Natural language processing artificial intelligence network and data security system

Patent metadata

Field	Value
Publication number	US-2019109878-A1
Application number	US-201715726137-A
Country	US
Kind code	A1
Filing date	Oct 5, 2017
Priority date	Oct 5, 2017
Publication date	Apr 11, 2019
Grant date	—

Abstract

According to an embodiment, a natural language processing artificial intelligence network and data security system determines an emotions model for one or more users from electronic natural language interactions of the users. The system includes a natural language processing decoder to determine textual features from the electronic natural language interactions that may be indicative of emotional states of the users. They system includes an emotions model encoder that generates an emotions model based on the emotional states of the users in the electronic natural language interactions retrieved from the data storage. The system also includes an artificial intelligence network and data security subsystem. The artificial intelligence network and data security subsystem may use the emotions model as a primitive for artificial intelligence based tasks including computer system security, network security, data security, proactive monitoring and preventive actions, that are moderated using the context provided by the emotional state of a user.

First claim

1 . A natural language processing artificial intelligence network and data security system comprising:
a processor; a memory to store machine readable instructions; wherein the processor is to execute the machine readable instructions to:
detect interactions of a user in a dataset using natural language processing, wherein the dataset comprises a plurality of interactions between a plurality of users on a network;
determine a plurality of textual features expressed in natural language in each of the interactions;
determine a plurality of emotional features for each of the interactions based on the plurality of textual features and an emotions index, wherein the emotions index associates textual features with emotional features;
determine an emotional state for each of the interactions based on an aggregation of the plurality of emotional features;
determine a chronological sequence of the emotional states for the user based on a time of each of the interactions and the determined emotional state for each of the interactions;
determine a probability of transitioning between different emotional states for the user based on the chronological sequence of emotional states for the user;
generate an emotions model as a primitive for artificial intelligence based on the probability of transitioning between different emotional states for the user;
determine an emotional stability value of the user based on the emotions model;
determine using an artificial intelligence subsystem whether the emotional stability value of the user is within a threshold, wherein the threshold is based on the emotional stability value of other uses in the dataset; and
in response to the determination that the emotional stability value is within the threshold the artificial intelligence and data security subsystem regulates network permissions of the user.

2 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein to determine the emotional stability value of the user the system is to:
determine transitions in the emotional state of the user in the chronological sequence of emotional states; determine a count of interactions in the chronological sequence of emotional states between each of the transitions of emotional states; and generate the emotional stability value of the user based on the transitions and the count of interactions between each of the transitions.

3 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein to determine the emotional stability value of the user the system is to:
generate the emotional stability value based on the entropy rate of the emotions model, wherein the entropy rate of the emotions model is based on the time density of the emotional states in the chronological sequence of emotional states for the user.

4 . The natural language processing natural language processing artificial intelligence network and data security system of claim 1 , wherein to determine the emotional state for each of the interactions the system is to:
determine type counts of each of the emotional features in each of the interactions based on the determined plurality of emotional features, wherein the type count is a count of each of the different types of emotional features in each of the interactions; determine a dominant emotional feature in each of the interactions based on the highest type count in each of the interactions; and detect the emotional state for each of the interactions based on the dominant emotional feature.

5 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein the emotions index comprises a types classification of emotional features and a plurality of textual features that maps to each of the types classification.

6 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein the emotions index comprises:
a types classification of emotional features based on cognitive processes; and a plurality of textual features that maps to each of the types classification.

7 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein the emotions index comprises:
a types classification of emotional features; a hedonicity value associated with each of the emotional features; and a plurality of textual features that maps to each of the types classification.

8 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein the emotions index comprises:
a types classification of emotional features, wherein the emotional features in the index are balanced using pairs of emotional features representing opposite emotions; and a plurality of textual features that maps to each of the types classification.

9 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein each of the interactions detected for the user comprises a plurality of textual features expressed by the user per unit time.

10 . The natural language processing artificial intelligence network and data security system of claim 1 , wherein the processor is to execute the machine readable instructions to:
determine emotional stability values for other users in the dataset based on interactions for other users in the dataset; and determine a threshold for the emotional stability value based on the emotional stability values for other users in the dataset.

11 . A natural language processing artificial intelligence network and data security system to regulate access to data for a user, the system comprising:
a data storage device to store a set of data, wherein the set of data comprises a plurality of interactions between a plurality of users; at least one processor to:
detect interactions of users in a dataset, wherein the interactions are in natural language;
determine a using natural language processing a plurality of textual features in each of the interactions for each of the users;
determine a plurality of emotional features for each of the interactions based on the textual features and an emotions index, wherein the emotions index associates textual features with emotional features;
determine an emotional state for each of the interactions based on an aggregation of the plurality of emotional features;
determine a chronological sequence of emotional states for each of the users based on a time for each of the interactions and the determined emotional state for each of the interactions;
determine a win-loss table of interactions between sub-sets of two users during the same period of time for each of the two users, based on the determined chronological sequence of emotional states for each of the users;
determine a relative ranking of interactions between sets of two of the users based on the plurality of win-loss tables of interactions;
determine whether the relative ranking of a user in the users exceeds a ranking threshold, wherein the ranking threshold is based on the deviation from the mean ranking of users in the dataset; and
based on the determination that the relative ranking of a user in the users exceeds the ranking threshold, the artificial intelligence network and data security subsystem may modify access to data on a data storage device for the user.

12 . The natural language processing artificial intelligence network and data security system of claim 11 , wherein to determine the emotional state for each of the interactions the system is to:
determine type counts of each of the emotional fea

Assignees

• Accenture Global Solutions Ltd
• Misram Llc

Inventors

• Boyadjiev Constantine T
• Chandramouli Rajarathnam
• Shao Zongru
• Subbalakshmi Koduvayur

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06F40/30

  Semantic analysis · CPC title

• G06F7/023

  adaptive, e.g. self learning · CPC title

• G06N5/022

  Knowledge engineering; Knowledge acquisition · CPC title