System and method for recommending keyword based on indirect selection

A method for recommending a keyword, the method comprising:

• determining whether the keyword is a direct click keyword by determining, referring to a click log, whether the keyword is associated with a first universal resource locator (URL) of an advertisement of an advertiser;
• determining whether the keyword is an indirect click keyword by determining that the keyword does not locate the first URL but locates a second URL having an attribute similar to the first URL;
• recommending the determined indirect click keyword;
• and
• computing selection probabilities that the first URL is selected using the direct click keyword and the indirect click keyword,
• wherein the recommending selects at least one indirect click keyword having a corresponding one of the selection probabilities greater than zero from among the plurality of indirect click keywords and then provides the selected at least one indirect click keyword to a user, and
• the computing includes,
• computing the selection probabilities based on selection rates of the first URL and the second URL with respect to the direct click keyword and the indirect click keyword, and a similarity probability between the first URL and the second URL, and
• updating the similarity probability based on the selection probabilities.

A method for recommending a keyword, the method comprising:

• determining whether a keyword is a direct click keyword by determining, referring to a click log, whether the keyword is associated with a first universal resource locator (URL) of an advertisement of an advertiser;
• determining whether the keyword is an indirect click keyword by determining that the keyword does not locate the first URL but locates a second URL having an attribute similar to the first URL;
• computing selection probabilities that the direct and indirect keywords are used for selecting the first URL or the second URL with the computing device based on selection rates of the first URL and the second URL with respect to the direct click keyword and the indirect click keyword;
• computing a similarity probability between the first URL and a second URL with the computing device;
• updating the similarity probability based on the selection probabilities;
• and
• selecting and recommending a target indirect keyword having a corresponding one of the selection probabilities greater than zero from a plurality of indirect click keywords.