WE CLAIM:
1. A method for identifying high potential products for a digital platform,
the method comprising:
- retrieving, by a processing unit [102] from a memory unit [108], a product data of one or more products, wherein the product data comprises a set of product attributes, and a set of image embeddings of the one or more products;
- processing, by the processing unit [102], the product data of the one or more products to generate a processed product data, wherein the processed product data comprises a processed set of product attributes, and a processed set of image embeddings;
- receiving, at a prediction unit [104], the processed product data, wherein the prediction unit [104] is a fine-tuned prediction unit;
- predicting, by the fine-tuned prediction unit [104], a potential score for each product of the one or more products, based on the processed product data; and
- identifying, by the processing unit [102], a set of high potential products based on the potential score for each product of the one or more products, and a potential score threshold value.
2. The method as claimed in claim 1, wherein:
the one or more products belong to a target category,
the set of product attributes comprises value for each attribute of
a plurality of attributes for the one or more products, and
the set of image embeddings comprises an embedding for each
image for the one or more products.
3. The method as claimed in claim 1, the method further comprising:
- generating, by a filtering unit [106], a filtered set of high potential
products from the set of high potential products, based on one or
more first parameters, and a first threshold;

- ranking, by a ranking unit [112], products of the filtered set of high potential products;
- displaying, via a user interface [110], the products of the filtered set of high potential products based on the ranking.
4. The method as claimed in claim 1, wherein the prediction unit [104] is
fine-tuned based on:
- receiving a training product data, wherein the training product data
comprises a training set of product attributes, a training set of image
embeddings, and a training product details of one or more training
products,
wherein
the one or more training products belong to the target category, the set of training product attributes comprises value for each
attribute of a plurality of attributes for the one or more training
products, and
the set of training image embeddings comprises an embedding for
each image of the one or more training products;
- splitting the training product data into a training dataset and a validation dataset, based on a predefined ratio; and
- fine-tuning the prediction unit [104] based on the training dataset and the validation dataset.

5. The method as claimed in claim 4, wherein the fine-tuned prediction unit [104] comprises a fine-tuned catboost model.
6. The method as claimed in claim 4, wherein the training product details comprises one or more of a historical sales number, an impressions number, and a product page visits number for each product of the one or more training products.
7. The method as claimed in claim 4, the method further comprising:
- identifying a set of relevant attributes using the fine-tuned prediction

unit [104].
8. The method as claimed in claim 7, the method further comprising:
- performing a target action based on the set of relevant attributes.
9. The method as claimed in claim 1, the method further comprising:
- receiving, by the processing unit [102], the set of high potential products, and a set of user cohorts based on one or more user characteristics, a clicks and conversions data for each user cohort;
- splitting, by the processing unit [102], the set of high potential products into one or more subsets of high potential products based on one or more second parameters;
- generating, by the processing unit [102], a rewards data based on the set of user cohorts, the clicks and conversions data, and the one or more subsets of high potential products; and
- displaying, via a user interface [110], a set of relevant products to users of the user cohorts based on the rewards data, the set of user cohorts, and the one or more subsets of high potential products.
10. A system for identifying high potential products for a digital platform, the
system comprising:
- a processing unit [102] configured to:
o retrieve from a memory unit [108], a product data of one or more products, wherein the product data comprises a set of product attributes, and a set of image embeddings of the one or more products;
o process the product data of the one or more products to
generate a processed product data, wherein the processed product data comprises a processed set of product attributes, and a processed set of image embeddings; and
- a prediction unit [104] configured to:
o receive the processed product data, wherein the prediction

unit [104] is a fine-tuned prediction unit; o predict a potential score for each product of the one or more
products, based on the processed product data; wherein the processing unit [102] is further configured to identify a set of high potential products based on the potential score for each product of the one or more products, and a potential score threshold value.
11. The system as claimed in claim 10, wherein:
the one or more products belong to a target category,
the set of product attributes comprises value for each attribute of
a plurality of attributes for the one or more products, and
the set of image embeddings comprises an embedding for each
image for the one or more products.
12. The system as claimed in claim 10, the system further comprising:
- a filtering unit [106] configured to generate a filtered set of high potential products from the set of high potential products, based on one or more first parameters, and a first threshold;
- a ranking unit [112] configured to rank products of the filtered set of high potential products;
- a user interface [110] configured to display the products of the filtered set of high potential products based on the ranking.
13. The system as claimed in claim 10, wherein the prediction unit [104], for
the fine-tuning, is configured to:
- receive a training product data, wherein the training product data
comprises a training set of product attributes, a training set of image
embeddings, and a training product details of one or more training
products,
wherein
the one or more training products belong to the target category,

the set of training product attributes comprises value for each attribute of a plurality of attributes for the one or more training products, and
the set of training image embeddings comprises an embedding for each image of the one or more training products;
- split the training product data into a training dataset and a validation dataset, based on a predefined ratio; and
- fine-tune the prediction unit [104], based on the training dataset and the validation dataset.

14. The system as claimed in claim 13, wherein the fine-tuned prediction unit [104] comprises a fine-tuned catboost model.
15. The system as claimed in claim 13, wherein the training product details comprises one or more of a historical sales number, an impressions number, and a product page visits number for each product of the one or more training products.
16. The system as claimed in claim 13, wherein the fine-tuned prediction unit [104] is further configured to:
- identify a set of relevant attributes.
17. The system as claimed in claim 16, wherein the fine-tuned prediction unit
[104] is further configured to:
- perform a target action based on the set of relevant attributes.
18. The system as claimed in claim 10, wherein:
- the processing unit [102] is further configured to:
o receive the set of high potential products, and a set of user
cohorts based on one or more user characteristics, a clicks and
conversions data for each user cohort; o split the set of high potential products into one or more
subsets of high potential products based on one or more
second parameters;

o generate a rewards data based on the set of user cohorts, the clicks and conversions data, and the one or more subsets of high potential products; and - a user interface [110] is configured to:
o display a set of relevant products to users of the user cohorts based on the rewards data, the set of user cohorts, and the one or more subsets of high potential products.