# US 10,484,541 B2 · Method and apparatus for improving goal-directed textual conversations between agents and customers > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,484,541 B2 - Document ID: US10484541B2 - Status: Granted - Kind: B2 - Application number: 16/046,045 - Filed: 2018-07-26 - Published: 2018-12-13 - Granted: 2019-11-19 - Earliest claimed: 2014-03-05 - Assignees: [24]7 .Ai, Inc. - Applicants: [24]7 .ai, Inc. - Inventors: Pallipuram V. Kannan; Ravi Vijayaraghavan; R. Mathangi Sri; Suchana Seth - Web page: https://patentgrep.com/patent/US10484541B2 ## CPC - H04M 3/5183: {Call or contact centers with computer-telephony arrangements} - H04L 12/1827: {Network arrangements for conference optimisation or adaptation} - H04L 51/52: for supporting social networking services - H04L 51/04: Real-time or near real-time messaging, e.g. instant messaging [IM] - H04M 3/42365: {Presence services providing information on the willingness to communicate or the ability to communicate in terms of media capability or network connectivity} - H04M 2203/2038: Call context notifications - H04M 2203/357: Autocues for dialog assistance - H04M: TELEPHONIC COMMUNICATION (circuits for controlling other apparatus via a telephone cable and not involving telephone switching apparatus G08) - H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M) ## Abstract In accordance with an example embodiment a computer-implemented method and an apparatus for predicting and tracking of mood changes in textual conversations are provided. The method includes determining, by a processor, one or more mood metrics in each of two or more chat stages of a real-time textual conversation between an agent and a customer. Changes in the one or more mood metrics across the two or more chat stages of the real-time textual conversation are tracked by the processor. Further, the method includes determining, by the processor, at least one action associated with the real-time textual conversation based on the changes in the one or more mood metrics. ## Claims ### Claim 1 A computer-implemented method, comprising: determining, by a processor, one or more mood metrics in each of two or more chat stages of a real-time textual conversation between an agent and a customer, wherein said determining the one or more mood metrics for a chat stage of the real-time textual conversation, by the processor, further comprises determining an overall mood for the chat stage based on a polarity based approach by: assigning polarity labels to features present in the chat stage; assigning polarity strength scores for the polarity labels assigned to the features present in the chat stage; calculating weighted polarity scores for the features based on aggregation of the polarity labels and the polarity strength scores to determine the overall mood for the chat stage; and determining the overall mood, by the processor, based on a subjectivity-based approach by removing terms classified as objective from the real-time textual conversation prior to assigning the polarity labels and the polarity strength scores. ### Claim 2 (depends on 1) The method of claim 1, further comprising: tracking, by the processor, changes in the one or more mood metrics across the two or more chat stages of the real-time textual conversation between the agent and the customer; determining, by the processor, at least one action associated with the real-time textual conversation based on the changes in the one or more mood metrics; and performing, by the processor, the at least one action associated with the real-time textual conversation, wherein performing the at least one action comprises any of: displaying, by the processor, information associated with the at least one action to a supervisor monitoring the real-time textual conversation and providing, by the processor, the information associated with the at least one action to the agent engaged in the real-time textual conversation based on an input received from the supervisor so as to enable the agent to perform the at least one action thereby causing a target outcome of the real-time textual conversation; monitoring an agent engagement score associated with the two or more chat stages of the real-time textual conversation; storing the real-time textual conversation with a timestamp of the real-time textual conversation; and displaying one or more textual recommendations to the agent so as to enable the agent to use the one or more textual recommendations in the real-time textual conversation to thereby cause a target outcome of the real-time textual conversation. ### Claim 3 (depends on 1) The method of claim 1, further comprising: predicting, by the processor, the two or more chat stages from among a plurality of chat stages associated with the real-time textual conversation for determining the one or more mood metrics in each of the two or more chat stages. ### Claim 4 (depends on 1) The method of claim 1, further comprising: displaying in a user interface a visual representation of the one or more mood metrics determined in each of the two or more chat stages for tracking the changes in the one or mood metrics. ### Claim 5 (depends on 4) The method of claim 4, wherein the visual representation of the one or mood metrics comprises representation of textual labels, the textual labels comprising a neutral sentiment, a positive sentiment and a negative sentiment. ### Claim 6 (depends on 4) The method of claim 4, wherein the visual representation of the one or mood metrics comprises one or more color coded representations in each of the two or more chat stages. ### Claim 7 (depends on 1) The method of claim 1, wherein said determining the one or more mood metrics comprises determining a customer engagement score based on at least one of a customer sentiment parameter, a customer response time, and a frequency of use of emoticons in the real-time textual conversation. ### Claim 8 (depends on 1) The method of claim 1, wherein said determining the one or more mood metrics comprises determining an agent engagement score based on at least one of an agent response time, an agent chat concurrency, and a parameter associated with agent's adherence to a recommended response template. ### Claim 9 (depends on 2) The method of claim 2, wherein determination of the at least one action is performed further based on historical information associated with one or more completed textual conversations. ### Claim 10 (depends on 2) The method of claim 2, wherein determination of the at least one action is performed further based on a statistical analysis of one or more completed textual conversations. ### Claim 11 (depends on 1) The method of claim 1, further comprising: accessing, by the processor, a real-time voice based conversation between the agent and the customer; and converting, by the processor, the real-time voice based conversation to the real-time textual conversation. ### Claim 12 (depends on 1) The method of claim 1, wherein said determining the one or more mood metrics for a chat stage further comprises determining an overall mood for the chat stage based on a supervised text classification approach. ### Claim 13 An apparatus, comprising: at least one processor; and a storage module having stored therein machine executable instructions, that when executed by the at least one processor, cause the apparatus to: determine one or more mood metrics in each of two or more chat stages of a real-time textual conversation between an agent and a customer, wherein said determining the one or more mood metrics for a chat stage of the real-time textual conversation, by the processor, further comprises determining an overall mood for the chat stage based on a polarity based approach by: assigning polarity labels to features present in the chat stage; assigning polarity strength scores for the polarity labels assigned to the features present in the chat stage; calculating weighted polarity scores for the features based on aggregation of the polarity labels and the polarity strength scores to determine the overall mood for the chat stage; and determining the overall mood based on a subjectivity-based approach by removing terms classified as objective from the real-time textual conversation prior to assigning the polarity labels and the polarity strength scores. ### Claim 14 (depends on 13) The apparatus of claim 13, said storage module having stored therein machine executable instructions, that when executed by the at least one processor, cause the apparatus to: track changes in the one or more mood metrics across the two or more chat stages of the real-time textual conversation between the agent and the customer; and determine at least one action associated with the real-time textual conversation based on the changes in the one or more mood metrics; and an input-output (I/O) module; wherein the at least one processor is further configured to cause the apparatus to perform the at least one action by enabling the I/O module to any of: displaying one or more textual recommendations to the agent so as to enable the agent to use the one or more textual recommendations in the real-time textual conversation to thereby cause a target outcome of the real-time textual conversation; displaying information associated with the at least one action to a supervisor monitoring the real-time textual conversation and causing the I/O module to provide the information associated with the at least one action to the agent engaged in the real-time textual conversation based on an input received from the supervisor so as to enable the agent to perform the at least one action thereby causing a target outcome of the real-time textual conversation; monitoring an agent engagement score associated with the two or more chat stages of the real-time textual conversation; storing the real-time textual conversation with a timestamp of the real-time textual conversation; and displaying one or more textual recommendations to the agent so as to enable the agent to use the one or more textual recommendations in the real-time textual conversation to thereby cause a target outcome of the real-time textual conversation. ### Claim 15 (depends on 13) The apparatus of claim 13, wherein the at least one processor comprises: a stage prediction module configured to predict the two or more chat stages from among a plurality of chat stages associated with the real-time textual conversation for determining the one or more mood metrics in each of the two or more chat stages. ### Claim 16 (depends on 13) The apparatus of claim 13, further comprising: a user interface module; wherein the at least one processor is further configured to cause the user interface module to display a visual representation of the one or more mood metrics determined in each of the two or more chat stages for tracking the changes in the one or mood metrics. ### Claim 17 (depends on 16) The apparatus of claim 16, wherein the visual representation of the one or mood metrics comprises representation of textual labels, the textual labels comprising a neutral sentiment, a positive sentiment and a negative sentiment. ### Claim 18 (depends on 16) The apparatus of claim 16, wherein the visual representation of the one or mood metrics comprises one or more color coded representations in each of the two or more chat stages. ### Claim 19 (depends on 13) The apparatus of claim 13, wherein the at least one processor comprises a mood tracking module to determine the one or more mood metrics, and wherein the one or more mood metrics comprises a customer engagement score determined based on at least one of a customer sentiment parameter, a customer response time, and a frequency of use of emoticons in the real-time textual conversation. ### Claim 20 (depends on 13) The apparatus of claim 13, wherein the at least one processor comprises a mood tracking module to determine the one or more mood metrics, and wherein the one or more mood metrics comprises an agent engagement score determined based on at least one of an agent response time, an agent chat concurrency, and a parameter associated with agent's adherence to a recommended response template. ### Claim 21 (depends on 14) The apparatus of claim 14, wherein the apparatus is further configured to determine the at least one action based on historical information associated with one or more completed textual conversations. ### Claim 22 (depends on 14) The apparatus of claim 14, wherein the apparatus is further configured to determine the at least one action based on a statistical analysis of one or more completed textual conversations. ### Claim 23 (depends on 13) The apparatus of claim 13, wherein the apparatus is further configured to: access a real-time voice based conversation between the agent and the customer; and convert the real-time voice based conversation to the real-time textual conversation. ### Claim 24 (depends on 13) The apparatus of claim 13, wherein the apparatus is further configured to determine the overall mood based on a subjectivity-based approach by removing terms classified as objective from the real-time textual conversation prior to assigning the polarity labels and the polarity strength scores. ### Claim 25 (depends on 13) The apparatus of claim 13, wherein the apparatus is further configured to determining an overall mood for a chat stage of the real-time textual conversation based on a supervised text classification approach.