Computer-readable recording medium having stored therein vector estimating program, apparatus for estimating vector, and method for estimating vector

1 . A non-transitory computer-readable recording medium having stored therein a vector estimating program that causes a computer to execute a process comprising:

obtaining a first vector and second entity information, the first vector being generated by using a first model with reference to graph structure data, the graph structure data representing a relationship of a first entity group and being obtained by using first entity information related to the first entity group, the second entity information being obtained by updating the first entity information and being related to the first entity group and a second entity not being included in the first entity group;

generating a second model based on the first vector and a feature value of the first entity group included in the second entity information, the second model being used for obtaining vector data from the second entity information; and

outputting, as an estimated value of a second vector corresponding to the second entity, a result output from the second model by inputting a feature value of the second entity included in the second entity information into the generated second model, wherein

the second model outputs the estimated value without the first model re-learning an entire knowledge graph based on both the updated first entity information and the second entity information, and

the feature value of the second entity includes at least one of a graph structure around a target node and a text label on nodes within the second entity information.

2 . The non-transitory computer-readable recording medium according to claim 1 , wherein the generating of the second model comprises terminating machine learning of the second model when a number of times of performing the machine learning becomes a given number of times or when an error between a vector and the first vector becomes smaller than a given value, the vector being output from the second model when the feature value of the first entity group included in the second entity information is used as an input.

3 . The non-transitory computer-readable recording medium according to claim 1 , wherein

the generating of the second model comprises generating the second model based on one or more feature values common to the first entity group and the second entity, the second model being used for estimating the second vector.

4 . An apparatus for estimating a vector, the apparatus comprising:

a memory; and

a processor coupled to the memory, the processor being configured to

obtain a first vector and second entity information, the first vector being generated by using a first model with reference to graph structure data, the graph structure data representing a relationship of a first entity group and being obtained by using first entity information related to the first entity group, the second entity information being obtained by updating the first entity information and being related to the first entity group and a second entity not being included in the first entity group;

generate a second model based on the first vector and a feature value of the first entity group included in the second entity information, the second model being used for obtaining vector data from the second entity information; and

output, as an estimated value of a second vector corresponding to the second entity, a result output from the second model by inputting a feature value of the second entity included in the second entity information into the generated second model, wherein

the second model outputs the estimated value without the first model re-learning an entire knowledge graph based on both the updated first entity information and the second entity information, and

the feature value of the second entity includes at least one of a graph structure around a target node and a text label on nodes within the second entity information.

5 . The apparatus according to claim 4 , wherein the processor terminates machine learning of the second model when a number of times of performing the machine learning becomes a given number of times or when an error between a vector and the first vector becomes smaller than a given value in the generating of the second model, the vector being output from the second model when the feature value of the first entity group included in the second entity information is used as an input.

6 . The apparatus according to claim 4 , wherein

the processor generates the second model based on one or more feature values common to the first entity group and the second entity in the generating of the second model, the second model being used for estimating the second vector.

7 . A computer-implemented method for estimating a vector, the method comprising:

obtaining a first vector and second entity information, the first vector being generated by using a first model with reference to graph structure data, the graph structure data representing a relationship of a first entity group and being obtained by using first entity information related to the first entity group, the second entity information being obtained by updating the first entity information and being related to the first entity group and a second entity not being included in the first entity group;

generating a second model based on the first vector and a feature value of the first entity group included in the second entity information, the second model being used for obtaining vector data from the second entity information; and

outputting, as an estimated value of a second vector corresponding to the second entity, a result output from the second model by inputting a feature value of the second entity included in the second entity information into the generated second model, wherein

the second model outputs the estimated value without the first model re-learning an entire knowledge graph based on both the updated first entity information and the second entity information, and

the feature value of the second entity includes at least one of a graph structure around a target node and a text label on nodes within the second entity information.

8 . The computer-implemented method according to claim 7 , wherein the generating of the second model comprises terminating machine learning of the second model when a number of times of performing the machine learning becomes a given number of times or when an error between a vector and the first vector becomes smaller than a given value, the vector being output from the second model when the feature value of the first entity group included in the second entity information is used as an input.

9 . The computer-implemented method according to claim 7 , wherein

the generating of the second model comprises generating the second model based on one or more feature values common to the first entity group and the second entity, the second model being used for estimating the second vector.