Topic clustering to generate formulations

Patent metadata

Field	Value
Publication number	US-10949470-B2
Application number	US-201916274815-A
Country	US
Kind code	B2
Filing date	Feb 13, 2019
Priority date	Feb 13, 2019
Publication date	Mar 16, 2021
Grant date	Mar 16, 2021

Abstract

A computer-implemented method is provided for generating a new formulation. The method includes dividing each of input formulations into constituent topics, based on analysis results for an analysis of the input formulations using a topic model algorithm. The method further incudes includes receiving an input query that specifies a set of fragrance. notes to he used to generate the new formulation, The method also includes choosing one of the input formulations which includes the set of fragrance notes to be used to generate the new formulation. The method additionally includes clustering the constituent topics of the chosen one of the input formulations based on a similarity metric. The method further includes generating the new formulation as a response to the input query by selecting, from the input formulations, materials for each of the clustered ones of the constituent topics.

First claim

What is claimed is:

1. A computer-implemented method for generating a new formulation, comprising:
dividing, by a hardware processor, each of input formulations into constituent topics, based on analysis results for an analysis of the input formulations using a topic model algorithm;
receiving, by the hardware processor, an input query that specifies a set of fragrance notes to be used to generate the new formulation;
choosing, by the hardware processor, one of the input formulations which includes the set of fragrance notes to be used to generate the new formulation;
clustering, by the hardware processor, the constituent topics of the chosen one of the input formulations based on a similarity metric; and
generating, by the hardware processor, the new formulation as a response to the input query by selecting, from the input formulations, materials for each of the clustered ones of the constituent topics.

2. The computer-implemented method of claim 1 , wherein the topic model algorithm is the Latent Dirichlet Allocation Algorithm.

3. The computer-implemented method of claim 1 , wherein said analyzing step is performed as a pre-processing step of the method.

4. The computer-implemented method of claim 1 , wherein said choosing step chooses the one of the input formulations, which includes at least the set of fragrance notes to be used to generate the new formulation, randomly from the input formulations.

5. The computer-implemented method of claim 4 , wherein the input formulations comprise at least two input formulations which include the set of fragrance notes to be used to generate the new formulation, one of which is randomly chosen by said choosing step.

6. The computer-implemented method of claim 1 , where said choosing step comprises:
for each of the clustered ones of the constituent topics, choosing one or more of the input formulations that include the each of the clustered ones of the constituent topics; and
extracting the materials for each of the clustered ones of the constituent topics from the chosen one or more of the input formulations.

7. The computer-implemented method of claim 1 , wherein said choosing step is biased to favor choosing any of the input formulations having more of the constituent topics than other ones of the input formulations having less of the constituent topics from among the constituent topics of the chosen one of the input formulations.

8. The computer-implemented method of claim 1 , wherein said clustering step is performed using an affinity propagation algorithm with the similarity metric.

9. The computer-implemented method of claim 1 , further comprising reducing a number of the materials selected for each of the clustered ones of the constituent topics by selecting only a subset of the materials in said selecting step.

10. The computer-implemented method of claim 1 , wherein said selecting step is biased to exclude pairs of the materials that are unpaired in any of the input formulations.

11. The computer-implemented method of claim 1 , further comprising manufacturing the new formulation.

12. A computer program product for generating a new formulation, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:
dividing, by a hardware processor, each of input formulations into constituent topics, based on analysis results for an analysis of the input formulations using a topic model algorithm;
receiving, by the hardware processor, an input query that specifies a set of fragrance notes to he used to generate the new formulation;
choosing, by the hardware processor, one of the input formulations which includes the set of fragrance notes to be used to generate the new formulation;
clustering, by the hardware processor, the constituent topics of the chosen one of the input formulations based on a similarity metric; and
generating, by the hardware processor, the new formulation as a response to the input query by selecting, from the input formulations, materials for each of the clustered ones of the constituent topics.

13. The computer program product of claim 12 , wherein the topic model algorithm is the Latent Dirichlet Allocation Algorithm.

14. The computer program product of claim 12 , wherein said analyzing step is performed as a pre-processing step of the method.

15. The computer program product of claim 12 , wherein said choosing step chooses the one of the input formulations, which includes at least the set of fragrance notes to be used to generate the new formulation, randomly from the input formulations.

16. The computer program product of claim 15 , wherein the input formulations comprise at least two input formulations which include the set of fragrance notes to be used to generate the new formulation, one of which is randomly chosen by said choosing step.

17. The computer program product of claim 12 , where said choosing step comprises:
for each of the clustered ones of the constituent topics, choosing one or more of the input formulations that include the each of the clustered ones of the constituent topics; and
extracting the materials for each of the clustered ones of the constituent topics from the chosen one or more of the input formulations.

18. The computer program product of claim 12 , wherein said choosing step is biased to favor choosing any of the input formulations having more of the constituent topics than other ones of the input formulations having less of the constituent topics from among the constituent topics of the chosen one of the input formulations.

19. The computer program product of claim 12 , wherein said clustering step is performed using an affinity propagation algorithm with the similarity metric.

20. A computer processing system for generating a new formulation, comprising:
a memory for storing program code; and
a processor for running the program code to
divide each of input formulations into constituent topics, based on analysis results for an analysis of the input formulations using a topic model algorithm;
receive an input query that specifics a set of fragrance notes to be used to generate the new formulation;
choose one of the input formulations which includes the set of fragrance notes to be used to generate the new formulation;
cluster the constituent topics of the chosen one of the input formulations based on a similarity metric; and
generate the new formulation as a response to the input query by selecting, from the input formulations, materials for each of the clustered ones of the constituent topics.

Assignees

• IBM

Inventors

• Yanagisawa Hiroki
• Obara Yachiko
• Imamichi Takashi
• Morimura Tetsuro

Classifications

• G06F16/90332 Primary

  Natural language query formulation or dialogue systems · CPC title

• G06F16/906 Primary

  Clustering; Classification · CPC title

• G06F9/5033

  considering data affinity · CPC title