IP Library › Granted Patent US 12,730,839
Granted Patent B2
US 12,730,839 · App. 18/570,539 · Granted Sep 8, 2026

Searching system, searching method, and searching program

Inventors: Takeshi Nagata (Tokyo, JP); Toshio Kasama (Tokyo, JP); Hidemasa Maekawa (Tokyo, JP); Masayoshi Shimomoto (Tokyo, JP); Toru Hagiwara (Tokyo, JP); Hiromitsu Tomozawa (Tokyo, JP); Misato Tsuchiya (Tokyo, JP)
Assignee: MIZUHO RESEARCH & TECHNOLOGIES, LTD.
G06F16/535
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,730,839
App. No.
18/570,539
Granted
Sep 8, 2026
Kind
B2
Abstract

A control unit outputs, to the user terminal, a first item including a first component value in a plurality of principal components composing items, and a plurality of item candidates including component values that are different from the first component value. The control unit specifies a second item selected in the user terminal from among the plurality of item candidates. The control unit calculates, for each principal component, the positional relationship between the first component value and a second component value of the second item. The control unit calculates, for each principal component, a distribution based on the component values according to the positional relationship. The control unit newly generates, for the second item, a plurality of item candidates on the basis of the distribution based on the component values. The new item candidates are output to the user terminal.

Claims (46)

1 . A searching system, comprising:

control circuitry that is connected to a user terminal, wherein

the control circuitry is configured to

output a first item and multiple item candidates to the user terminal based on multiple principal components representing the items, the first item having a first component value, and each of the item candidates having a component value different from the first component value,

identify a second item selected in the user terminal from the item candidates, the second item having a second component value as a component value different from the first component value,

calculate a positional relationship between the first component value and the second component value in each of the principal components,

calculate a component value distribution according to the positional relationship in each of the principal components,

newly generate multiple item candidates based on the component value distribution, and

output the newly generated item candidates to the user terminal, wherein:

the component value distribution comprises a probability distribution having a spread parameter determined based on the positional relationship, and

the multiple item candidates are randomly generated by sampling from the probability distribution.

2 . The searching system according to claim 1 , wherein the control circuitry is configured to calculate the component value distribution according to a distance between the first component value and the second component value.

3 . The searching system according to claim 2 , wherein the control circuitry is configured to

calculate a composite distribution using the component value distribution based on the first component value and the component value distribution based on the second component value, and

newly generate multiple item candidates by using the composite distribution.

4 . The searching system according to any one of claim 1 , wherein the control circuitry is configured to

identify a preferred item and a disliked item selected in the user terminal as the second items from the item candidates,

calculate the positional relationship by using the first item, the preferred item, and the disliked item, and

calculate the component value distribution such that overlap between the component value distribution and a component value of the disliked item is prevented.

5 . The searching system of claim 1 , wherein:

the spread parameter defines a boundary of a range of values of the probability distribution.

6 . The searching system of claim 1 , wherein:

the spread parameter comprises a standard deviation of the probability distribution.

7 . The searching system of claim 6 , wherein:

the standard deviation increases as the positional relationship increases.

8 . The searching system of claim 6 , wherein:

the standard deviation is calculated using a function that multiplies the positional relationship by a proportionality factor.

9 . A method for searching for an item that meets preference of a user by using a searching system, the searching system including control circuitry connected to a user terminal, the method comprising:

outputting, with the control circuitry, a first item and multiple item candidates to the user terminal based on multiple principal components representing the items, the first item having a first component value, and each of the item candidates having a component value different from the first component value;

identifying, with the control circuitry, a second item selected in the user terminal from the item candidates, the second item having a second component value as a component value different from the first component value;

calculating, with the control circuitry, a positional relationship between the first component value and the second component value in each of the principal components;

calculating, with the control circuitry, a component value distribution according to the positional relationship in each of the principal components;

newly generating, with the control circuitry, multiple item candidates for the second item based on the component value distribution; and

outputting, with the control circuitry, the newly generated item candidates to the user terminal, wherein:

the component value distribution comprises a probability distribution having a spread parameter determined based on the positional relationship, and

the multiple item candidates are randomly generated by sampling from the probability distribution.

10 . A non-transitory computer-readable storage medium that stores a searching program causing a processor to execute a searching process for searching for an item that meets preference of a user by using a searching system, the searching system including control circuitry connected to a user terminal, wherein

the control circuitry includes the processor, and the searching process includes

outputting, with the control circuitry, a first item and multiple item candidates to the user terminal based on multiple dimensionality-reduction elements of feature quantities, the dimensionality-reduction elements including multiple dimensions representing the items, the first item having a first component value, and each of the item candidates having a component value different from the first component value;

identifying, with the control circuitry, a second item selected in the user terminal from the item candidates, the second item having a second component value as a component value different from the first component value;

calculating, with the control circuitry, a positional relationship between the first component value and the second component value in each of the elements;

calculating, with the control circuitry, a component value distribution according to the positional relationship in each of the elements;

newly generating, with the control circuitry, multiple item candidates for the second item based on the component value distribution; and

outputting, with the control circuitry, the newly generated item candidates to the user terminal, wherein:

the component value distribution comprises a probability distribution having a spread parameter determined based on the positional relationship, and

the multiple item candidates are randomly generated by sampling from the probability distribution.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 14, 2023
From: NAGATA, TAKESHI; KASAMA, TOSHIO; MAEKAWA, HIDEMASA; SHIMOMOTO, MASAYOSHI; HAGIWARA, TORU; TOMOZAWA, HIROMITSU; TSUCHIYA, MISATO
To: MIZUHO RESEARCH & TECHNOLOGIES, LTD.
Reel/Frame 065876/0091 →
Priority Claims (1)
JP 2021-103506 · Jun 22, 2021 · national
Continuity (1)
Related Publication 20240281464A1 · Aug 22, 2024
References Cited (14)
US 9799041B2 · Karty · 2017 [cited by examiner]
US 11640684B2 · Abdal · 2023 [cited by examiner]
US 11663265B2 · Lin · 2023 [cited by examiner]
US 11698715B2 · Sugawara · 2023 [cited by examiner]
US 20220365636A1 · Sugawara · 2022 [cited by applicant]
JP 2011039944A · 2011 [cited by applicant]
JP 2012008617A · 2012 [cited by applicant]
JP 2021005208A · 2021 [cited by applicant]
WO 2022270203A1 · 2022 [cited by applicant]
Erik Härkönen, Aaron Hertzmann, Jaakko Lehtinen, Sylvain Paris. GANSpace: Discovering Interpretable GAN Controls. Advances in Neural Information Processing Systems 33 (NeurIPS 2020). https://proceedings.neurips.cc/paper… [cited by examiner]
Tejeda-Ocampo, C.; López-Cuevas, A.; Terashima-Marin, H. Improving Deep Interactive Evolution with a Style-Based Generator for Artistic Expression and Creative Exploration. Entropy 2021, 23, 11. https://doi.org/10.3390/… [cited by examiner]
S. Liu, D. Li, T. Cao, Y. Sun, Y. Hu and J. Ji. GAN-Based Face Attribute Editing. IEEE Access, vol. 8, pp. 34854-34867, 2020, doi: 10.1109/ACCESS.2020.2974043. (Year: 2020). [cited by examiner]
Alireza Zaeemzadeh, Shabnam Ghadar, Baldo Faieta, Zhe Lin, Nazanin Rahnavard, Mubarak Shah, Ratheesh Kalarot; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2021, p. 12116-12125 (Year: 2… [cited by examiner]
International Search Report for PCT/JP2022/021344 mailed Jul. 26, 2022, all pages. [cited by applicant]