Methods and apparatus for grouping items
This application relates to apparatus and methods for automatically grouping the same or similar items provided by various suppliers that may use various supplier identification systems to identify the items to a retailer. In some examples, a computing device receives a supplier identifier from a supplier. The supplier identifier identifies an item that the supplier provides to the retailer for sale. The computing device may determine a group identifier for the item, where the group identifier identifies the item within a group of the retailer. The computing device may update the supplier identifier's association to the retailer identifier based on whether the group identifier is also associated with the retailer identifier. In some examples, the computing device executes a machine learning model to determine anomalies within a current database of supplier identifiers and associated retailer identifiers. In some examples, the computing device indicates changes to associations and any detected anomalies.
1. A system comprising:
a database having a hierarchical structure defined by a plurality of supplier-group associations, a plurality of supplier-item associations, and a plurality of group-item associations;
a memory having instructions stored thereon; and
a processor configured to read the memory and coupled to the database, wherein the processor configured to read the instructions to:
receive first supplier data including a first supplier identifier that identifies a first item in a first supplier's numbering system;
store the first supplier data in the database;
determine a first retailer identifier associated with the first supplier identifier, wherein the first retailer identifier identifies the first item in a first retail numbering system, and wherein the first retail identifier corresponds to first item data stored within the database;
generate a first supplier-item association defining a hierarchical relationship between the first supplier data and the first item data in the database;
determine a first group identifier associated with the first supplier identifier, wherein the first group identifier identifies a plurality of items in a second retail numbering system, wherein the first group identifier corresponds to first group data stored within the database, wherein each item of the plurality of items are assigned to the first group identifier, and wherein the plurality of items includes the first item;
generate a first supplier-group association defining a hierarchical relationship between the first supplier data and first group data in the database;
determine a second retailer identifier associated with the first group identifier based on one or more group-item associations defining a hierarchical relationship between at least the first group identifier and an associated set of retailer identifiers, wherein the second retailer identifier corresponds to second item data within the database;
train a machine learning model to detect anomalies in the hierarchical structure of the database, wherein the machine learning model is trained by a supervised training dataset based on attributes of items in the plurality of items, wherein the machine learning model is configured to generate an anomaly score between the first item data and the second item data, and wherein the machine learning model is configured to generate the anomaly score by:
generating a first word embedding representative of at least one attribute of the first item data;
generating a second word embedding representative of the at least one attribute of the second item data; and
comparing the first word embedding and the second word embedding to generate the anomaly score;
generate a second supplier-item association defining a hierarchical relationship between first supplier data and the second item data in the database when the second retailer identifier matches the first retailer identifier, and when the second retailer identifier does not match the first retailer identifier and the anomaly score is below a predetermined threshold;
in response to determining the anomaly score is equal to or above the predetermined threshold, transmit an electronic communication identifying the first item data and the second item data; and
train an additional machine learning model to detect anomalies in the hierarchical structure of the database, wherein the additional machine learning model is trained by a supervised training dataset based on attributes of items in the plurality of items and at least the second supplier-item association.
2. The system of claim 1 , wherein the processor is configured to read the instructions to:
determine that the plurality of group-item associations does not include a group-item association between the first group data and item data; and
generate the group-item association between the first supplier identifier and the first group data.
3. The system of claim 1 , wherein the trained machine learning model is configured to:
determine a score associated with each of a plurality of anomalies; and
compare the score associated with each of the plurality of anomalies to the predetermined threshold, wherein the second supplier-item association is generated when the score associated with each of the plurality of anomalies is below the predetermined threshold.
4. The system of claim 1 , wherein:
the retailer's first numbering system associates a plurality of supplier identifiers to a first plurality of group identifiers and a plurality of retailer identifiers; and
the retailer's second numbering system associates a second plurality of group identifiers and at least a portion of the plurality of supplier identifiers.
5. The system of claim 1 , wherein the processor is further configured to read the instructions to:
determine at least a first attribute for the first item;
determine at least a second attribute of a second item corresponding to a second supplier identifier and a third attribute of a third item corresponding to a third supplier identifier; and
generate the second supplier-item association when a first number of anomalies based on the at least first attribute and the at least second attribute and a second number of anomalies based on the at least first attribute and the at least third attribute are each below the predetermined threshold.
6. The system of claim 1 , wherein each of the at least one-attribute of the first item data and the at least one attribute of the second item data comprise item descriptions.
7. A computer-implemented method comprising:
receiving, by a processor, first supplier data including a first supplier identifier that identifies a first item in a first supplier's numbering system;
storing, by the processor, the first supplier data in a database, wherein the database includes a hierarchical structure defined by a plurality of supplier-group associations, a plurality of supplier-item associations, and a plurality of group-item associations;
determining, by the processor, a first retailer identifier associated with the first supplier identifier, wherein the first retailer identifier identifies the first item in a first retail numbering system, and wherein the first retail identifier corresponds to first item data stored within the database;
generating, by the processor, a first supplier-item association defining a hierarchical relationship between the first supplier data and the first item data in the database;
determining, by the processor, a first group identifier associated with the first supplier identifier, wherein the first group identifier identifies a plurality of items in a second retail numbering system, wherein the first group identifier corresponds to first group data stored within the database, wherein each item of the plurality of items are assigned to the first group identifier, and wherein the plurality of items includes the first item;
generating, by the processor, a first supplier-group association defining a hierarchical relationship between the first supplier data and first group data in the database;
determining, by the processor, a second retailer identifier associated with the first group identifier based on one or more group-item associations defining a hierarchical relationship between at least the first group identifier and an associated set of retailer identifiers, wherein the second retailer identifier corresponds to second item data within the database;
training, by the processor, a machine learning model to detect anomalies in the hierarchical structure of the database, wherein the machine learning model is trained by a supervised training dataset based on attributes of items in the plurality of items, wherein the machine learning model is configured to generating an anomaly score between the first item data and the second item data, and wherein the machine learning model is configured to generate the anomaly score by:
generating a first word embedding representative of at least one attribute of the first item data;
generating a second word embedding representative of the at least one attribute of the second item data; and
comparing the first word embedding and the second word embedding to generate the anomaly score;
generate, by the processor, a second supplier-item association defining a hierarchical relationship between first supplier data and the second item data when the second retailer identifier matches the first retailer identifier, and when the second retailer identifier does not match the first retailer identifier and the anomaly score is below a predetermined threshold;
in response to determining the anomaly score is equal to or above the predetermined threshold, transmitting, by the processor, an electronic communication identifying the first item data and the second item data; and
training, by the processor, an additional machine learning model to detect anomalies in the hierarchical structure of the database, wherein the additional machine learning model is trained by a supervised training dataset based on attributes of items in the plurality of items and at least the second supplier-item association.
8. The method of claim 7 , further comprising:
determining, by the processor, that the plurality of group-item associations does not include a group-item association between the first group data and the first item data; and
generating, by the processor, the group-item association between the first supplier identifier-and the first group data.
9. The method of claim 7 , wherein the trained machine learning model is configured to:
determine a score associated with each of a plurality of anomalies; and
compare the score associated with each of the plurality of anomalies to the predetermined threshold wherein the second supplier-item association is generated when the score associated with each of the plurality of anomalies is below the predetermined threshold.
10. The method of claim 7 wherein:
the first retail numbering system associates a plurality of supplier identifiers to a first plurality of group identifiers and a plurality of retailer identifiers; and
the second retail numbering system associates a second plurality of group identifiers and at least a portion of the plurality of supplier identifiers.
11. The method of claim 7 , comprising
determining, by the processor, at least a first attribute for the first item;
determining, by the processor, at least a second attribute of a second item corresponding to a second supplier identifier and a third attribute of a third item corresponding to a third supplier identifier; and
generating, by the processor, the second supplier-item association when a first number of anomalies based on the at least first attribute and the at least second attribute and a second number of anomalies based on the at least first attribute and the at least third attribute are each below the predetermined threshold.
12. The method of claim 7 , wherein each of the at least one-attribute of the first item data and the at least one attribute of the second item data comprise item descriptions.