Life plan proposal device, life plan proposal method, and program storage medium
An acquisition unit of a life plan proposal device acquires attribute information indicating a user and living condition information relevant to a living condition of the user. An expansion unit adds additional attribute information of the user by an information generation method using the acquired attribute information. An estimation unit estimates a life event that is likely to happen in the user's life and a time of the life event by using the attribute information including the added attribute information, and the living condition information of the user. An output unit outputs a life plan generated by using information of the estimated life event.
1 . A life plan proposal device, comprising:
a memory configured to store instructions; and
at least one processor configured to execute the instructions to:
acquire correctness/incorrectness information on information of life event from a user;
generate a model by executing machine-learning based on the information of the life event associated with the correctness/incorrectness information;
acquire attribute information indicating a user and living condition information relevant to a living condition of the user;
generate, based on the acquired attribute information, a query by inserting the acquired attribute information into a predetermined query template;
input the generated query to a language model configured to output candidate additional attribute information in response to the generated query;
obtain, from the language model, (i) the candidate additional attribute information and (ii) confidence information indicating an appearance probability associated with each candidate additional attribute information;
select, as additional attribute information of the user, one or more candidates whose confidence information satisfies a predetermined threshold;
store, in association with the user, the selected additional attribute information together with the confidence information;
estimate a life event that is likely to happen in the user's life and a time of the life event by using the generated model, based on the attribute information including the selected additional attribute information, the confidence information, and the living condition information of the user; and
output a life plan generated by using information of the estimated life event.
2 . The life plan proposal device according to claim 1 ,
wherein the at least one processor is further configured to execute the instructions to:
generate the life plan of the user by using information of the estimated time of the life event.
3 . The life plan proposal device according to claim 1 ,
further comprising an information generation comprising using a language model that outputs attribute information in response to input of a query including the acquired attribute information.
4 . The life plan proposal device according to claim 1 ,
wherein the at least one processor is further configured to execute the instructions to:
acquire a collection of pieces of information including the living condition information on a plurality of users, group the living condition information of the plurality of users in the collection of pieces of information by using a clustering technology, and acquire the living condition information of each user by using a feature analyzed for each group.
5 . The life plan proposal device according to claim 1 ,
wherein the at least one processor is further configured to execute the instructions to:
acquire information of whether information of the estimated life event is correct from the user.
6 . A life plan proposal method, comprising:
by a computer:
acquiring correctness/incorrectness information on information of life event from a user;
generating a model by executing machine-learning based on the information of the life event associated with the correctness/incorrectness information;
acquiring attribute information indicating a user and living condition information relevant to a living condition of the user;
generating, based on the acquired attribute information, a query by inserting the acquired attribute information into a predetermined query template;
inputting the generated query to a language model configured to output candidate additional attribute information in response to the generated query;
obtaining, from the language model, (i) the candidate additional attribute information and (ii) confidence information indicating an appearance probability associated with each candidate additional attribute information;
selecting, as additional attribute information of the user, one or more candidates whose confidence information satisfies a predetermined threshold;
storing, in association with the user, the selected additional attribute information together with the confidence information;
estimating the life event that is likely to happen in the user's life and a time of the life event by using the generated model, based on the attribute information including the selected additional attribute information, the confidence information, and the living condition information of the user; and
outputting a life plan generated by using information of the estimated life event.
7 . A non-transitory computer readable program storage medium storing a computer program for causing a computer to execute:
acquiring correctness/incorrectness information on information of life event from a user;
generating a model by executing machine-learning based on the information of the life event associated with the correctness/incorrectness information;
acquiring attribute information indicating a user and living condition information relevant to a living condition of the user;
generating, based on the acquired attribute information, a query by inserting the acquired attribute information into a predetermined query template;
inputting the generated query to a language model configured to output candidate additional attribute information in response to the generated query;
obtaining, from the language model, (i) the candidate additional attribute information and (ii) confidence information indicating an appearance probability associated with each candidate additional attribute information;
selecting, as additional attribute information of the user, one or more candidates whose confidence information satisfies a predetermined threshold;
storing, in association with the user, the selected additional attribute information together with the confidence information;
estimating the life event that is likely to happen in the user's life and a time of the life event by using the generated model, based on the attribute information including the selected additional attribute information, the confidence information, and the living condition information of the user; and
outputting a life plan generated by using information of the estimated life event.
8 . The life plan proposal device according to claim 1 ,
wherein the at least one processor is further configured to execute the instructions to:
update the life plan at predetermined timing;
output the updated life plan in a case where an updated life plan provision request is acquired.