Server computer and networked computer system for evaluating, storing, and managing labels for classification model evaluation and training

1. An evaluation server computer for use in monitoring an operation of a classification server computer, the classification server computer configured to perform labeling operations including assigning respective estimated labels to respective item records, the respective estimated labels for use in a search engine and including respective information associated with the respective item records, the evaluation computer server computer comprising:

a database including an item list including a plurality of item records, each respective item record of the plurality of item records including a respective initial selection probability associated with a respective previous sampling operation; and

a processing device including a processor coupled to a memory device, the processor programmed to:

receive a first number of trusted labels available for use in a current sampling operation;

determine a respective current selection probability for each respective item record of the plurality of item records included in the item list, the respective current selection probability being a function of:

the first number of trusted labels; and

each respective corresponding initial selection probability of each respective label of the first number of trusted labels;

select a respective portion of the plurality of item records from the item list as a function of each respective corresponding current selection probability of each respective label of the first number of trusted labels;

generate a sampling list including the respective portion of the plurality of item records, wherein a number of the plurality of item records in the sampling list is equal to the first number of trusted labels;

determine a respective estimated label associated with each respective item record of the plurality of item records in the sampling list, wherein each respective estimated label is determined as a function of a current labeling operation performed by the classification server computer;

generate a training group and a testing group using each respective item record of the plurality of item records in the sampling list;

train an accuracy model on the training group, the accuracy model configured to output an accuracy associated with the current labeling operation;

determine the accuracy associated with the current labeling operation using the accuracy model and the testing group;

display the accuracy associated with the current labeling operation on an operator computing device, wherein the accuracy associated with the current labeling operation is determined as a function of:

the respective estimated labels; and

each respective corresponding current selection probability;

update at least one of the respective item records to include at least one of the respective estimated labels as a respective new trusted label of the first number of trusted labels for the at least one of the respective item records;

receive a search query from a user computing device;

analyze the search query using the search engine; and

return an item record of the respective item records, as updated, to the user computing device in response to the search query, wherein:

the accuracy associated with the current labeling operation indicates an accuracy of the respective estimated labels as compared to trusted labels associated with the plurality of item records; and

the plurality of item records are configured to be displayed to a consumer on a website in an ordered list based on a relevancy score in response to receiving a product search request received via the website.

2. An evaluation server computer in accordance with claim 1 , wherein the processor is further programmed to:

determine each respective corresponding current selection probability of each respective label of the first number of trusted labels as a function of a second number of trusted labels associated with the respective previous sampling operation.

3. An evaluation server computer in accordance with claim 1 , wherein the processor is further programmed to:

determine each respective corresponding current selection probability of each respective label of the first number of trusted labels as a function of the number of the plurality of item records included in the item list.

4. An evaluation server computer in accordance with claim 1 , wherein the processor is further programmed to:

determine the accuracy associated with the current labeling operation as a function of the number of the plurality of item records included in the item list.

5. An evaluation server computer in accordance with claim 1 , wherein:

the processor is further programmed to:

generate a labeling list including the plurality of item records in the sampling list;

transmit the labeling list to a labeling server computer; and

receive respective trusted labels for each respective item record in the labeling list; and

determine the accuracy associated with the current labeling operation further comprises:

determine the accuracy associated with the current labeling operation as a function of the respective trusted labels, as received.

6. An evaluation server computer in accordance with claim 1 , wherein:

the plurality of item records include information associated with consumer products.

7. An evaluation server computer in accordance with claim 1 , wherein the processor is further programmed to:

update each respective item record of the training group to indicate a respective associated group selected from the training group or the testing group; and

update each respective item record of the testing group to indicate a respective associated group selected from the training group or the testing group.

8. An evaluation server computer in accordance with claim 1 , wherein:

the training group comprises respective trusted labels assigned to each respective item record of the plurality of item records in the sampling list, as received; and

the testing group comprises respective estimated labels as determined for each respective item record of the plurality of item records in the sampling list.

9. An evaluation server computer in accordance with claim 1 , wherein the accuracy model comprises:

a loss function configured to measure error between respective trusted labels assigned to each respective item record of the plurality of item records in the sampling list, as received, and estimated labels as determined for each respective item record of the plurality of item records in the sampling list.

10. An evaluation server computer in accordance with claim 1 , wherein processor is further programmed to:

receive, from the operator computing device, a respective trusted label for at least one respective item record of the plurality of item records, the respective trusted label determined using the accuracy associated with the current labeling operation; and

store the respective new trusted label in the database.

11. An evaluation server computer in accordance with claim 1 , wherein:

analyze the search query using the search engine further comprises:

comparing the search query to the respective new trusted label of the first number of trusted labels to generate the relevance score for the at least one of the respective item records.

12. A networked computer system including:

a classification server computer including a first processor programmed to perform labeling operations including assigning respective estimated labels to respective item records, the respective estimated labels for us in a search engine and including respective information associated with the respective item records;

a database including an item list including a plurality of item records, each respective item record of the plurality of item records including a respective initial selection probability associated with a respective previous sampling operation; and

an evaluation server computer including a second processor programmed to:

receive a first number of trusted labels available for use in a current sampling operation;

determine a respective current selection probability for each respective item record of the plurality of item records included in the item list, the respective current selection probability being a function of:

the first number of trusted labels; and

each respective corresponding initial selection probability of each respective label of the first number of trusted labels;

select a respective portion of the plurality of item records from the item list as a function of each respective corresponding current selection probability of each respective label of the first number of trusted labels;

generate a sampling list including the respective portion of the plurality of item records, wherein a number of the plurality of item records in the sampling list is equal to the first number of trusted labels;

determine a respective estimated label associated with each respective item record of the plurality of item records in the sampling list, wherein each respective estimated label is determined as a function of a current labeling operation performed by the classification server computer;

generate a training group and a testing group from each respective item record of the plurality of item records in the sampling list;

train an accuracy model on the training group, the accuracy model configured to output an accuracy associated with the current labeling operation;

determine the accuracy associated with the current labeling operation using the accuracy model;

display the accuracy associated with the current labeling operation on an operator computing device, the accuracy associated with the current labeling operation being determined as a function of:

the respective estimated labels; and

each respective corresponding current selection probability

update at least one of the respective item records to include at least one of the respective estimated labels as a respective new trusted label of the first number of trusted labels for the at least one of the respective item records;

receive a search query from a user computing device;

analyze the search query using the search engine; and

return the at least one of the respective item records of the respective item records, as updated, to the user computing device in response to the search query, wherein:

the accuracy associated with the current labeling operation indicates an accuracy of the respective estimated labels as compared to trusted labels associated with the plurality of item records; and

the plurality of item records are configured to be displayed to a consumer on a website in an ordered list based on a relevancy score in response to receiving a product search request received via the website.

13. A networked computer system in accordance with claim 12 , wherein the second processor is further programmed to:

determine each respective corresponding current selection probability of each respective label of the first number of trusted labels as a function of a second number of trusted labels associated with the respective previous sampling operation.

14. A networked computer system in accordance with claim 12 , wherein the second processor is further programmed to:

determine each respective corresponding current selection probability of each respective label of the first number of trusted labels as a function of the number of the plurality of item records included in the item list.

15. A networked computer system in accordance with claim 12 , further comprising a labeling server computer comprising a third processor, the labeling computer server configured to provide trusted labels associated with item records, the third processor is programmed to:

generate a labeling list including the plurality of item records in the sampling list; and

transmit the labeling list to the labeling server computer, wherein:

the labeling server computer is programmed to:

determine a respective trusted label for each respective item record of the plurality of item records in the labeling list; and

transmit the respective trusted label for each respective item record of the plurality of item records to the evaluation server computer; and

the evaluation server computer is programmed to determine the accuracy associated with the current labeling operation as a function of the respective trusted label for each respective item record of the plurality of item records, as received.

16. A networked computer system in accordance with claim 12 , wherein:

the plurality of item records include information associated with consumer products.

17. A networked computer system in accordance with claim 12 , wherein the second processor is further programmed to:

update each respective item record of the training group to indicate a respective associated group selected from the training group or the testing group; and

update each respective item record of the testing group to indicate a respective associated group selected from the training group or the testing group.

18. A networked computer system in accordance with claim 12 , wherein:

the training group comprises respective trusted labels assigned to each respective item record of the plurality of item records in the sampling list, as received; and

the testing group comprises respective estimated labels as determined for each respective item record of the plurality of item records in the sampling list.

19. A networked computer system in accordance with claim 12 , wherein the accuracy model comprises:

a loss function configured to measure error between respective trusted labels assigned to each respective item record of the plurality of item records in the sampling list, as received, and estimated labels as determined for each respective item record of the plurality of item records in the sampling list.

20. A networked computer system in accordance with claim 12 , wherein the second processor is further programmed to:

receive, from the operator computing device, a respective trusted label for at least one respective item record of the plurality of item records, the respective new trusted label determined using the accuracy associated with the current labeling operation; and

store the respective new trusted label in the database.