Generating work products using work product metrics and predicted constituent availability

• US 10,467,638 B2
• Filed: 08/14/2014
• Issued: 11/05/2019
• Est. Priority Date: 08/14/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bhattacharjya, Debarun, Varshney, Kush R., Varshney, Lav R.
• Primary Examiner(s): Miller, Alan S
• Assistant Examiner(s): Ullah, Arif
• Application Number:
  US14/459,848
• Publication Number:
  US 20160048853A1
• Time in Patent Office: 1,909 Days
• Field of Search: None
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities

Abstract

A method includes predicting availability of a plurality of constituents for one or more future epochs, obtaining one or more metrics for each of a plurality of existing work products, each of the plurality of existing work products using at least one constituent, and generating at least one work product for each of the one or more future epochs based in part on the predicted availability of the constituents and the one or more metrics for the existing work products. The metrics for the existing work products may include quality metrics and novelty metrics.

Claims

1. A method comprising:

predicting availability of a plurality of constituents for one or more future epochs, wherein predicting the availability of the plurality of constituents comprises;

querying a knowledge database to obtain information associated with current availability of respective ones of the plurality of constituents for a current epoch and to receive information associated with historical availability of respective ones of the plurality of constituents for one or more previous epochs; and

creating probability distributions of future availability of respective ones of the plurality of constituents for the one or more future epochs utilizing one or more constraints associated with supply chains of respective ones of the plurality of constituents and the received information associated with current and historical availability of respective ones of the plurality of constituents;

obtaining one or more metrics for each of a plurality of existing work products, each of the plurality of existing work products using at least one of the plurality of constituents, wherein obtaining the one or more metrics comprises;

querying the knowledge database to obtain information associated with the plurality of existing work products;

utilizing the obtained information associated with the plurality of existing work products to determine, for respective ones of the existing work products, one or more novelty metrics and one or more quality metrics, the one or more novelty metrics comprising temporal novelty metrics associated with respective constituents, the novelty metric for a given constituent for a given future epoch being based on selection of the given constituent for one or more previous epochs; and

generating at least one work product for each of the one or more future epochs based in part on the predicted availability of the constituents and the one or more metrics for the existing work products, wherein generating the at least one work product comprises;

building an optimization model comprising two or more constraints, the two or more constraints comprising an availability-based constraint determined utilizing the predicted availability of the constituents and a creativity-based constraint determined using a computational creativity model that utilizes the quality metrics and novelty metrics associated with the existing work products to compute a creativity metric for respective ones of a plurality of candidate work products different from the existing work products in the knowledge database; and

utilizing the optimization model to select the at least one work product for each of the one or more future epochs from the plurality of candidate work products;

wherein the method is performed by at least one processing device comprising a processor coupled to a memory;

wherein the creativity-based constraint is a function of;

amounts of one or more constituents utilized in a given candidate work product;

whether the given candidate work product is selected for a given epoch; and

whether the given candidate work product is selected for two or more consecutive ones of the future epochs; and

wherein the optimization model comprises an objective function that takes into account a transfer cost associated with producing the given candidate work product for the given epoch, the transfer cost being based at least in part on whether the given work product was selected for an epoch preceding the given epoch.

2. The method of claim 1, wherein the constituents comprise raw materials for food dishes and the existing work products and generated work products comprise menus of one or more food dishes.

3. The method of claim 1, wherein the constituents comprise events and the existing work products and generated work products comprise itineraries.

4. The method of claim 1, wherein the constituents comprise raw materials for clothing and the existing work products and generated work products comprise catalogs of clothing items.

5. The method of claim 1, wherein creating the probability distributions is further based on one or more of delivery times for the constituents, seasonal constraints relating to availability of the constituents and climate and weather constraints relating to availability of the constituents.

6. The method of claim 1, wherein the probability distributions comprises stochastic forecasts of availability for respective ones of the constituents for each of the one or more future epochs.

7. The method of claim 1, wherein creating the probability distributions uses at least one of a time series model and a regression forecast.

8. The method of claim 1, wherein the quality metrics are obtained using human hedonic perception scoring metrics based on historical data regarding the constituents and the existing work products.

9. The method of claim 1, wherein the novelty metrics are obtained based on a temporal Bayesian surprise metric.

10. The method of claim 1, wherein the novelty metrics evolve over respective ones of the future epochs.

11. The method of claim 1, wherein:

the optimization model comprising a sequential decision problem for two or more future epochs;

utilizing the optimization model to generate the at least one work product for each of the one or more future epochs comprises utilizing a backward recursion over the two or more future epochs to solve the optimization model; and

the optimization model is formulated to maximize total expected profits over time subject to the two or more constraints.

12. The method of claim 11, wherein the creativity-based constraint specifies that creativity of the at least one generated work product exceeds a user-defined threshold.

13. The method of claim 11, wherein the availability-based constraint specifies that the amounts of constituents used by the at least one generated work product in a given epoch does not exceed predicted availability of the constituents for the given epoch.

14. The method of claim 1, further comprising utilizing the at least one generated work product for each of the one or more future epochs to generate, for each of the one or more future epochs, one or more corresponding inventory orders, a given inventory order specifying respective amounts of constituents to order in a given epoch to satisfy the at least one generated work product associated with the given epoch.

15. The method of claim 1, wherein generating the at least one work product for each of the one or more future epochs comprises, for a given future epoch, generating two or more work products, and further comprising providing the two or more work products to a user device over a network for selection of a given work product to utilize in the given future epoch.

20. The method of claim 14, wherein the objective function couples generation of the at least one work product for each of the one or more future epochs with generation of the corresponding inventory orders for the one or more future epochs.

16. An article of manufacture comprising a computer readable storage medium for storing computer readable program code which, when executed, causes a computer to:

predict availability of a plurality of constituents for one or more future epochs, wherein predicting the availability of the plurality of constituents comprises;

querying a knowledge database to obtain information associated with current availability of respective ones of the plurality of constituents for a current epoch and to receive information associated with historical availability of respective ones of the plurality of constituents for one or more previous epochs; and

creating probability distributions of future availability of respective ones of the plurality of constituents for the one or more future epochs utilizing one or more constraints associated with supply chains of respective ones of the plurality of constituents and the received information associated with current and historical availability of respective ones of the plurality of constituents;

obtain one or more metrics for each of a plurality of existing work products, each of the plurality of existing work products using at least one of the plurality of constituents, wherein obtaining the one or more metrics comprises;

querying the knowledge database to obtain information associated with the plurality of existing work products;

utilizing the obtained information associated with the plurality of existing work products to determine, for respective ones of the existing work products, one or more novelty metrics and one or more quality metrics, the one or more novelty metrics comprising temporal novelty metrics associated with respective constituents, the novelty metric for a given constituent for a given future epoch being based on selection of the given constituent for one or more previous epochs; and

generate at least one work product for each of the one or more future epochs based in part on the predicted availability of the constituents and the one or more metrics for the existing work products, wherein generating the at least one work product comprises;

building an optimization model comprising two or more constraints, the two or more constraints comprising an availability-based constraint determined utilizing the predicted availability of the constituents and a creativity-based constraint determined using a computational creativity model that utilizes the quality metrics and novelty metrics associated with the existing work products to compute a creativity metric for respective ones of a plurality of candidate work products different from the existing work products in the knowledge database; and

utilizing the optimization model to select the at least one new work product for each of the one or more future epochs from the plurality of candidate work products;

wherein the creativity-based constraint is a function of;

amounts of one or more constituents utilized in a given candidate work product;

whether the given candidate work product is selected for a given epoch; and

whether the given candidate work product is selected for two or more consecutive ones of the future epochs; and

wherein the optimization model comprises an objective function that takes into account a transfer cost associated with producing the given candidate work product for the given epoch, the transfer cost being based at least in part on whether the given work product was selected for an epoch preceding the given epoch.

17. The article of manufacture of claim 16, wherein the computer readable program code, when executed, further causes a computer to utilize the at least one generated work product for each of the one or more future epochs to generate, for each of the one or more future epochs, one or more corresponding inventory orders, a given inventory order specifying respective amounts of constituents to order in a given epoch to satisfy the at least one generated work product associated with the given epoch.

18. An apparatus comprising:

a memory; and

a processing device operatively coupled to the memory and configured to;

predict availability of a plurality of constituents for one or more future epochs, wherein predicting the availability of the plurality of constituents comprises;

querying a knowledge database to obtain information associated with current availability of respective ones of the plurality of constituents for a current epoch and to receive information associated with historical availability of respective ones of the plurality of constituents for one or more previous epochs; and

creating probability distributions of future availability of respective ones of the plurality of constituents for the one or more future epochs utilizing one or more constraints associated with supply chains of respective ones of the plurality of constituents and the received information associated with current and historical availability of respective ones of the plurality of constituents;

obtain one or more metrics for each of a plurality of existing work products, each of the plurality of existing work products using at least one of the plurality of constituents, wherein obtaining the one or more metrics comprises;

querying the knowledge database to obtain information associated with the plurality of existing work products;

utilizing the obtained information associated with the plurality of existing work products to determine, for respective ones of the existing work products, one or more novelty metrics and one or more quality metrics, the one or more novelty metrics comprising temporal novelty metrics associated with respective constituents, the novelty metric for a given constituent for a given future epoch being based on selection of the given constituent for one or more previous epochs; and

generate at least one work product for each of the one or more future epochs based in part on the predicted availability of the constituents and the one or more metrics for the existing work products, wherein generating the at least one work product comprises;

building an optimization model comprising two or more constraints, the two or more constraints comprising an availability-based constraint determined utilizing the predicted availability of the constituents and a creativity-based constraint determined using a computational creativity model that utilizes the quality metrics and novelty metrics associated with the existing work products to compute a creativity metric for respective ones of a plurality of candidate work products different from the existing work products in the knowledge database; and

utilizing the optimization model to select the at least one new work product for each of the one or more future epochs from the plurality of candidate work products;

wherein the creativity-based constraint is a function of;

amounts of one or more constituents utilized in a given candidate work product;

whether the given candidate work product is selected for a given epoch; and

whether the given candidate work product is selected for two or more consecutive ones of the future epochs; and

wherein the optimization model comprises an objective function that takes into account a transfer cost associated with producing the given candidate work product for the given epoch, the transfer cost being based at least in part on whether the given work product was selected for an epoch preceding the given epoch.

19. The apparatus of claim 18, wherein the processing device is further configured to utilize the at least one generated work product for each of the one or more future epochs to generate, for each of the one or more future epochs, one or more corresponding inventory orders, a given inventory order specifying respective amounts of constituents to order in a given epoch to satisfy the at least one generated work product associated with the given epoch.