Disclosure storage and decision system and application for resolving a transaction claim
View Patent ↗Disclosed herein are system, method, and computer program product aspects for resolving a transaction claim associated with a merchant system. An example implementation receives, from a user device, a transaction claim associated with a merchant system. The implementation performs an information search within a database storing disclosure information between the user device and the merchant system. The implementation retrieves, from the database, specific disclosure information associated with the transaction claim. The implementation generates a prompt that directs an LLM to generate one or more actions associated with the transaction claim. The implementation determines whether at least one of the one or more actions applies to the user device. When the at least one of the one or more actions applies to the user device, the implementation determines whether it applies to the merchant system. The implementation resolves the transaction claim based on the at least one of the one or more actions.
1 . A method, comprising:
performing, by an action generation engine using one or more processors, an information search within a database storing disclosure information between a user device and a merchant system, wherein the disclosure information is stored as one or more data chunks, and wherein the merchant system does not share the disclosure information with the user device without the action generation engine;
retrieving, by the action generation engine, a subset of the one or more data chunks from the database representing specific disclosure information associated with a transaction claim received from the user device and associated with the merchant system, wherein the specific disclosure information comprises at least part of the disclosure information;
generating, by the action generation engine, a prompt based on combining the transaction claim, the subset of the one or more data chunks, and one or more instructions for querying a large language model (LLM);
generating, by the action generation engine using the LLM and the prompt, one or more actions to resolve the transaction claim from the user device; and
in response to the one or more actions being generated, resolving, by the action generation engine, the transaction claim associated with the merchant system, comprising:
transmitting, to the merchant system by the action generation engine based on the one or more actions applying to the merchant system, the one or more actions to cause the merchant system to execute the one or more actions to address the transaction claim; or
generating, by the action generation engine, a dispute associated with the transaction claim based on the one or more actions not applying to the merchant system.
2 . The method according to claim 1 , wherein the disclosure information comprises a transaction record of the user device associated with the merchant system, an agreement between the user device and the merchant system, a merchant policy associated with the transaction claim, a historical transaction claim, and an action associated with the historical transaction claim.
3 . The method according to claim 1 , further comprising:
obtaining the disclosure information between the user device and the merchant system over a period of time to extract a change of the disclosure information, wherein the change is identified from at least an agreement between the user device and the merchant system or a merchant policy associated with the transaction claim; and
updating the disclosure information in the database to incorporate the change of the disclosure information.
4 . The method according to claim 1 , further comprising:
analyzing, within the database, one or more historical transaction claims and one or more actions associated with the one or more historical transaction claims to generate a user behavior pattern associated with the user device and a merchant behavior pattern associated with the merchant system;
generating a user score indicating a user trustworthiness level from the user behavior pattern; and
generating a merchant score indicating a merchant trustability level from the merchant behavior pattern, wherein the user score and the merchant score are factored into at least resolving the transaction claim or receiving a new transaction claim associated with the merchant system.
5 . The method according to claim 1 , wherein the database is generated by the action generation engine and updated over a period of time to incorporate new disclosure information between the user device and the merchant system, comprising:
tokenizing the disclosure information and the new disclosure information to obtain the one or more data chunks;
generating a set of data embedding associated with the one or more data chunks, wherein each data embedding is associated with a data chunk;
generating one or more vector indexes to store the set of data embedding; and
storing the one or more vector indexes into the database.
6 . The method according to claim 1 , wherein retrieving the subset of the one or more data chunks comprises:
generating a second data embedding associated with the transaction claim;
calculating a set of distance metrics between the second data embedding and a set of data embedding associated with the one or more data chunks; and
identifying the subset of the one or more data chunks stored in the database based on determining the set of distance metrics within a threshold.
7 . A system, comprising:
an action generation engine configured to generate an action to resolve a transaction claim, wherein the transaction claim is received from a user device and is associated with a merchant system;
a memory configured to store operations; and
one or more processors configured to perform the operations, the operations comprising:
performing, by the action generation engine, an information search within a database storing disclosure information between the user device and the merchant system, wherein the disclosure information is stored as one or more data chunks, and wherein the merchant system does not share the disclosure information with the user device without the action generation engine;
retrieving, by the action generation engine, a subset of the one or more data chunks from the database representing specific disclosure information associated with the transaction claim, wherein the specific disclosure information comprises at least part of the disclosure information;
generating, by the action generation engine, a prompt based on combining the transaction claim, the subset of the one or more data chunks, and one or more instructions querying a large language model (LLM);
generating, by the action generation engine using the LLM and the prompt, one or more actions to resolve the transaction claim from the user device; and
in response to the one or more actions being generated, resolving, by the action generation engine, the transaction claim associated with the merchant system, comprising:
transmitting, to the merchant system by the action generation engine based on the one or more actions applying to the merchant system, the one or more actions to cause the merchant system to execute the one or more actions to address the transaction claim; or
generating, by the action generation engine, a dispute associated with the transaction claim based on the one or more actions not applying to the merchant system.
8 . The system according to claim 7 , wherein the disclosure information comprises a transaction record of the user device associated with the merchant system, an agreement between the user device and the merchant system, a merchant policy associated with the transaction claim, a historical transaction claim, and an action associated with the historical transaction claim.
9 . The system according to claim 7 , wherein the one or more processors are further configured to perform the operations comprising:
obtaining the disclosure information between the user device and the merchant system over a period of time to extract a change of the disclosure information, wherein the change is identified from at least an agreement between the user device and the merchant system or a merchant policy associated with the transaction claim; and
updating the disclosure information in the database to incorporate the change of the disclosure information.
10 . The system according to claim 7 , wherein the one or more processors are further configured to perform the operations comprising:
analyzing, within the database, one or more historical transaction claims and one or more actions associated with the one or more historical transaction claims to generate a user behavior pattern associated with the user device and a merchant behavior pattern associated with the merchant system;
generating a user score indicating a user trustworthiness level from the user behavior pattern; and
generating a merchant score indicating a merchant trustability level from the merchant behavior pattern, wherein the user score and the merchant score are factored into at least resolving the transaction claim or receiving a new transaction claim associated with the merchant system.
11 . The system according to claim 7 , wherein the database is generated by the action generation engine and updated over a period of time to incorporate new disclosure information between the user device and the merchant system, comprising:
tokenizing the disclosure information and the new disclosure information to obtain the one or more data chunks;
generating a set of data embedding associated with the one or more data chunks, wherein each data embedding is associated with a data chunk;
generating one or more vector indexes to store the set of data embedding; and
storing the one or more vector indexes into the database.
12 . The system according to claim 7 , wherein retrieving the subset of the one or more data chunks comprises:
generating a second data embedding associated with the transaction claim;
calculating a set of distance metrics between the second data embedding and a set of data embedding associated with the one or more data chunks; and
identifying the subset of the one or more data chunks stored in the database based on determining the set of distance metrics within a threshold.
13 . A non-transitory computer-readable storage device having instructions stored thereon, execution of which, by one or more processing devices of an action generation engine, causes one or more processors to perform operations comprising:
performing, by the action generation engine, an information search within a database storing disclosure information between a user device and a merchant system, wherein the disclosure information is stored as one or more data chunks, and wherein the merchant system does not share the disclosure information with the user device without the action generation engine;
retrieving, by the action generation engine, a subset of the one or more data chunks from the database representing specific disclosure information associated with a transaction claim received from the user device and associated with the merchant system, wherein the specific disclosure information comprises at least part of the disclosure information;
generating, by the action generation engine, a prompt based on combining the transaction claim, the subset of the one or more data chunks, and one or more instructions querying a large language model (LLM);
generating, by the action generation engine using the LLM and the prompt, one or more actions to resolve the transaction claim from the user device; and
in response to the one or more actions being generated, resolving, by the action generation engine, the transaction claim associated with the merchant system, comprising:
transmitting, to the merchant system by the action generation engine based on the one or more actions applying to the merchant system, the one or more actions to cause the merchant system to execute the one or more actions to address the transaction claim; or
generating, by the action generation engine, a dispute associated with the transaction claim based on the one or more actions not applying to the merchant system.
14 . The non-transitory computer-readable storage device according to claim 13 , wherein the operations further comprise:
obtaining the disclosure information between the user device and the merchant system over a period of time to extract a change of the disclosure information, wherein the change is identified from at least an agreement between the user device and the merchant system or a merchant policy associated with the transaction claim; and
updating the disclosure information in the database to incorporate the change of the disclosure information.
15 . The non-transitory computer-readable storage device according to claim 13 , wherein the operations further comprise:
analyzing, within the database, one or more historical transaction claims and one or more actions associated with the one or more historical transaction claims to generate a user behavior pattern associated with the user device and a merchant behavior pattern associated with the merchant system;
generating a user score indicating a user trustworthiness level from the user behavior pattern; and
generating a merchant score indicating a merchant trustability level from the merchant behavior pattern, wherein the user score and the merchant score are factored into at least resolving the transaction claim or receiving a new transaction claim associated with the merchant system.
16 . The non-transitory computer-readable storage device according to claim 13 , wherein the database is generated by the action generation engine and updated over a period of time to incorporate new disclosure information between the user device and the merchant system, the operations further comprising:
tokenizing the disclosure information and the new disclosure information to obtain the one or more data chunks;
generating a set of data embedding associated with the one or more data chunks, wherein each data embedding is associated with a data chunk;
generating one or more vector indexes to store the set of data embedding; and
storing the one or more vector indexes into the database.
17 . The non-transitory computer-readable storage device according to claim 13 , wherein retrieving the subset of the one or more data chunks comprises:
generating a second data embedding associated with the transaction claim;
calculating a set of distance metrics between the second data embedding and a set of data embedding associated with the one or more data chunks; and
identifying the subset of the one or more data chunks stored in the database based on determining the set of distance metrics within a threshold.
18 . The method according to claim 1 , wherein the one or more actions are generated by the action generation engine using the LLM and a chaining of the prompt and a response of the LLM being generated based on the prompt, wherein the LLM is first queried by the prompt to generate the response, and wherein in response to receiving the generated response from the LLM, the LLM is second queried by the response to generate the one or more actions.
19 . The system according to claim 7 , wherein the one or more actions are generated by the action generation engine using the LLM and a chaining of the prompt and a response of the LLM being generated based on the prompt, wherein the LLM is first queried by the prompt to generate the response, and wherein in response to receiving the generated response from the LLM, the LLM is second queried by the response to generate the one or more actions.
20 . The non-transitory computer-readable storage device according to claim 13 , wherein the one or more actions are generated by the action generation engine using the LLM and a chaining of the prompt and a response of the LLM being generated based on the prompt, wherein the LLM is first queried by the prompt to generate the response, and wherein in response to receiving the generated response from the LLM, the LLM is second queried by the response to generate the one or more actions.