Back to results
US 2026/0289707 A1Published application
System
AssigneeSoftBank Group Corp.
InventorNobufusa KOBAYASHI
Life of this patent
Estimating the term…
Abstract
A system includes a processor that is configured to: receive inheritance-related information from a decedent and from an heir; analyze the received information by using a natural language processing technique; and input a prompt sentence to a generative AI model based on a result of the analysis so as to automatically generate a required procedure or a required document.
Claims20
A system comprising:
- circuitry configured to
- receive, via a packet-switched network, a plurality of heterogeneous data records from a terminal device, each heterogeneous data record comprising at least one natural language character sequence and a plurality of structured attribute fields associated with entity identifier data;
- store the plurality of heterogeneous data records in a storage device by indexing each heterogeneous data record with a case identifier;
- apply a natural language processing operation to the at least one natural language character sequence to extract a set of structured feature data comprising entity attribute pairs and candidate classification labels with associated confidence score values;
- determine, based on the set of structured feature data and a set of rule condition definitions stored in the storage device, a list of candidate procedure type codes satisfying respective rule condition predicates;
- generate a structured input sequence for a generative neural network model by inserting values from the structured attribute fields, the set of structured feature data, and the list of candidate procedure type codes into a prompt template data structure comprising a role instruction portion, a task instruction portion, a case summary portion, and an output format specification portion;
- input the structured input sequence to the generative neural network model to obtain output character sequence data;
- parse the output character sequence data according to the output format specification portion to extract a plurality of structured result objects comprising procedure list data, document list data, and draft text data;
- map the plurality of structured result objects to internal data type codes and populate electronic document template data with the draft text data and the structured attribute fields to generate electronic document data;
- and
- transmit, via the packet-switched network, the electronic document data and a progress state indicator to the terminal device for rendering on a display of the terminal device.
8 more elements
A system comprising:
- circuitry configured to
- receive, via a packet-switched network, a plurality of heterogeneous data records from a terminal device, each heterogeneous data record comprising a natural language character sequence and structured attribute fields associated with entity identifier data;
- store the plurality of heterogeneous data records in a storage device indexed by a case identifier;
- apply a natural language processing operation comprising tokenization, named entity recognition, and semantic similarity computation to the natural language character sequence to extract structured feature data comprising entity attribute pairs, candidate classification labels, and confidence score values;
- determine, based on the structured feature data and rule condition definitions stored in the storage device, a list of candidate procedure type codes;
- generate a structured input sequence for a generative neural network model by populating a prompt template data structure with the structured attribute fields, the structured feature data, and the list of candidate procedure type codes, the prompt template data structure comprising a role instruction portion, a task instruction portion, a case summary portion, and an output format specification portion with labeled section delimiters;
- input the structured input sequence to the generative neural network model to obtain output character sequence data;
- parse the output character sequence data by detecting the labeled section delimiters to extract structured result objects comprising procedure list data, document list data, and draft text data;
- populate electronic document templates with the draft text data and the structured attribute fields to generate electronic document data;
- transmit, via the packet-switched network, at least one electronic document data item to an external system interface address;
- and
- transmit, via the packet-switched network, remaining electronic document data and a progress state indicator to the terminal device for rendering on a display of the terminal device.
9 more elements
A method comprising:
- receiving, via a packet-switched network, a plurality of heterogeneous data records from a terminal device, each heterogeneous data record comprising at least one natural language character sequence and a plurality of structured attribute fields associated with entity identifier data;
- storing the plurality of heterogeneous data records in a storage device by indexing each heterogeneous data record with a case identifier;
- applying a natural language processing operation to the at least one natural language character sequence to extract a set of structured feature data comprising entity attribute pairs and candidate classification labels with associated confidence score values;
- determining, based on the set of structured feature data and a set of rule condition definitions stored in the storage device, a list of candidate procedure type codes satisfying respective rule condition predicates;
- generating a structured input sequence for a generative neural network model by inserting values from the structured attribute fields, the set of structured feature data, and the list of candidate procedure type codes into a prompt template data structure comprising a role instruction portion, a task instruction portion, a case summary portion, and an output format specification portion;
- inputting the structured input sequence to the generative neural network model to obtain output character sequence data;
- parsing the output character sequence data according to the output format specification portion to extract a plurality of structured result objects comprising procedure list data, document list data, and draft text data;
- mapping the plurality of structured result objects to internal data type codes and populating electronic document template data with the draft text data and the structured attribute fields to generate electronic document data;
- and
- transmitting, via the packet-switched network, the electronic document data and a progress state indicator to a terminal device for rendering on a display of the terminal device.
7 more elements
Dependencies are inferred from the claim text.