Intelligent virtual assistant for communication management and automated response generation
Systems and methods are provided for facilitating inviting a conversation administrator to a message exchange (i.e., a text conversation) between chat participants. The conversation administrator is configured to use natural language processing and machine learning algorithms to provide support to human chat participants by assisting with various tasks or inquiries, resolving disputes, and acting in a mediator or counselor capacity.
1 . A computer-implemented method for providing assistance to multiple simultaneously using conversational-style inputs, the method comprising:
receiving, by one or more processors, conversational-style inputs exchanged within a group conversation hosted by a messaging application, the group conversation including a plurality of users;
receiving, by the one or more processors, an invitation command to add a conversation administrator account to the group conversation, and, in response, joining the conversation administrator account to the group conversation such that the conversation administrator account is presented as a participant in the group conversation;
maintaining and switching context for multiple user sessions by storing, for each of the plurality of users, a user-specific session context associated with the group conversation, and selecting a user-specific session context based on a user identifier associated with an incoming conversational-style input;
analyzing conversational-style inputs received and assessing an emotional state of each of the plurality of users using a natural language processing module; and
generating, based on the assessed emotional state, at least one response for the group conversation, and transmitting the at least one response into the group conversation via the messaging application.
2 . The method of claim 1 , wherein the conversational-style inputs between multiple users comprise at least one conversational-style input sent by a user.
3 . The method of claim 1 , wherein the at least one conversational-style input sent by the user comprises text data.
4 . The method of claim 1 , wherein the natural language processing module comprises one or more deep learning systems applied to the at least one conversational-style input to extract feature representations from the at least one conversational-style input, wherein the feature representations comprise one or more language features.
5 . The method of claim 1 , wherein the natural language processing module comprises one or more machine learning models to assess emotional state of the user based at least on the feature representations and generate a prediction confidence of the emotional state, wherein the one or more machine learning models comprise at least one of an or a natural language processing model.
6 . The method of claim 1 , wherein a descriptive feature is incorporated into a machine learning model of the one or more machine learning models, wherein the descriptive feature comprises a direct measurement that characterizes the at least one conversational-style spoken response.
7 . The method of claim 6 , wherein the direct measurement is a semantic pattern, speech fluency, use of particular words, or speech quality.
8 . The method of claim 5 , wherein the response to the one conversational-style input sent by the user is generated based on the emotional state of the user.
9 . The method of claim 8 , wherein the response to the one conversational-style input sent by the user generated based on the emotional state of the user comprises an empathetic feature.
10 . A system for providing assistance to multiple users simultaneously using conversational-style inputs, the system comprising:
one or more computing processors; and
a non-transitory memory storing instructions;
wherein the instructions, when executed by the one or more processors, implement:
a context management module that maintains and switches context for multiple user sessions by storing, for each of a plurality of users participating in a group conversation, a user-specific session context associated with the group conversation and selecting a user-specific session context based on a user identifier associated with an incoming conversational-style input;
an natural language processing module configured to analyze conversational-style inputs exchanged within the group conversation and assess an emotional state of each user of the plurality of users; and
a response module configured to generate, based on output of the natural language processing module, at least one response, and transmit the at least one response into the group conversation via a messaging application.
11 . The system of claim 10 , wherein the conversational-style inputs between multiple users comprise at least one conversational-style input sent by a user.
12 . The system of claim 10 , wherein the at least one conversational-style input sent by the user comprises text data.
13 . The system of claim 10 , wherein the natural language processing module comprises one or more deep learning systems applied to the at least one conversational-style input to extract feature representations from the at least one conversational-style input, wherein the feature representations comprise one or more language features.
14 . The system of claim 10 , wherein the user NLP natural language processing module comprises one or more machine learning models to assess emotional state of the user based at least on the feature representations and generate a prediction confidence of the emotional state, wherein the one or more machine learning models comprise at least one of an or a natural language processing model.
15 . The system of claim 10 , wherein a descriptive feature is incorporated into a machine learning model of the one or more machine learning models, wherein the descriptive feature comprises a direct measurement that characterizes the at least one conversational-style spoken response.
16 . The system of claim 14 , wherein the direct measurement is a semantic pattern, speech fluency, use of particular words, or speech quality.
17 . The system of claim 16 , wherein the response to the at least one conversational-style input sent by the user is generated based on the emotional state of the user.
18 . The system of claim 17 , wherein the response to the one conversational-style input sent by the user generated based on the emotional state of the user comprises an empathetic feature.