×

Predictive publishing of internet digital content

  • US RE47,167 E1
  • Filed: 07/23/2015
  • Issued: 12/18/2018
  • Est. Priority Date: 02/28/2008
  • Status: Active Grant
First Claim
Patent Images

1. A computer-implemented method system comprising:

  • at least one digital processor and a memory coupled to the processor, the memory storing instructions executable in the processor, wherein the instructions are configured to cause the processor, in operation, to carry out the steps of accessing a list of individual subscribers to an Internet digital content feed service, the listed subscribers together defining a subscriber audience, and each subscriber on the list having an associated subscriber user profile;

    selecting an article to publish to the subscriber audience based on the subscriber profiles;

    publishing the selected article to the subscriber audience via the Internet digital content feed service;

    maintaining an author profile for at least one author, including an author of the selected article;

    analyzing the selected article to generate corresponding article information, the article information including a plurality of attributes, wherein the attributes include a category and a hardness metric;

    storing the article information in a data store in association with an identifier of a corresponding the author of the selected article and an identifier of the selected article;

    trackingmonitoring individual subscriber actions associated with the selected article in a software reader application that executes on a subscriber'"'"'s client device, to form attention data, wherein tracking includes determining whether a status identifier associated with the selected article indicates that the article is unreadwherein the attention data formed by the reader application comprises indicia of the corresponding subscriber'"'"'s interactions with the selected article in one or more of the following ways, namely, reading the article, flagging the article, tagging the article, emailing the article, clicking through a link in the article, or deleting the article;

    analyzing the attention data to form an indication of each of the one or more individual subscribers'"'"' response responses to the selected article;

    wherein analyzing the attention data includes computing a response variable value as a ratio of a number of positive interactions over a total number of interactions with the selected article over all of the subscriber audience;

    storing the indicia of response variable for each of the one or more individual subscribers'"'"' responses to the selected article;

    andusing the stored article information together with the corresponding indicia of attention data, to predict a response of the one or more of the individual subscribers of the subscriber audience to a new, an unpublished article as followsproviding a software based feed-forward back-propagation neural network;

    inputting to the neural network, for each of the said plurality of attributes, an average of the user response variable values as determined by analyzing the corresponding attention data over plural articles previously published to the subscriber audience; and

    training the neural network using the attention data for all of the subscriber audience over all the articles published by the author, so that the neural network generates an overall predicted response value for the unpublished article for the author.

View all claims
  • 2 Assignments
Timeline View
Assignment View
    ×
    ×