System and method for recommending computational commands based on a weighted, ranked hierarchical graph
Inventors
COUTINHO, Roberto • Silveira, Roberto • GOMES, GUILHERME • Nascimento, Carlos
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12393622-B2
Publication Date
2025-08-19
Expiration Date
Abstract
Systems, methods, and computer-readable storage media for recommending computational commands based on a weighted, ranked hierarchical graph. The system converts a map of intended user actions to canonical actions (commands) into a graph representation. Then, via a chatbot algorithm, the system receives an initial intent as a request by a user of the chatbot algorithm. Based on the popularity of alternative canonical actions, the system makes recommendations of alternative actions which are similar to the initial intent provided by the user.
Core Innovation
A chatbot command recommendation system receives a request comprising an intent for a chatbot and an indication of an authorization level. The system identifies a plurality of intents based on the intent, where the plurality of intents are connected to a common higher node in an electronic graph via a plurality of canonical actions configured to execute via one or more processors. Based on edges in the electronic graph, the system determines authorization levels to access the plurality of intents.
Using the determined authorization levels, the system identifies one or more intents from the plurality of intents that satisfy the authorization level. The system selects a canonical action of the plurality of canonical actions that maps to the one or more intents that satisfy the authorization level and provides the canonical action for execution. Recommended canonical actions are ranked and selected based on execution counts and graph or semantic proximity.
The system reformats an intent-to-canonical-action mapping into a weighted hierarchical graph representing intent nodes and canonical action nodes connected by graph edges. The system generates and ranks recommended canonical actions for an input intent based on graph distance measures and semantic similarity, including comparisons using cosine distance and BERT embeddings from canonical action titles. User selection is used, and optional reinforcement learning or Multi-Armed Bandits preference learning can be applied.
Claims Coverage
The document includes three independent claims (system, method, and non-transitory computer-readable medium). Across them, the core inventive coverage is authorization-aware identification of intents in an electronic graph connected to canonical actions, followed by selecting and providing a canonical action for execution.
Authorization-filtered intent identification in an electronic graph
receive a request comprising an intent for a chatbot and an indication of an authorization level; identify a plurality of intents based on the intent, the plurality of intents connected to a common higher node in an electronic graph via a plurality of canonical actions that are configured to execute via the one or more processors; determine, based on edges in the electronic graph, authorization levels to access the plurality of intents; identify, based on the determined authorization levels, one or more intents from the plurality of intents that satisfy the authorization level;
Canonical action mapping and providing for execution
select a canonical action of the plurality of canonical actions that maps to the one or more intents of the plurality of intents that satisfy the authorization level; and provide, for execution, the canonical action.
Non-transitory processor-executable instructions implementing the authorization-aware graph-to-canonical-action selection
a non-transitory computer-readable medium storing processor executable instructions, the processor executable instructions configured to cause one or more processors to receive a request comprising an intent for a chatbot and an indication of an authorization level; identify a plurality of intents based on the intent, the plurality of intents connected to a common higher node in an electronic graph via a plurality of canonical actions that are configured to execute via the one or more processors; determine, based on edges in the electronic graph, authorization levels to access the plurality of intents; identify, based on the determined authorization levels, one or more intents from the plurality of intents that satisfy the authorization level; select a canonical action of the plurality of canonical actions that maps to the one or more intents of the plurality of intents that satisfy the authorization level; and provide, for execution, the canonical action.
The independent claims consistently require an electronic graph connecting intents to canonical actions, determining authorization levels from graph edges, selecting canonical actions that map to authorization-satisfying intents, and providing the selected canonical action for execution. Dependent claim refinements further cover selection criteria such as ranking by execution counts and shortest-distance selection using natural-language-description distances including cosine distance, as well as graph-structure constraints and optional user interaction for selection and execution.
Stated Advantages
Provides authorization-level-based access to intents when selecting canonical actions for a chatbot request.
Selects a canonical action that maps to intents satisfying the authorization level for execution.
Supports ranking of recommended canonical actions using execution counts and graph/semantic proximity.
Documented Applications
Chatbot command recommendation, including recommending and providing canonical executable chatbot actions based on an input intent while enforcing authorization levels.
Interested in licensing this patent?