Granted Patent B1

US 11,005,843 · App. 16/668,920 · Granted May 11, 2021

Inventors: Yalin Evren Sagduyu (Gaithersburg, MD); Ziad El-Jamous (Rockville, MD); Min Ding (Poolesville, MD); Vikram Manikonda (Bethesda, MD); Yi Shi (Germantown, MD)

Assignee: Intelligent Automation, Inc.

H04L63/0876G06Q50/01H04L9/3247H04L51/046H04L67/306

Abstract

Claims (47)

1. A computer implemented system of computing scores for features based on the pattern of historical social media usage and profile information, comprising:

a computer having a processor and data storage capabilities including a non-transitory medium having computer executable instructions stored thereon, the computer receiving social media data;

software means to obtain social media data sets with features representing historical user profile with a time of running loot detection;

software means to assign a score to each feature;

software means to compute a user account age;

software means to assign the feature score inversely proportional to the user account age;

software means to compute a number of times the username or user profile includes any bot-characteristic word; and

software means to assign the feature score directly proportional to a non-decreasing function of the number of times the username or user profile includes any bot-characteristic word.

2. The system of claim 1 , wherein the assignment of scores to features of a user includes:

software means to compute a life post rate as the number of posts by the user normalized by a window size starting from the time of user account creation; and

software means to assign the feature score directly proportional to the life post rate.

3. The system of claim 1 , wherein the assignment of scores to features of a user includes:

software means to compute an asymmetry of followers and friends counts, as a followers/friends ratio; and

software means to assign the feature score directly proportional to the followers/friends ratio.

4. The system of claim 1 , wherein the assignment of scores to features of a user includes:

software means to compute a length of username; and

software means to assign the feature score directly proportional to the length of username.

5. The system of claim 1 , wherein the assignment of scores to features of a user includes:

software means to compute a count of numbers in a username normalized by the maximum count of characters allowed in the username; and

software means to assign the feature score directly proportional to the count of numbers in the username.

6. A computer implemented system of computing scores for features based on the pattern of historical social media usage and profile information, comprising:

a computer receiving social media data;

software means to obtain social media data sets with features representing historical user profile until a time of running bot detection;

software means to assign a score to each feature; software means to compute a return of indicator function when the user uses API or auto-piloted programs for social-media posts; and

software means to assign the feature score directly proportional to the return of indicator function when user uses API or auto-piloted programs for social-media posts.

7. A computer implemented system of computing scores for features based on the pattern of historical social media usage and profile information, comprising:

a computer having a processor and data storage capabilities including a non-transitory medium having computer executable instructions stored thereon; the computer receiving social media data;

software means to obtain social media data sets with features representing historical user profile until a time of running bot detection;

software means to assign a score to each feature;

software means to compute a change in locations of the user;

software means to assign the feature score inversely proportional to the change in locations of the user;

software means to compute a number of times the social media posts of a user include are a bot-characteristic word; and

software rears to assign the feature score directly proportional to a non-decreasing function of the number of times the social media posts of a user include any bot-characteristic word.

8. A computer implemented system of detecting hots in a social media data set, comprising:

a computer having a processor and data storage capabilities including a non-transitory medium having computer executable instructions stored thereon, the computer receiving social media data;

software means to compute scores for features of each user in the social media data set based on social media posts and interactions and assigning a score to each feature;

software means to compute scores for features of each user in the social media data set based on the historical pattern of social media use and profile information, and assigning a score to each feature;

software means to aggregate the scores for different bot features of each user;

software means to assign the hot score to each user in the social media data set as the likelihood that a user is a hot or a human;

software means to map each user's social media interactions to bot features, and assign scores to features as relative indicators of bot behavior specific to a given social media data set;

software means to map each user's historical pattern of social media use and profile information to bot features as absolute indicators of bot behavior over the account lifetime;

software means to compute the score of a feature separately per user; and

software means to compute the score of a feature by normalizing it relative to other users' social media interactions, historical patterns and profiles.

9. The system of claim 8 , wherein assignment of bot scores to users includes:

software means to check verified users and users in white and black lists;

software means to combine weighted sum of scores of features to the bot score assigned to each user; and

software means to train a classifier to classify each user to a bot or human with the confidence of classifier assigned as the bot score to each user.

Assignments (11)

Cited By (2)

US 12,306,881 US 12,541,560