Information Processing Device and Method, and Program

An information processing method for an information processing device which, based on an estimated expression made up of a plurality of operators including an operator for performing computation of collaborative filtering which employs computation employing an evaluation matrix made up of an evaluation value indicating an evaluation for each of a plurality of users as to each of a plurality of evaluation targets to estimate said evaluation value, generates said estimated expression newly employed for estimation of said evaluation value of an unevaluated target included in said evaluation targets which has not been evaluated by said users, said information processing device comprising:

• estimated expression candidate generating means configured to take said estimated expression as an estimated expression candidate which is a candidate of said estimated expression newly employed for estimation of said evaluation value of said unevaluated target, and generate a new estimated expression wherein a portion of said estimated expression is modified, and an arbitrary new estimated expression, as said estimated expression candidates;
• estimation result generating means configured to estimate said evaluation value as to each of said plurality of evaluation targets of each of said users with said evaluation matrix for each of said estimated expression candidates based on said estimated expression candidate and said evaluation matrix to generate an estimation result made up of a predictive evaluation value which is said estimated evaluation value;
• evaluating means configured to calculate the evaluation of said estimated expression candidate based on said estimation result and said evaluation matrix;
• and
• selecting means configured to select said estimated expression candidate having the highest evaluation calculated by said evaluating means, of said estimated expression candidates, as said estimated expression newly employed for estimation of said evaluation value of said unevaluated target;
• said estimated expression candidate generating means taking said estimated expression as said estimated expression candidate, and generating a new estimated expression in which a part of said estimated expression is modified, and an arbitrary new estimated expression, as said estimated expression candidates;
• said estimation result generating means generating said estimation result based on said estimated candidate and said evaluation matrix;
• said evaluating means calculating the evaluation of said estimated expression candidate;
• and
• said selecting means selecting said estimated expression candidate having the highest evaluation calculated by said evaluating means, of said estimated expression candidates, as said estimated expression newly employed for estimation of said evaluation value of said unevaluated target.

A program for information processing which, based on an estimated expression made up of a plurality of operators including an operator for performing computation of collaborative filtering which employs computation employing an evaluation matrix made up of an evaluation value indicating an evaluation for each of a plurality of users as to each of a plurality of evaluation targets to estimate said evaluation value, generates said estimated expression newly employed for estimation of said evaluation value of an unevaluated target included in said evaluation targets which has not been evaluated by said users, said program causing a computer to execute processing including the steps of:

• taking said estimated expression as an estimated expression candidate which is a candidate of said estimated expression newly employed for estimation of said evaluation value of said unevaluated target, and generating a new estimated expression wherein a portion of said estimated expression is modified, and an arbitrary new estimated expression, as said estimated expression candidates;
• estimating said evaluation value as to each of said plurality of evaluation targets of each of said users with said evaluation matrix for each of said estimated expression candidates based on said estimated expression candidate and said evaluation matrix to generate an estimation result made up of a predictive evaluation value which is said estimated evaluation value;
• calculating the evaluation of said estimated expression candidate based on said estimation result and said evaluation matrix;
• and
• selecting said estimated expression candidate having the highest evaluation calculated, of said estimated expression candidates, as said estimated expression newly employed for estimation of said evaluation value of said unevaluated target.

An information processing device which, based on an estimated expression made up of a plurality of operators including an operator for performing computation of collaborative filtering, which employs computation employing an evaluation matrix made up of an evaluation value indicating an evaluation for each of a plurality of users as to each of a plurality of evaluation targets to estimate said evaluation value, generates said estimated expression newly employed for estimation of said evaluation value of an unevaluated target included in said evaluation targets which has not been evaluated by said users, comprising:

• an estimated expression candidate generating unit configured to take said estimated expression as an estimated expression candidate which is a candidate of said estimated expression newly employed for estimation of said evaluation value of said unevaluated target, and generate a new estimated expression wherein a portion of said estimated expression is modified, and an arbitrary new estimated expression, as said estimated expression candidates;
• an estimation result generating unit configured to estimate said evaluation value as to each of said plurality of evaluation targets of each of said users with said evaluation matrix for each of said estimated expression candidates based on said estimated expression candidates and said evaluation matrix to generate an estimation result made up of a predictive evaluation value which is said estimated evaluation value;
• an evaluating unit configured to calculate the evaluation of said estimated expression candidate based on said estimation result and said evaluation matrix;
• and
• a selecting unit configured to select said estimated expression candidate having the highest evaluation calculated by said evaluating unit, of said estimated expression candidates, as said estimated expression newly employed for estimation of said evaluation value of said unevaluated target.

An information processing method for an information processing device which, based on an estimated expression made up of a plurality of operators including an operator for performing computation of collaborative filtering, which employs computation employing an evaluation matrix made up of an evaluation value indicating an evaluation for each of a plurality of users as to each of a plurality of evaluation targets to estimate said evaluation value, generates said estimated expression newly employed for estimation of said evaluation value of an unevaluated target included in said evaluation targets which has not been evaluated by said users, said information processing device comprising:

• an estimated expression candidate generating unit configured to take said estimated expression as an estimated expression candidate which is a candidate of said estimated expression newly employed for estimation of said evaluation value of said unevaluated target, and generate a new estimated expression wherein a portion of said estimated expression is modified, and an arbitrary new estimated expression, as said estimated expression candidates;
• an estimation result generating unit configured to estimate said evaluation value as to each of said plurality of evaluation targets of each of said users with said evaluation matrix for each of said estimated expression candidates based on said estimated expression candidate and said evaluation matrix to generate an estimation result made up of a predictive evaluation value which is said estimated evaluation value;
• an evaluating unit configured to calculate the evaluation of said estimated expression candidate based on said estimation result and said evaluation matrix;
• and
• a selecting unit configured to select said estimated expression candidate having the highest evaluation calculated by said evaluating unit, of said estimated expression candidates, as said estimated expression newly employed for estimation of said evaluation value of said unevaluated target;
• said estimated expression candidate generating unit taking said estimated expression as said estimated expression candidate, and generating a new estimated expression in which a part of said estimated expression is modified, and an arbitrary new estimated expression, as said estimated expression candidates;
• said estimation result generating unit generating said estimation result based on said estimated candidate and said evaluation matrix;
• said evaluating unit calculating the evaluation of said estimated expression candidate;
• and
• said selecting unit selecting said estimated expression candidate having the highest evaluation calculated by said evaluating unit, of said estimated expression candidates, as said estimated expression newly employed for estimation of said evaluation value of said unevaluated target.