Scoring quality of traffic to network sites

Abstract

A software and/or hardware facility for scoring the quality of traffic to a site accessible via the Internet or other network. The facility may evaluate traffic based on multiple agent actions in order to detect bogus agent actions. The agent actions may be generic agent actions that are applicable to substantially all network sites. The facility may characterize the quality of an agent's interaction with the network site based on whether one or more of the multiple agent actions occurred during the agent's interaction. The facility may generate non-binary generic and custom scores, classifying traffic as fraudulent if its custom score is substantially higher than its generic score. The facility may characterize the quality of traffic to a network site based on dual algorithms. A first algorithm may be applied to traffic that is statistically significant, while a second algorithm may be applied to traffic that is not statistically significant.

I/We claim: 
     
         1 . A method of characterizing the quality of traffic received by a network site based on multiple agent actions, the method comprising:
 selecting a plurality of actions that an agent may take during an interaction with the network site, wherein each of the plurality of actions may generate value for the network site;   evaluating whether two or more of the plurality of actions occurred during the agent's interaction with the network site; and   characterizing the quality of the agent's interaction with the network site based on whether two or more of the plurality of actions occurred during the agent's interaction with the network site, wherein the quality of the agent's interaction having two or more actions is characterized as more valuable than an agent's interaction having only one action.   
     
     
         2 . The method of  claim 1 , wherein one of the plurality of actions is an action taken by the agent prior to completing a commercial transaction via the network site. 
     
     
         3 . The method of  claim 1 , wherein one of the plurality of actions is an action taken by the agent after completing a commercial transaction via the network site. 
     
     
         4 . The method of  claim 1 , wherein one of the plurality of actions is the agent bookmarking the network site via a browser that does not automatically bookmark each network site visited by the agent. 
     
     
         5 . The method of  claim 1 , wherein one of the plurality of actions is the agent visiting a highly-ranked network site page, not including a network site page initially accessed by the agent. 
     
     
         6 . The method of  claim 1 , wherein one of the plurality of actions is the agent visiting at least one other network site page in addition to a network site page initially accessed by the agent. 
     
     
         7 . The method of  claim 1 , wherein one of the plurality of actions is the agent accessing an HTTPS server associated with the network site. 
     
     
         8 . The method of  claim 1 , wherein one of the plurality of actions is the agent accessing a network site page that contains a particular string of characters. 
     
     
         9 . The method of  claim 1 , wherein one of the plurality of actions is the agent accessing at least two uniform resource locators (URLs) in addition to a network site page initially accessed by the agent. 
     
     
         10 . The method of  claim 1 , wherein one of the plurality of actions is the agent requesting a file in a particular file format from the network site. 
     
     
         11 . The method of  claim 1 , wherein one of the plurality of actions is the agent accessing a file in a particular directory of the network site. 
     
     
         12 . The method of  claim 1 , wherein one of the plurality of actions is the agent accessing an area of the network site accessible only to members of the network site. 
     
     
         13 . The method of  claim 1 , wherein one of the plurality of actions is the agent downloading a file from the network site. 
     
     
         14 . The method of  claim 1 , wherein one of the plurality of actions is the agent performing a POST request via the network site. 
     
     
         15 . The method of  claim 1 , further comprising calculating a generic, non-binary score for the agent's interaction with the network site based on whether one or more of the plurality of actions occurred during the agent's interaction with the network site. 
     
     
         16 . The method of  claim 15 , further comprising calculating a custom, non-binary score for the agent's interaction with the network site based on a custom scoring algorithm associated with the network site. 
     
     
         17 . The method of  claim 16 , further comprising evaluating whether the calculated custom score is substantially higher than the calculated generic score. 
     
     
         18 . The method of  claim 17 , further comprising classifying the agent's interaction with the network site as fraudulent if the custom score is substantially higher than the generic score. 
     
     
         19 . A method of characterizing the quality of traffic received by a network site based on dual algorithms, the method comprising:
 classifying as statistically significant one or more agent interactions with the network site that have been demonstrated to be meaningful indicators of traffic that is valuable to the network site;   classifying as not statistically significant one or more agent interactions with the network site that have not been demonstrated to be meaningful indicators of traffic that is valuable to the network site;   applying a first algorithm to the one or more agent interactions classified as statistically significant;   applying a second algorithm to the one or more agent interactions classified as not statistically significant; and   characterizing the quality of the traffic received by the network site based on the application of the first and second algorithms.   
     
     
         20 . The method of  claim 19 , further comprising:
 calculating a non-binary score for the one or more agent interactions classified as statistically significant, wherein the non-binary score is calculated based on the result generated by applying the first algorithm; and   calculating a non-binary score for the one or more agent interactions classified as not statistically significant, wherein the non-binary score is calculated based on the result generated by applying the second algorithm.   
     
     
         21 . The method of  claim 20 , wherein calculating a non-binary score further comprises normalizing the non-binary score to fall within a standard range. 
     
     
         22 . The method of  claim 19 , wherein the first algorithm comprises:
 selecting a plurality of rules, wherein each rule evaluates one or more parameters associated with an agent interaction with a network site and generates a result based on the value of the one or more parameters;   for each session representing an agent interaction with a network site:
 applying the plurality of rules to data representative of the agent interaction to generate a plurality of results; and 
 characterizing whether the agent interaction represented a desired quality of agent interaction with the network site; 
   evaluating the plurality of results and the characterized quality of agent interaction to determine which combinations of parameters are predictors of the desired quality of agent interaction with the network site; and   generating a non-binary score for one or more sessions based on the combinations of parameters associated with the one or more sessions.   
     
     
         23 . The method of  claim 19 , wherein the second algorithm comprises:
 selecting a plurality of rules that, when applied to data characterizing an agent interaction with the network site, provide an indication of whether the agent interaction is likely to be of value to the network site;   applying each of the plurality of rules to data characterizing an agent interaction with the network site to determine a plurality of results;   multiplying each of the plurality of results by each of a plurality of correlation coefficients to determine a plurality of correlated results, wherein each correlation coefficient reflects a likelihood that the associated rule will be indicative of the value of the agent interaction; and   aggregating the plurality of correlated results to generate a non-binary score.