# US 2019/0377790 A1 · Supporting Combinations of Intents in a Conversation > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2019/0377790 A1 - Document ID: US20190377790A1 - Status: Published application - Kind: A1 - Application number: 16/001,777 - Filed: 2018-06-06 - Published: 2019-12-12 - Earliest claimed: 2018-06-06 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Robert N. Redmond; Florian Pinel; Donna K. Byron; Will Spann - Web page: https://patentgrep.com/patent/US20190377790A1 ## CPC - G06F 40/30: Semantic analysis - G06F 40/35: Discourse or dialogue representation - G06F 40/205: Parsing - G06F 40/295: Named entity recognition - G06F 40/56: Natural language generation - H04L 51/02: using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages - G10L 13/00: Speech synthesis; Text to speech systems - G10L 15/02: Feature extraction for speech recognition; Selection of recognition unit - G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue - G10L 2015/223: {Execution procedure of a spoken command} - H04L 51/04: Real-time or near real-time messaging, e.g. instant messaging [IM] - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M) - G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING ## Abstract A computer-implemented method for supporting combinations of intents in a conversation service. In one embodiment, the computer-implemented method provides a conversation service using a set of conversation trees. Each conversation tree within the set of conversation trees comprises a list of intents that define how to answer to a certain type of user input. The computer-implemented method creates an intent graph using the list of intents of each conversation tree within the set of conversation trees. The computer-implemented method adds composite intents to the set of conversation trees using the intent graph to generate a set of updated conversation trees. The computer-implemented method configures the conversation service to use the set of updated conversation trees that comprises composite intents for supporting combinations of intents. ## Claims ### Claim 1 A computer-implemented method for supporting combinations of intents, the computer-implemented method comprising: providing a conversation service using a set of conversation trees, each conversation tree within the set of conversation trees comprises a list of intents that define how to answer to a certain type of user input; creating an intent graph using the list of intents of each conversation tree within the set of conversation trees; adding composite intents to the set of conversation trees using the intent graph to generate a set of updated conversation trees; and configuring the conversation service to use the set of updated conversation trees that comprises composite intents. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein creating the intent graph using the list of intents of each conversation tree within the set of conversation trees comprises: determining an entity input and entity outputs for each intent in the list of intents for each conversation tree within the set of conversation trees; and assembling the entity input and the entity outputs for each intent in the list of intents for each conversation tree within the set of conversation trees to create the intent graph. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein adding composite intents to the set of conversation trees comprises generating a sample utterance for a new composite intent. ### Claim 4 (depends on 3) The computer-implemented method of claim 3, wherein generating the sample utterance for the new composite intent comprises: removing output entities from a first sample utterance for a first intent node of the new composite intent to produce a first portion of the sample utterance for the new composite intent; removing input entities from a second sample utterance for a second intent node of the new composite intent to produce a second portion of the sample utterance for the new composite intent; and combining the first portion of the sample utterance for the new composite intent and second portion of the sample utterance for the new composite intent to generate the sample utterance for the new composite intent. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, wherein adding composite intents to the set of conversation trees comprises: assessing a user interest in a new composite intent prior to adding the new composite intent to the set of conversation trees by determining whether the user interest in the new composite intent exceeds a predetermined threshold; and adding the new composite intent to the set of conversation trees in response to a determination that the user interest in the new composite intent exceeds the predetermined threshold. ### Claim 6 (depends on 5) The computer-implemented method of claim 5, wherein assessing the user interest in the new composite intent comprises: parsing a conversation log; and determining a number of user inputs that would be routed to the new composite intent. ### Claim 7 (depends on 5) The computer-implemented method of claim 5, wherein assessing the user interest in the new composite intent comprises: parsing a conversation log; and determining how often a first intent and a second intent that constitute the new composite intent are called in sequence within a predetermined period. ### Claim 8 (depends on 1) The computer-implemented method of claim 1, wherein adding composite intents to the set of conversation trees comprises adding a new intent node to at least one of the conversation trees in the set of conversation trees that includes an original intent that forms part of a new composite intent. ### Claim 9 (depends on 1) The computer-implemented method of claim 1, wherein adding composite intents to the set of conversation trees comprises adding the composite intents to a new conversation tree dedicated to composite intents. ### Claim 10 A computer-implemented method for dynamically supporting combinations of intents in a conversation service, the computer-implemented method comprising: receiving a user input from a user via a user device; determining input entities and output entities of the user input; determining a path in an intent graph that connects the input entities to the output entities; determining a response to the user input using intent nodes in the path in the intent graph that connects the input entities to the output entities; and providing the response to the user via the user device. ### Claim 11 (depends on 10) The computer-implemented method of claim 10, further comprising collecting feedback regarding the response to the user input. ### Claim 12 (depends on 11) The computer-implemented method of claim 11, wherein collecting feedback regarding the response to the user input comprises requesting the feedback from the user. ### Claim 13 (depends on 11) The computer-implemented method of claim 11, wherein collecting feedback regarding the response to the user input comprises inferring the feedback from a user response to the response to the user input. ### Claim 14 (depends on 10) The computer-implemented method of claim 10, further comprising adding composite intent nodes that are frequently requested by users to a set of conversation trees of the intent graph, where composite intent nodes are frequently requested by users when use of the composite intent nodes exceeds a predetermine threshold. ### Claim 15 (depends on 10) The computer-implemented method of claim 10, further comprising requesting additional information from the user in response to a determination that the path includes at least one input entity that requires the additional information from the user. ### Claim 16 (depends on 10) The computer-implemented method of claim 10, wherein determining the path in the intent graph that connects the input entities to the output entities comprises: identifying multiple paths in the intent graph that connects the input entities to the output entities; and selecting a preferred path from the multiple paths as the path. ### Claim 17 (depends on 16) The computer-implemented method of claim 16, wherein the preferred path is selected from the multiple paths based on a set of criteria comprising path length, number of required additional input entities, occurrences of similar paths in conversation logs, and past user feedback. ### Claim 18 A system configured to support combinations of intents, the system comprising memory for storing instructions, and a processor configured to execute the instructions to: provide a conversation service using a set of conversation trees, each conversation tree within the set of conversation trees comprises a list of intents that define how to answer to a certain type of user input; create an intent graph using the list of intents of each conversation tree within the set of conversation trees; add composite intents to the set of conversation trees using the intent graph to generate a set of updated conversation trees; and configure the conversation service to use the set of updated conversation trees that comprises composite intents. ### Claim 19 (depends on 18) The system of claim 18, wherein the processor further executes instructions to: extract an entity input and entity outputs for each intent in the list of intents for each conversation tree within the set of conversation trees; and assemble the entity input and the entity outputs for each intent in the list of intents for each conversation tree within the set of conversation trees to create the intent graph. ### Claim 20 (depends on 18) The system of claim 18, wherein the processor further executes instructions to generate a sample utterance for a new composite intent.