Adjusting a master build plan based on events using machine learning
As an example, a server may receive and/or retrieve news items and process the news items using natural language processing to identify news related to entities (e.g., people, locations, and organizations) extracted from an enterprise resource planning system. A term frequency-inverse document frequency algorithm may be used to identify critical news items that may impact one or more supply chains associated with at least one product that is to be manufactured. A long short-term memory artificial recurrent neural network may be used to determine a confidence score for each critical news item. The confidence scores of the critical news items may be used to adjust replenishment planning and a master build plan that includes a plan to build the at least one product. In this way, news items may be used to automatically (e.g., without human interaction) adjust the master build plan.
1. A method comprising:
obtaining, by one or more processors, a set of news items associated with entities stored in an entity database, wherein the set of news items is obtained from at least one website via the Internet, wherein at least a portion of the set of news items is pushed from the at least one website to a machine learning system, wherein an intelligent web scraper associated with the machine learning system identifies and pulls the at least a portion of the set of news items associated with the entities stored in the entity database, and wherein natural language processing provided by a natural processing language system applying one or more of support vector machines, Bayesian networks, maximum entropy, conditional random field, neural networks, and deep learning to the at least a portion of the set of news items to extract news features that are associated with the entities stored in the entity database;
processing, by the one or more processors, the extracted news features to identify critical features of the news items that is are predicted to affect one or more supply chains used to supply components to manufacture at least one product;
determining, by the one or more processors, a confidence score associated with each news item in the subset of news items;
modifying, by the one or more processors, a replenishment plan based at least in part on the predicted affect to the one or more supply chains to create a modified replenishment plan; and
performing a modification, by the one or more processors, to a master build plan that includes a build plan associated with the at least one product based at least in part on the modified replenishment plan to modify an impact to the master build plan taking into account the critical features of the news items.
2. The method of claim 1 , wherein: the plurality of entities are extracted from an enterprise resourcing planning (ERP) system.
3. The method of claim 1 , wherein processing the extracted news features to identify critical features of the news items that is predicted to affect the one or more supply chains used to supply components to manufacture the at least one product comprises:
performing a term frequency-inverse document frequency (TF-IDF) analysis of the set of news items associated with the entities;
determining a TF-IDF value that increases proportionally to the number of times a particular entity of the entities appears in each news item offset by a number of news items in the set of news items that includes the word; and
determining the subset of the news items based at least in part on the TF-IDF value associated with each of the entities.
4. The method of claim 1 , wherein determining the confidence score associated with each news item in the subset of news items comprises: determining the confidence score using a long short-term memory (LSTM) recurrent neural network (RNN) algorithm.
5. The method of claim 1 , wherein the plurality of entities comprises: a first plurality of people; a second plurality of locations; and a third plurality of organizations.
6. The method of claim 1 , wherein obtaining the set of news items associated with the entities stored in the entity database comprises: selecting a particular news item;
using natural language processing (NLP) to determine that the particular news item is associated with a particular one of the entities; and
including the particular news item in the set of news items.
7. The method of claim 1 , further comprising: determining the master build plan based at least in part on processing historical sales data using a Naive Bayes algorithm.
8. A server comprising: one or more processors; and one or more non-transitory computer readable media to store instructions executable by the one or more processors to perform operations comprising:
obtaining a set of news items associated with entities stored in an entity database, wherein the plurality of entities are extracted from an enterprise resourcing planning (ERP) system, wherein the set of news items is obtained from at least one website via the Internet, wherein at least a portion of the set of news items is pushed from the at least one website to a machine learning system, wherein an intelligent web scraper associated with the machine learning system identifies and pulls the at least a portion of the set of news items associated with the entities stored in the entity database, and wherein natural language processing provided by a natural processing language system applying one or more of support vector machines, Bayesian networks, maximum entropy, conditional random field, neural networks, and deep learning to the at least a portion of the set of news items to extract news features that are associated with the entities stored in the entity database;
processing the extracted news features to identify critical features of the news items that is are predicted to affect one or more supply chains used to supply components to manufacture at least one product;
determining a confidence score associated with each news item in the subset of news items;
modifying a replenishment plan based at least in part on the predicted affect to the one or more supply chains to create a modified replenishment plan; and
performing a modification to a master build plan that includes a build plan associated with the at least one product based at least in part on the modified replenishment plan to modify an impact to the master build plan taking into account the critical features of the news items.
9. The server of claim 8 , wherein processing the extracted news features to identify critical features of the news items that are predicted to affect the one or more supply chains used to supply components to manufacture the at least one product comprises:
performing a term frequency-inverse document frequency (TF-IDF) analysis of the set of news items associated with the entities; determining a TF-IDF value that increases proportionally to the number of times a particular entity of the entities appears in each news item offset by a number of news items in the set of news items that includes the word; and
determining the subset of the news items based at least in part on the TF-IDF value associated with each of the entities.
10. The server of claim 8 , wherein determining the confidence score associated with each news item in the subset of news items comprises: determining the confidence score using a long short-term memory (LSTM) recurrent neural network (RNN) algorithm.
11. The server of claim 8 , wherein the plurality of entities comprises: a first plurality of people; a second plurality of locations; and a third plurality of organizations.
12. The server of claim 8 , wherein obtaining the set of news items associated with the entities stored in the entity database comprises:
selecting a particular news item;
using natural language processing (NLP) to determine that the particular news item is associated with a particular one of the entities; and
including the particular news item in the set of news items.
13. The server of claim 8 , the operations further comprising: determining the master build plan based at least in part on processing historical sales data using a Naive Bayes algorithm.
14. One or more non-transitory computer readable media storing executable by one or more processors to perform operations comprising:
obtaining, by one or more processors, a set of news items associated with entities stored in an entity database, wherein the set of news items is obtained from at least one website via the Internet, wherein at least a portion of the set of news items is pushed from the at least one website to a machine learning system, wherein an intelligent web scraper associated with the machine learning system identifies and pulls the at least a portion of the set of news items associated with the entities stored in the entity database, and wherein natural language processing provided by a natural processing language system applying one or more of support vector machines, Bayesian networks, maximum entropy, conditional random field, neural networks, and deep learning to the at least a portion of the set of news items to extract news features that are associated with the entities stored in the entity database;
processing, by the one or more processors, the extracted news features to identify critical features of the news items that is are predicted to affect one or more supply chains used to supply components to manufacture at least one product;
determining a confidence score associated with each news item in the subset of news items; modifying a replenishment plan based at least in part on the predicted affect to the one or more supply chains to create a modified replenishment plan; and
performing a modification to a master build plan that includes a build plan associated with the at least one product based at least in part on the modified replenishment plan to modify an impact to the master build plan taking into account the critical features of the news items.
15. The one or more non-transitory computer readable media of claim 14 , wherein: the plurality of entities are extracted from an enterprise resourcing planning (ERP) system.
16. The one or more non-transitory computer readable media of claim 14 , wherein processing the extracted news features to identify critical features of the news items that is are predicted to affect the one or more supply chains used to supply components to manufacture the at least one product comprises:
performing a term frequency-inverse document frequency (TF-IDF) analysis of the set of news items associated with the entities;
determining a TF-IDF value that increases proportionally to the number of times a particular entity of the entities appears in each news item offset by a number of news items in the set of news items that includes the word; and
determining the subset of the news items based at least in part on the TF-IDF value associated with each of the entities.
17. The one or more non-transitory computer readable media of claim 14 , wherein determining the confidence score associated with each news item in the subset of news items comprises: determining the confidence score using a long short-term memory (LSTM) recurrent neural network (RNN) algorithm.
18. The one or more non-transitory computer readable media of claim 14 , wherein the plurality of entities comprises: a first plurality of people; a second plurality of locations; and a third plurality of organizations.
19. The one or more non-transitory computer readable media of claim 14 , wherein obtaining the set of news items associated with the entities stored in the entity database comprises:
selecting a particular news item;
using natural language processing (NLP) to determine that the particular news item is associated with a particular one of the entities; and
including the particular news item in the set of news items.
20. The one or more non-transitory computer readable media of claim 14 , the operations further comprising:
determining the master build plan based at least in part on processing historical sales data using a Naive Bayes algorithm.