Processing

Please wait...

Settings

Settings

Goto Application

1. WO2020113169 - METHOD AND SYSTEM OF FOOD PRODUCT COMPARISON

Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

[ EN ]

CLAIMS

1 . A method comprising:

training a food product machine learning model using a plurality of data items thereby establishing a trained food product model, the plurality of data items each labeled with values corresponding to a set of attributes, the attributes corresponding to a food product, the data items including any of:

images;

ordering catalog entries; or

unique identifiers;

receiving, by the trained food product model, a query including a food product; and identifying, by the trained food product model, a confidence score that the query has a first value corresponding to a first food product attribute of the set of attributes, wherein the confidence score is based on the training of the food product model.

2. The method of claim 1 , further comprising:

verifying, by the trained food product model, that the confidence score for the query has exceeded a threshold amount, corresponding with the first food product attribute thereby establishing a verified first food product attribute.

3. The method of claim 2, further comprising:

detecting, by the trained food product model, a second food product attribute that is typically associated with a verified first food product attribute;

identifying, by the trained food product model, a confidence score that the query has a second value corresponding to a second food product attribute of the set of attributes, wherein the confidence score is based on the training of the food product model; and

verifying, by the trained food product model, that the confidence score for the query has exceeded a threshold amount, corresponding with the second food product attribute thereby establishing a verified second food product attribute.

4. The method of claim 3, further comprising:

identifying, by the food product model, that values associated with the query corresponding to the first food product attribute and the second food product attribute belong to a set of attributes corresponding to a first discrete food product; and

classifying the query as the first discrete food product.

5. The method of claim 4, further comprising:

identifying, by the trained food product model, values for additional attributes of the set of attributes of the first discrete food product;

comparing, by the trained food product model, values associated with the first food product attribute, the second food product attribute, and the additional attributes of the first discrete food product with values associated with the set of attributes of other discrete food products;

generating, by the trained food product model, a search rank score between the first discrete food product and other discrete food products based on similarity between respective values of attributes; and

returning, by the trained food product model, search results including a second discrete food product of the other discrete food products.

6. The method of claim 5, wherein said generating the search rank score further includes a scoring weight applied to each attribute of the set of attributes.

7. The method of claim 6, further comprising:

displaying the search results on a hosted marketplace application user interface.

8. The method of claim 1 , wherein the trained food product model is based on a:

convolutional neural network architecture;

hidden Markov model architecture; or

few-shot model architecture.

9. A method for recommending a food product substitute, the method comprising:

receiving, by a trained food product model, a query, the query referring to a first discrete food product, wherein discrete food products are associated in the trained food product model with a combination of values corresponding to a set of attributes;

comparing, by the trained food product model, values associated with the set of attributes of the first discrete food product with values associated with the set of attributes of other discrete food products;

generating, by the trained food product model, a search rank score between the first discrete food product and other discrete food products based on similarity between respective values of attributes; and

returning, by the trained food product model, search results including a second discrete food product of the other discrete food products.

10. The method of claim 9, wherein the query is embodied as any of:

an image;

an ordering catalog entry; or

a unique identifier.

1 1. The method of claim 10, wherein the ordering catalog entry and the unique identifier are each values of the set of attributes.

12. The method of claim 9, wherein said generating the search rank score further includes a scoring weight applied to each attribute of the set of attributes.

13. The method of claim 9, further comprising:

displaying the search results on a hosted marketplace application user interface.

14. The method of claim 9, wherein said generating the search rank score is further based on a proximity of a user address to a delivery vehicle route associated with a given discrete food product.

15. A system comprising:

a food product machine learning model configured using a plurality of data items including images, ordering catalog entries, and unique identifiers, the plurality of data items each labeled with values corresponding to a set of attributes, the attributes corresponding to a food product;

a memory storing an underlying dataset including the plurality of data items for the food product machine learning model;

a search engine network server configured to receive a query including a food product, the search engine configured to use the food product machine learning model to link the query to a first discrete food product via identification of the values corresponding to the set of attributes.

15. The system of claim 14, wherein the search engine further executes instructions in response to the query to:

compare values associated with the set of attributes of the first discrete food product with values associated with the set of attributes of other discrete food products;

generate a search rank score between the first discrete food product and other discrete food products based on similarity between respective values of attributes; and return search results including a second discrete food product of the other discrete food products.

16. The system of claim 15, wherein said generation of the search rank score further includes a scoring weight applied to each attribute of the set of attributes.

17. The system of claim 16, further comprising:

a user interface associated with a client application configured to display the search results.

18. The system of claim 14, wherein the trained food product model is based on a:

convolutional neural network architecture;

hidden Markov model architecture; or

few-shot model architecture.

19. The system of claim 15, wherein said generation of the search rank score is further based on a proximity of a user address to a delivery vehicle route associated with a given discrete food product.

20. The system of Claim 17, wherein, displayed search results on the user interface are linked to an order form for the second discrete food product within the client application.