ESTIMATION APPARATUS, ESTIMATION METHOD, AND PROGRAM

1 . An estimation apparatus, comprising:

a memory; and

at least one processor connected to the memory, wherein

the processor is configured to

create a divided utterance text and a divided script, the divided utterance text being created by dividing an utterance text expressing an utterance content into a predetermined unit and the divided script being created by dividing a script expressing a predetermined utterance content into a predetermined unit, and

estimate, based on the divided utterance text and the divided script, compliance, non-compliance, or both between the utterance content expressed by the utterance text and the utterance content expressed by the script.

2 . The estimation apparatus according to claim 1 , wherein

the processor is configured to estimate

a range of the utterance text in which the utterance text complies with the utterance content expressed by the script,

a range of the utterance text in which the utterance text does not comply with the utterance content expressed by the script,

a range of the script in which there exists the utterance text that complies with the utterance content expressed by the script,

a range of the script in which there does not exist the utterance text that complies with the utterance content expressed by the script, or

any combination thereof.

3 . The estimation apparatus according to claim 1 , wherein

the script is a script in which a predetermined item and the utterance content are associated with each other, and

the processor is configured to

divide the script by a unit of the item to create the divided script.

4 . The estimation apparatus according to claim 3 , wherein

the processor is configured to

aggregate an estimated result of the compliance, the non-compliance, or both for the item or the divided script,

estimate compliance, non-compliance, or both between the utterance content expressed by the divided utterance text and the utterance content expressed by the divided script, and

in a case in which utterance contents of multiple divided utterance texts comply with utterance contents of divided scripts expressing same item, integrate the multiple divided utterance texts.

5 . The estimation apparatus according to claim 4 , wherein

the processor is configured to

calculate a rating score including a precision degree of the utterance text, a recall degree of the utterance text, a precision degree of the script, a recall degree of the script, or any combination thereof, based on the estimated result of the compliance, the non-compliance, or both.

6 . The estimation apparatus according to claim 5 , wherein

the processor is configured to

calculate a degree of compliance between the utterance content expressed by the divided utterance text and the utterance content expressed by the divided script, as the rating score, based on the estimated result of the compliance, the non-compliance, or both, and

visualize the item on a predetermined terminal in a highlighted manner in accordance with the rating score.

7 . The estimation apparatus according to claim 1 , wherein

the script includes the utterance content determined in a node or a link of a graph structure or a tree structure, and

the processor is configured to

create multiple divided utterance texts, each being the divided utterance text, by sequentially arranging the utterance content determined on a path of from an initial node through an end node of the graph structure or the tree structure.

8 . The estimation apparatus according to claim 1 , wherein

the processor is configured to

estimate the compliance, the non-compliance, or both based on an utterance order of the utterance content expressed by the script, auxiliary information on the utterance content, or both.

9 . The estimation apparatus according to claim 8 , wherein

the auxiliary information includes

turn information expressing a number of times of interaction in a case in which two or more persons take turns of uttering or utter alternately.

10 . The estimation apparatus according to claim 1 , wherein

the processor is configured to

perform association between the utterance content expressed by the utterance text and the utterance content expressed by the script using a neural network that is previously trained so as to output a correspondence relation between the utterance text and the script in response to an input of the utterance text and the script, and

estimate the compliance, the non-compliance, or both based on the association.

11 . An estimation method, comprising:

creating a divided utterance text and a divided script, the divided utterance text being created by dividing an utterance text expressing an utterance content into a predetermined unit and the divided script being created by dividing a script expressing a predetermined utterance content into a predetermined unit, and

estimating, based on the divided utterance text and the divided script, compliance, non-compliance, or both between the utterance content expressed by the utterance text and the utterance content expressed by the script,

the creation and the estimation being executed by a computer.

12 . A non-transitory recording medium storing a computer-executable program so as to execute an estimation process, the estimation process including:

creating a divided utterance text and a divided script, the divided utterance text being created by dividing an utterance text expressing an utterance content into a predetermined unit and the divided script being created by dividing a script expressing a predetermined utterance content into a predetermined unit, and

estimating, based on the divided utterance text and the divided script, compliance, non-compliance, or both between the utterance content expressed by the utterance text and the utterance content expressed by the script.