In-Text Embedded Advertising

Abstract

Computer program products, devices, and methods for generating in-text embedded advertising are described. Embedded advertising is “hidden” or embedded into a message by matching an advertisement to the message and identifying a place in the message to insert the advertisement. For textual messages, statistical analysis of individual sentences is performed to determine where it would be most natural to insert an advertisement. Statistical rules of grammar derived from a language model may be used choose a natural and grammatical place in the sentence for inserting the advertisement. Insertion of the advertisement creates a modified sentence without degrading a meaning of the original sentence, yet also includes the advertisement as a part of a new sentence.

1 . A computer-readable storage medium comprising computer-executable instructions that, when executed by one or more processors, cause a computing device to perform acts comprising:
 extracting a sentence from a text;   associating an advertisement stored in an advertisement database with the sentence, wherein the advertisement comprises an advertising text;   creating a plurality of compositions of the advertisement and the sentence by inserting the advertising text into the sentence;   calculating a probability for each of a plurality of compositions based at least in part on a language model; and   selecting a composition from the plurality of compositions having a highest probability relative to the remaining ones of the plurality of compositions.   
     
     
         2 . The computer-readable storage medium of  claim 1 , wherein the text comprises a webpage having multiple text blocks, and
 extracting the sentence comprises identifying, from the multiple text blocks, a main text block that contains more words than other text blocks and extracting the sentence from the main text block of the webpage.   
     
     
         3 . The computer-readable storage medium of  claim 1 , wherein associating the advertisement stored in the advertisement database with the sentence further comprises:
 identifying a list of similar words for words in the advertising text; and   matching one or more words in the sentence with words in the list of similar words.   
     
     
         4 . The computer-readable storage medium of  claim 1 , wherein associating the advertisement stored in the advertisement database with the sentence further comprises:
 identifying words associated with the advertising text, wherein the words associated with the advertisement comprise at least one of text from the advertisement, a textual description of the advertisement, or a word bid on by an advertiser associated with the advertisement; and   matching one or more words in the sentence with the words associated with the advertising text.   
     
     
         5 . The computer-readable storage medium of  claim 1 , wherein the language model comprises a corpus of text. 
     
     
         6 . The computer-readable storage medium of  claim 5 , wherein calculating the probability for each of the plurality of compositions comprises determining a statistical probability of word orders and relationships between words based at least in part on analysis of the corpus of text. 
     
     
         7 . The computer-readable storage medium of  claim 1 , wherein calculating the probability for each of the plurality of compositions comprises:
 calculating the probability for word strings based at least in part on a frequency of the word strings occurring in the language model;   computing a product of the probabilities of the word strings that comprise each of plurality of compositions; and   taking the product of the probabilities as the probability of each of the plurality of compositions.   
     
     
         8 . The computer-readable storage medium of  claim 1 , wherein the acts further comprise generating a parse tree for the sentence, the parse tree constructed such that each word in the sentence is a node in the parse tree and creating the plurality of compositions comprises inserting the advertising text adjacent to one or more nodes of the parse tree. 
     
     
         9 . The computer-readable storage medium of  claim 8 , wherein generating the parse tree for the sentence comprises:
 identifying a head-word of the sentence;   generating modifiers to a right of the head-word; and   generating modifiers to a left of the head-word.   
     
     
         10 . The computer-readable storage medium of  claim 1 , wherein the acts further comprise rendering the text with the sentence replaced by the composition having the highest probability. 
     
     
         11 . A computing device comprising:
 one or more processors;   memory coupled to the one or more processors;   a sentence extraction module, stored at least in part in the memory and executable by the one or more processors, to extract a sentence from a text;   an advertisement-keyword matching module, stored at least in part in the memory and executable by the one or more processors, to match an advertisement with a keyword in the sentence;   an advertisement-sentence composition module, stored at least in part in the memory and executable by the one or more processors, to insert the advertisement in the sentence; and   a rendering module, stored at least in part in the memory and executable by the one or more processors, to render a modified version of the text that includes the sentence with the advertisement.   
     
     
         12 . The computing device of  claim 11 , wherein the plurality of advertisement-sentence compositions is created at least in part by insertion of the advertisement into a parse tree adjacent to nodes in the parse tree, the parse tree constructed such that each word in the sentence is a node in the parse tree. 
     
     
         13 . The computing device of  claim 11 , further comprising a composition filtering module, stored at least in part in the memory and executable by the one or more processors, to select from a plurality of advertisement-sentence compositions a one of the advertisement-sentence compositions with a highest probability as compared to the other advertisement-sentence compositions, wherein the probability of an advertisement-sentence composition is based at least in part on a probability of word strings in the sentence as determined by a language model. 
     
     
         14 . The computing device of  claim 13 , wherein the language model comprises a corpus of text related to a specific topic. 
     
     
         15 . The computing device of  claim 11 , further comprising a segmentation module, stored at least in part in the memory and executable by the one or more processors, to segment the text into blocks, to identify a main text block, and to extract the sentence from the main text block. 
     
     
         16 . A method implemented at least partially by one or more processors, the method comprising:
 extracting a sentence from a text;   associating, by the one or more processors, keywords in the sentence with an advertisement;   creating compositions comprising the sentence and the advertisement; and   embedding the compositions into the text.   
     
     
         17 . The method of  claim 16 , wherein the advertisement comprises at least one of textual content, a hyperlink, a pop-up window, a still image, a moving image, or a sound. 
     
     
         18 . The method of  claim 16 , wherein associating keywords in the sentence with the advertisement further comprises:
 comparing keywords in the sentence to words associated with the advertisement, wherein the words associated with the advertisement comprise at least one of text from the advertisement, a textual description of the advertisement, or a word bid on by an advertiser associated with the advertisement; and   associating a keyword in the sentence with the advertisement when the keyword matches at least one of the words associated with the advertisement.   
     
     
         19 . The method of  claim 16 , wherein creating compositions comprising the sentence and the advertisement further comprises:
 creating a parse tree for the sentence constructed such that nodes in the parse tree represent words in the sentence;   creating a plurality of modified parse trees by inserting the advertisement adjacent to nodes in the parse tree; and   selecting one of the plurality of modified parse trees based at least in part on a probability of a word order found in the modified parse tree.   
     
     
         20 . The method of  claim 19 , wherein the probability of the word order is determined at least in part by reference to a language model.