Granted Patent B1

US 12,699,727 · App. 19/184,423 · Granted Aug 4, 2026

Inventors: Ron Kimmel (Haifa, IL); Ron Slossberg (Raanana, IL); Aviad Zabatani (Kfar Vitkin, IL); Sagi Ben Moshe (Ein Ayala, IL); Ofir Mulla (Talme Elazar, IL); Noam Rotstein (Haifa, IL); Tomer Weiss (Kadarim, IL)

Assignee: LUMANA INC.

G06F16/435G06F16/243G06F16/2438

Abstract

Claims (52)

1 . A method for structured querying, comprising:

embedding text of a request in order to create an embedding of the text of the request;

comparing the embedding of the text of the request to at least a portion of each of a plurality of query structure examples during a comparison, wherein each query structure example of the plurality of query structure examples includes a natural language text portion and a structured query portion, wherein the structured query portion of each query structure example is structured according to a format used by a database, wherein the comparison includes comparing the embedding of the text of the request to a plurality of embeddings of the plurality of query structure examples, wherein the embedding of the at least a portion of each of the plurality of query structure examples is a reduced representation of the at least a portion of the query structure example;

identifying at least one matching query structure example among the plurality of query structure examples based on the comparison;

generating a first query based on the at least one matching query structure example and the text of the request;

providing the first query to a language model in order to obtain a set of outputs from the language model; and

querying at least a portion of the database using a second query, wherein the second query is based on the set of outputs from the language model, wherein the second query is structured according to the format used by the database, wherein the at least a portion of the database is queried by a first system, wherein a plurality of media content items is stored in a plurality of storages accessible to a plurality of second systems;

causing transmission of at least a portion of at least one media content item among the plurality of media content items from the plurality of second systems to a third system, wherein the at least one media content item is determined based on the results of querying the at least a portion of the database using the second query performed by the first system, wherein causing the transmission of the at least a portion of the identified at least one media content item further includes sending a corresponding identifier for each of the at least one media content item to a respective second system of the plurality of second systems which is configured to access a respective storage among the plurality of storages in which the media content item is stored.

2 . The method of claim 1 , further comprising:

providing at least one function to the language model, wherein the at least one function is configured to access the database when called by the language model, wherein the set of outputs from the language model includes at least one call to the at least one function.

3 . The method of claim 2 , wherein the request is from a first entity of a plurality of entities, wherein the database includes a plurality of subsets, wherein each subset of the database corresponds to a respective entity of the plurality of entities, wherein the at least one function is configured to limit access by the language model to only at least one subset of the plurality of subsets corresponding to the first entity.

4 . The method of claim 1 , wherein the request is from a first entity of a plurality of entities, wherein the database includes a plurality of subsets, wherein each subset of the database corresponds to a respective entity of the plurality of entities, further comprising:

identifying at least one first subset among the plurality of subsets which corresponds to the first entity, wherein only the at least one first subset of the database is queried using the second query.

5 . The method of claim 1 , wherein the request indicates at least one attribute of a plurality of attributes in natural language, further comprising:

defining an event with respect to the at least one attribute;

analyzing media content in order to detect the event in at least one portion of the media content; and

causing streaming of the at least one portion of the media content when the event is detected.

6 . The method of claim 1 , wherein the request is received via a first language model session with a first language model, wherein the language model to which the first query is provided is a second language model, wherein the first query is provided to the second language model during a second language model session.

7 . The method of claim 1 , wherein the database stores attribute data indicating a plurality of attributes identified within a plurality of media content items, wherein results of querying the database using the second query include at least one instance of at least one attribute among the plurality of attributes, further comprising:

identifying the at least one media content item among the plurality of media content items as including the at least one attribute based on the results of querying the database using the second query.

8 . The method of claim 1 , wherein the first system is a cloud server, wherein each of the plurality of second systems is an edge system, wherein the first system is remote from each of the plurality of second systems.

9 . A non-transitory computer-readable medium having stored thereon instructions for causing a processing circuitry to execute a process, the process comprising:

embedding text of a request in order to create an embedding of the text of the request;

comparing the embedding of the text of the request to at least a portion of each of a plurality of query structure examples during a comparison, wherein each query structure example of the plurality of query structure examples includes a natural language text portion and a structured query portion, wherein the structured query portion of each query structure example is structured according to a format used by a database, wherein the comparison includes comparing the embedding of the text of the request to a plurality of embeddings of the plurality of query structure examples, wherein the embedding of the at least a portion of each of the plurality of query structure examples is a reduced representation of the at least a portion of the query structure example;

identifying at least one matching query structure example among the plurality of query structure examples based on the comparison;

generating a first query based on the at least one matching query structure example and the text of the request;

providing the first query to a language model in order to obtain a set of outputs from the language model; and

querying at least a portion of the database using a second query, wherein the second query is based on the set of outputs from the language model, wherein the second query is structured according to the format used by the database, wherein the at least a portion of the database is queried by a first system, wherein a plurality of media content items is stored in a plurality of storages accessible to a plurality of second systems;

causing transmission of at least a portion of at least one media content item among the plurality of media content items from the plurality of second systems to a third system, wherein the at least one media content item is determined based on the results of querying the at least a portion of the database using the second query performed by the first system, wherein causing the transmission of the at least a portion of the identified at least one media content item further includes sending a corresponding identifier for each of the at least one media content item to a respective second system of the plurality of second systems which is configured to access a respective storage among the plurality of storages in which the media content item is stored.

10 . A system for structured querying, comprising:

a processing circuitry; and

a memory, the memory containing instructions that, when executed by the processing circuitry, configure the system to:

embed text of a request in order to create an embedding of the text of the request;

compare the embedding of the text of the request to at least a portion of each of a plurality of query structure examples during a comparison, wherein each query structure example of the plurality of query structure examples includes a natural language text portion and a structured query portion, wherein the structured query portion of each query structure example is structured according to a format used by a database, wherein the system is further configured to compare the embedding of the text of the request to a plurality of embeddings of the plurality of query structure examples, wherein the embedding of the at least a portion of each of the plurality of query structure examples is a reduced representation of the at least a portion of the query structure example;

identify at least one matching query structure example among the plurality of query structure examples based on the comparison;

generate a first query based on the at least one matching query structure example and the text of the request;

provide the first query to a language model in order to obtain a set of outputs from the language model; and

query at least a portion of the database using a second query, wherein the second query is based on the set of outputs from the language model, wherein the second query is structured according to the format used by the database, wherein the at least a portion of the database is queried by a first system, wherein a plurality of media content items is stored in a plurality of storages accessible to a plurality of second systems;

cause transmission of at least a portion of at least one media content item among the plurality of media content items from the plurality of second systems to a third system, wherein the at least one media content item is determined based on the results of querying the at least a portion of the database using the second query performed by the first system, wherein causing the transmission of the at least a portion of the identified at least one media content item further includes sending a corresponding identifier for each of the at least one media content item to a respective second system of the plurality of second systems which is configured to access a respective storage among the plurality of storages in which the media content item is stored.

11 . The system of claim 10 , wherein the system is further configured to:

provide at least one function to the language model, wherein the at least one function is configured to access the database when called by the language model, wherein the set of outputs from the language model includes at least one call to the at least one function.

12 . The system of claim 11 , wherein the request is from a first entity of a plurality of entities, wherein the database includes a plurality of subsets, wherein each subset of the database corresponds to a respective entity of the plurality of entities, wherein the at least one function is configured to limit access by the language model to only at least one subset of the plurality of subsets corresponding to the first entity.

13 . The system of claim 10 , wherein the request is from a first entity of a plurality of entities, wherein the database includes a plurality of subsets, wherein each subset of the database corresponds to a respective entity of the plurality of entities, wherein the system is further configured to:

identify at least one first subset among the plurality of subsets which corresponds to the first entity, wherein only the at least one first subset of the database is queried using the second query.

14 . The system of claim 10 , wherein the request indicates at least one attribute of a plurality of attributes in natural language, wherein the system is further configured to:

define an event with respect to the at least one attribute;

analyze media content in order to detect the event in at least one portion of the media content; and

cause streaming of the at least one portion of the media content when the event is detected.

15 . The system of claim 10 , wherein the request is received via a first language model session with a first language model, wherein the language model to which the first query is provided is a second language model, wherein the first query is provided to the second language model during a second language model session.

16 . The system of claim 10 , wherein the database stores attribute data indicating a plurality of attributes identified within a plurality of media content items, wherein results of querying the database using the second query include at least one instance of at least one attribute among the plurality of attributes, wherein the system is further configured to:

identify the at least one media content item among the plurality of media content items as including the at least one attribute based on the results of querying the database using the second query.

17 . The system of claim 10 , wherein the first system is a cloud server, wherein each of the plurality of second systems is an edge system, wherein the first system is remote from each of the plurality of second systems.

Assignments (1)

References Cited (62)