# US 2026/0094020 A1 · Generation method and information processing apparatus

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0094020 A1
- Document ID: US20260094020A1
- Status: Published application
- Kind: A1
- Application number: 19/336,421
- Filed: 2025-09-22
- Published: 2026-04-02
- Earliest claimed: 2024-09-27
- Assignees: Fujitsu Limited
- Applicants: Fujitsu Limited
- Inventors: Akihiro WADA; Tatsuya KIKUZUKI; Masatoshi OGAWA; Teruhisa NINOMIYA
- Web page: https://patentgrep.com/patent/US20260094020A1

## CPC

- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06N 20/00: Machine learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A processing unit acquires a combination of a first entity included in a first knowledge graph and a second entity included in a second knowledge graph. The processing unit inputs the first entity and the second entity to a machine learning model and instructs the machine learning model to decrease the similarity between the first entity and the second entity if the relationship between the first entity and the second entity is a predetermined relationship. The processing unit acquires the similarity between the first entity and the second entity output by the machine learning model. The processing unit generates a third knowledge graph by merging the first knowledge graph and the second knowledge graph, treating the first entity and the second entity as identical entities if their similarity is greater than a threshold.

## Claims

### Claim 1

A generation method comprising:
acquiring, by a processor, from a first knowledge graph and a second knowledge graph, a combination of a first entity included in the first knowledge graph and a second entity included in the second knowledge graph, each of the first knowledge graph and the second knowledge graph representing a causal relationship between a plurality of entities;
inputting, by the processor, the first entity and the second entity to a machine learning model and instructing the machine learning model to decrease a similarity between the first entity and the second entity upon determining that a relationship between the first entity and the second entity is a predetermined relationship, the machine learning model being capable of outputting a similarity between two entities in response to an input of the two entities;
acquiring, by the processor, the similarity between the first entity and the second entity, the similarity being output by the machine learning model; and
generating, by the processor, a third knowledge graph by merging the first knowledge graph and the second knowledge graph, treating the first entity and the second entity whose similarity is greater than a threshold as identical entities.

### Claim 2 (depends on 1)

The generation method according to claim 1, wherein the predetermined relationship includes at least one of a cause-effect relationship, a simultaneity relationship, or a subject-predicate relationship.

### Claim 3 (depends on 1)

The generation method according to claim 1, wherein the instructing to the machine learning model includes at least one of an instruction to determine the similarity between the first entity and the second entity through consensus decision-making, an example of the predetermined relationship, or an example answer indicating a similarity between two example sentences.

### Claim 4 (depends on 1)

The generation method according to claim 1, further comprising:
selecting, by the processor, based on a cosine similarity between the first entity and the second entity, a target combination for which a similarity is to be acquired by the machine learning model.

### Claim 5 (depends on 1)

The generation method according to claim 1, wherein the generating of the third knowledge graph includes:
acquiring, by the processor, from the first knowledge graph and the second knowledge graph, a plurality of combinations each containing the first entity and the second entity, the plurality of combinations each having a similarity greater than the threshold;
selecting, by the processor, for each of first entities, a combination having a maximum similarity from the plurality of combinations;
selecting, by the processor, for each of second entities, a combination having a maximum similarity from combinations selected for said each of the first entities; and
determining, by the processor, that the first entity and the second entity belonging to the combination selected for said each of the second entities are identical entities.

### Claim 6 (depends on 4)

The generation method according to claim 4, further comprising:
acquiring, by the processor, for each combination of a plurality of combinations each containing the first entity and the second entity, the plurality of combinations being acquired from the first knowledge graph and the second knowledge graph, a determination result of determining semantic equivalence between the first entity and the second entity belonging to said each combination, based on a comparison between the similarity corresponding to said each combination and the threshold; and
performing, by the processor, fine tuning of a language model using the determination result, the language model being used to calculate the cosine similarity.

### Claim 7 (depends on 1)

The generation method according to claim 1, further comprising:
acquiring, by the processor, from the third knowledge graph and a fourth knowledge graph, a combination of a third entity included in the third knowledge graph and a fourth entity included in the fourth knowledge graph;
acquiring, by the processor, a similarity between the third entity and the fourth entity using the machine learning model; and
generating, by the processor, a fifth knowledge graph by merging the third knowledge graph and the fourth knowledge graph, treating the third entity and the fourth entity whose similarity i s greater than the threshold as identical entities.

### Claim 8 (depends on 1)

The generation method according to claim 1, wherein at least one of the first entity or the second entity is a sentence.

### Claim 9

A non-transitory computer-readable storage medium storing a computer program that causes a computer to perform a process comprising:
acquiring, from a first knowledge graph and a second knowledge graph, a combination of a first entity included in the first knowledge graph and a second entity included in the second knowledge graph, each of the first knowledge graph and the second knowledge graph representing a causal relationship between a plurality of entities;
inputting the first entity and the second entity to a machine learning model and instructing the machine learning model to decrease a similarity between the first entity and the second entity upon determining that a relationship between the first entity and the second entity is a predetermined relationship, the machine learning model being capable of outputting a similarity between two entities in response to an input of the two entities;
acquiring the similarity between the first entity and the second entity, the similarity being output by the machine learning model; and
generating a third knowledge graph by merging the first knowledge graph and the second knowledge graph, treating the first entity and the second entity whose similarity is greater than a threshold as identical entities.

### Claim 10

An information processing apparatus comprising:
a memory configured to store information on a first knowledge graph and a second knowledge graph, each of the first knowledge graph and the second knowledge graph representing a causal relationship between a plurality of entities; and
a processor coupled to the memory and the processor configured to:
acquire a combination of a first entity included in the first knowledge graph and a second entity included in the second knowledge graph;
input the first entity and the second entity to a machine learning model and instruct the machine learning model to decrease a similarity between the first entity and the second entity upon determining that a relationship between the first entity and the second entity is a predetermined relationship, the machine learning model being capable of outputting a similarity between two entities in response to an input of the two entities;
acquire the similarity between the first entity and the second entity, the similarity being output by the machine learning model; and
generate a third knowledge graph by merging the first knowledge graph and the second knowledge graph, treating the first entity and the second entity whose similarity is greater than a threshold as identical entities.
