculating a plurality of weights is calculated according to an active learning algorithm which analyzes each text segment of said plurality of text segments. 
     
     
         19 . A computerized method of evaluating a semantic relatedness of terms, comprising:
 identifying a plurality of text segments extracted from a plurality of documents associated with at least one targeted user;   calculating, using a processor, a plurality of weights, each one of said plurality of weights is calculated for a text segments of said plurality of text segments based on an analysis of the behavior of said plurality of text segments;   calculating a prevalence of a co-appearance of each of a plurality of pairs of a plurality of terms in said plurality of text segments;   evaluating a semantic relatedness between said terms of each said pair according to said prevalence, and the weights of said text segments in which a co-appearance of each of said pairs occurs,   generating a semantic relatedness between said terms of each said pair according to said prevalence, and the weights of said text segments in which a co-appearance of each of said pair occurs,   generating a semantic relatedness dataset mapping said semantic relatedness between at least some of said plurality of terms, wherein said semantic relatedness dataset is subjective to said at least one target user; and   using said semantic relatedness dataset in conjunction with inputs of said at least one user for at least one of aggregating personalized content, searching for content, and providing services to said at least one targeted user.   
     
     
         20 . A system of evaluating a semantic relatedness of terms, comprising:
 a processor;   an input interface which receives a plurality of text segments extracted from a plurality of documents associated with at least one user;   a weighting module calculating a plurality of weights, each one of said plurality of weights is calculated for a text segment of said plurality of text segments based on an analysis of the behavior of said at least one user with reference to said each text segment; and   a dataset generation module which, using said processor, A) calculates a prevalence of a co-appearance of each of a plurality of pairs of terms in said plurality of text segments, B) evaluates a semantic relatedness between of each said pair according to a combination of: 1) a prevalence of the said plurality of text segments, and 2) a weight of each text segment of said plurality of text segments in which a co-appearance of said pair occurs, said semantic relatedness subjective to said at least one user and, C) generates a semantic relatedness dataset mapping said semantic relatedness between said terms of each said pair.