FIELD OF THE DISCLOSURE

The present disclosure relates generally to the field of rating systems for products on digital platforms. More particularly, the disclosure relates to methods and systems for providing quality product listings to users on digital platforms.

BACKGROUND

The following description of related art is intended to provide background information pertaining to the field of the disclosure. This section may include certain aspects of the art that may be related to various features of the present disclosure. However, it should be appreciated that this section be used only to enhance the understanding of the reader with respect to the present disclosure, and not as admissions of prior art.

With the advent of increasing reach of online digital platforms such as ecommerce platforms, and the changing dynamics of user base such as requirements of customers and sellers on these digital platforms, it is important to satisfy the requirements of customers as well as sellers. The customers want to look at the best results of product listings in their feed and the sellers want their products/listings to be shown higher in the list of recommendations shown to the customers. The increased usage of digital platforms for shopping purposes by customers has also led to an enormous rise in the number of sellers selling on these platforms. In this case, one product that is manufactured by the manufacturer is sold by multiple sellers on the platform. So, a lay man user might not be able to figure out the difference in the experience that he/she might have on the same product sold by different sellers on the platform. Some of the sellers provide satisfactory services to the customers, while many others do not. Now, if the sellers who provide unsatisfactory services to the customers are displayed at priority in the listings shown to the users, the users might have bad experience which will affect the business of the digital platforms. Further, if the listings of sellers providing good services are listed at lower ranks in the results shown to
the users, the good sellers will not get optimum orders which they deserve and thus will not be motivated to provide better services to customers. This is more so possible in case if a new seller enters on the digital platforms market. For example, if a new seller is providing good products and services and starts selling his products on the digital platform, the listings of products created by this new seller might not be shown at priority in the results shown to the users as these listings do not have sufficient user ratings or feedback. Now, since more and more sellers are already selling the same products on the digital platforms, it is getting even more difficult for the new seller to display his listing at priority despite providing good products and services.

Further, quality of a products is one of the key considerations that users have while making a purchase decision. They can estimate the quality of a product by its touch and feel during offline shopping, however, in e-commerce, users do not have the ability to touch-and-feel products before purchasing. Therefore, it is important to rate product listings to provide best product listings to the users at priority for the benefit of both customers as well as sellers.

There are a variety of existing methods and mechanisms by which digital platforms can estimate better quality products being sold on digital platforms. These existing solutions take into account data related to various parameters. For example, the net promoter score (NPS) system, which measures customer experience with, say, product and seller, and predicts business growth based on the same. However, this approach and other existing approaches are possible when the data on product listings is available beforehand, and not applicable when no or limited data is available for the same, i.e., during cold-start of a product listing created on the digital platform.

Thus, there exists an imperative need in the art to provide a system and method for first estimating quality of product listings when atleast no or limited data is available and providing quality product listings to users on digital platforms. This will help especially in saving a lot of time of consumers on the
digital platforms that might be spent on researching for a good quality product listing or the time spent on ordering a product and then returning it due to dissatisfaction, and will improve their experience. Also, it will help in providing product listings that are of good quality and are available on the digital platforms for the users, but might not otherwise be shown to them at priority.

SUMMARY

This section is intended to introduce certain objects and aspects of the disclosed method and system in a simplified form and is not intended to identify the key advantages or features of the present disclosure.

Accordingly, a first object of the present disclosure is to obtain a method and system for providing quality product listings to users on a digital platform that overcomes the limitations of the existing approaches. Another object of the present disclosure is to obtain a method and system that provides ratings for the product listings that do not have sufficient feedback data from users. Yet another object of the present disclosure is to obtain a method and system that predicts more accurate ratings of product listings being based on higher number of inputs, such as seller quality score based on various parameters related to seller quality (for example, time to deliver, cancellations done by the seller, etc.), return rate of a product listing, average rating of a product listing, etc. Yet another object of the present disclosure is to obtain a method and system that is able to provide new product listings that are of good quality to the users at priority irrespective of the fact that sufficient data is not available to calculate the actual rating of the product listing on the digital platform.

One aspect of the present disclosure relates to a system for providing quality product listings to users on a digital platform. The system comprises a processing unit configured to retrieve a set of seller ratings, a set of one or more mature listings, and a set of one or more non-mature listings from a memory unit, a set of rating indices from a rating index unit, and a set of return indices from a returns index unit. Further, the processing unit is configured to generate a
set of one or more listing quality scores for each of the one or more non-mature listings based on the set of seller ratings, the set of rating indices retrieved from the rating index unit, and the set of return indices retrieved from the returns index unit. Further, the processing unit is configured to determine a set of listing bands comprising a listing band for each of the one or more non-mature listings and the one or more mature listings. This set of listing bands is determined based on one or more pre-defined rules. Further, the system comprises a user interface unit coupled with the processing unit. The user interface unit is configured to provide quality product listings on the digital platform based on the set of listing bands.

Another aspect of the present disclosure relates to a method for providing quality product listings to users on a digital platform. The method comprises retrieving, by a processing unit, a set of seller ratings, a set of one or more mature listings and a set of one or more non-mature listings from a memory unit, a set of rating indices from a rating index unit, and a set of return indices from a returns index unit. Further, the method comprises generating, by the processing unit, a set of one or more listing quality scores for each of the one or more non-mature listings based on the set of seller ratings, the set of rating indices, and the set of return indices. Thereafter, the method comprises determining, by the processing unit, a set of listing bands comprising a listing band for each of the one or more non-mature listings and the one or more mature listings. This set of listing bands is determined based on one or more predefined rules. Finally, the method comprises providing, by a user interface unit, quality product listings on the digital platform based on the set of listing bands. BRIEF DESCRIPTION OF DRAWINGS

The accompanying drawings, which are incorporated herein, and constitute a part of this disclosure, illustrate exemplary embodiments of the disclosed methods and systems in which like reference numerals refer to the same parts throughout the different drawings. Components in the drawings are
not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the present disclosure. Some drawings may indicate the components using block diagrams and may not represent the internal circuitry of each component. It will be appreciated by those skilled in the art that disclosure of such drawings includes disclosure of electrical components, electronic components or circuitry commonly used to implement such components.

FIG.1 illustrates an architecture of a system for providing quality product listings to users on a digital platform, in accordance with exemplary embodiments of the present disclosure.

FIG.2A illustrates an architecture of a rating index unit forming part of a system for providing quality product listings to users on a digital platform, in accordance with exemplary embodiments of the present disclosure.

FIG.2B illustrates an architecture of a returns index unit forming part of a system for providing quality product listings to users on a digital platform, in accordance with exemplary embodiments of the present disclosure.

FIG.3 illustrates an exemplary method flow diagram depicting a method for providing quality product listings to users on a digital platform, in accordance with exemplary embodiments of the present disclosure.

The foregoing shall be more apparent from the following more detailed description of the disclosure.

DETAILED DESCRIPTION

In the following description, for the purposes of explanation, various specific details are set forth in order to provide a thorough understanding of embodiments of the present disclosure. It will be apparent, however, that embodiments of the present disclosure may be practiced without these specific details. Several features described hereafter can each be used independently of one another or with any combination of other features. An individual feature may not address any of the problems discussed above or might address only some of the problems discussed above. Some of the problems discussed above
might not be fully addressed by any of the features described herein. Example embodiments of the present disclosure are described below, as illustrated in various drawings in which like reference numerals refer to the same parts throughout the different drawings.

The ensuing description provides exemplary embodiments only, and is not intended to limit the scope, applicability, or configuration of the disclosure. Rather, the ensuing description of the exemplary embodiments will provide those skilled in the art with an enabling description for implementing an exemplary embodiment. It should be understood that various changes may be made in the function and arrangement of elements without departing from the spirit and scope of the disclosure as set forth.

Specific details are given in the following description to provide a thorough understanding of the embodiments. However, it will be understood by one of ordinary skill in the art that the embodiments may be practiced without these specific details. For example, circuits, systems, networks, processes, and other components may be shown as components in block diagram form in order not to obscure the embodiments in unnecessary detail. In other instances, well-known circuits, processes, algorithms, structures, and techniques may be shown without unnecessary detail in order to avoid obscuring the embodiments.

Also, it is noted that individual embodiments may be described as a process which is depicted as a flowchart, a flow diagram, a data flow diagram, a structure diagram, or a block diagram. Although a flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations may be rearranged. A process is terminated when its operations are completed but could have additional steps not included in a figure.

The word "exemplary" and/or "demonstrative" is used herein to mean serving as an example, instance, or illustration. For the avoidance of doubt, the subject matter disclosed herein is not limited by such examples. In addition, any
aspect or design described herein as "exemplary" and/or "demonstrative" is not necessarily to be construed as preferred or advantageous over other aspects or designs, nor is it meant to preclude equivalent exemplary structures and techniques known to those of ordinary skill in the art. Furthermore, to the extent that the terms "includes," "has," "contains," and other similar words are used in either the detailed description or the claims, such terms are intended to be inclusive—in a manner similar to the term "comprising" as an open transition word—without precluding any additional or other elements.

As used herein, a "processor" or "processing unit" includes processing unit, wherein processor refers to any logic circuitry for processing instructions. A processor may be a general-purpose processor, a special purpose processor, a conventional processor, a digital signal processor, a plurality of microprocessors, one or more microprocessors in association with a DSP core, a controller, a microcontroller, Application Specific Integrated Circuits, Field Programmable Gate Array circuits, any other type of integrated circuits, etc. The processor may perform signal coding data processing, input/output processing, and/or any other functionality that enables the working of the system according to the present disclosure.

As used herein, a "set" refers to a group of one or more units. A set of ratings may comprise one or more ratings, a set of listings may comprise one listing or more than one listings. The term "set" refers to one or more units unless indicated otherwise in the specification. For example, a set of seller ratings means a set of one or more seller ratings, and a set of two or more product listings means this set of product listings comprises atleast 2 product listings. Also, a set may not comprise physical products/entities only. The disclosure also encompasses non-tangible entities included in a set.

The present disclosure provides method for providing quality product listings to users on a digital platform, especially useful on ecommerce platforms. Here, product listing refers to a product that is listed on an ecommerce
platform/website by a particular seller. Since a product manufactured by a manufacturer may be sold on an ecommerce platform by multiple sellers, one product can have multiple listings on the same platform. For various reasons, for example, availability of several products that are not available in the offline market, competitive prices of products, ease of ordering and receiving the products at home, etc., the number of users on ecommerce platforms are increasing day by day. Due to this more and more sellers and manufacturers are starting to sell their products on digital platforms. Therefore a large number of sellers are already selling on the digital platforms, and many more are starting to associate with the digital platforms. Now, a main concern of customers on the digital platforms is about the quality of products which they assess using the rating of products available on these platforms. Products may be shown on the digital platforms in accordance with their ratings that are provided by the users. And, the products are rated by the users based on their experience when they order them on the digital platforms. So, users tend to order and purchase mostly the higher rated products and again the higher rated product listings will get further ratings and reviews from the users and therefore are shown at priority on the digital platform. But there is no solution for a new product listing that is created by a seller. A new product listing maybe of a very good quality but is not shown at priority on the digital platform only because it does not have sufficient rating and reviews. A good quality product listing is not rated atleast in its initial period after creation as the system of the digital platform does not have any mechanism that automatically generates a rating/score for the product listing without user inputs such as through user feedback and surveys. The prior existing solutions such as Net Promoter score (NPS) rating provide rating of product listings based on user reviews and various other factors. However, these existing solutions are not very helpful in the cases atleast where a new product is listed on the digital platform, or a new product listing is created buy a new seller. This is because the product listings for being shown at priority in the list on the
digital platform need to be mature enough, and the new product listings that are created are always non-mature product listings.

As used herein, mature product listings refers to the product listings that have been ordered and/or returned and/or reviewed by the customers at least a minimum threshold number of times. For example, a product listing may be considered as a mature product listing if it has received atleast 20 user ratings on the digital platform. A person skilled in the art would appreciate that the value of 20 user ratings is only exemplary and does not restrict the disclosure in any possible manner. And, non-mature product listings refers to the product listings that have not been ordered and/or rated by the customers at least a minimum threshold number of times. The present disclosure provides a solution to this problem by using the ratings of matured product listings in a category of products to predict the ratings for non-matured product listings. In this way, users will be able to see non-matured product listings of good quality at priority on digital platforms and will save a lot of time by not needing to search the same. Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the solution provided by the present disclosure.

Figure 1 illustrates an architecture of a system for providing quality product listings to users on a digital platform. As shown, the system [100] comprises a processing unit [102], a memory unit [104], a rating index unit [106], a returns index unit [108], a user interface unit [110]. All the components of the system [100] should be construed to be operably connected to each other unless indicated in the disclosure. The system [100] based on the interconnection of and interworking of the various components thereof, is configured to provide quality product listings to users on a digital platform.

The processing unit [102] is configured to retrieve a set of seller ratings, a set of one or more mature listings, and a set of one or more non-mature listings from a memory unit [104], a set of rating indices from a rating index unit [106],
and a set of return indices from a returns index unit [108]. In an implementation, the processing unit [102] retrieves the data of a pre-defined period of time, say, of last 365 days, etc.

The set of seller ratings comprises ratings provided by users to the sellers that are selling on the digital platform, also called as seller quality score or SQS score. A seller on the platform can be an old seller or a new seller. An old seller is the one who has been associated with the digital platform by listings its products on the platform, and is therefore expected to have received a number of reviews and ratings from the users on the platform. These reviews/ratings may be present in the form of an SQS score and saved in the memory unit [104].

Similarly, a product listing can be an old product listing that has been previously sold and/or is being sold on the digital platform for quite some time and therefore has rating and reviews from users on the platform. Also, the same product can be sold by multiple sellers on the platform, and therefore as used herein, a product listing refers to a product that is sold on the platform by a particular seller, i.e., the product listing comprises a product and seller combination. Further, this product listing can be a mature listing or a non-mature listing. A mature listing refers to the product listing that has been ordered and/or purchased and/or reviewed by the customers at least a minimum threshold number of times. And, a non-mature listing refers to the product listing that has not been ordered and/or rated by the customers at least a minimum threshold number of times. For example, a particular product is listed on the digital platform and is ordered by some customers. The customers provide reviews and ratings to the product that they ordered. Receiving the ratings and reviews from the customers, the product listing gradually becomes mature when the number of ratings/ reviews/ orders made by customers reaches a minimum threshold criteria, such as, a criteria may be of 20 orders placed by customers on the platform. So, as soon as 20 orders of the product are placed by customers on the platform, the product listing becomes a mature listing on the platform. In the
above example, till the point when users have not placed atleast 20 orders on the platform for the same product listing, the product listing is referred to as a non-mature listing. A person skilled in the art would appreciate that the threshold of 20 orders is exemplary and only for understanding purposes, and does not restrict the present disclosure in any possible manner.

Also, the rating index unit [106] is configured to generate a set of rating indices. The rating index of a product indicates the quality of a product listing. The rating index is a relative measure, an indication of the product listing's rating in comparison with other similar products. A higher rating index means that it has been rated higher by the users, and implies a better quality of the product listing. The rating index is determined by the rating index unit [106] based on the equation (1) given below.

Rating index of the listing = —Rating of the — ... (1)

Median rating of the cohort

For this purpose of generating rating indices of various products, the processing unit [102] first generates a median rating of various products in a cohort which comprises products similar to the product listing of which the rating index is being calculated. The processing unit [102] sends this information to the rating index unit [106] which uses this information received from the processing unit [102] to calculate a rating index for the product listing. Similarly, the returns index unit [108] is configured to generate a set of return indices. The returns index is a relative measure, an indication of the listing's returns in comparison with other similar products. A lower returns index of a product listing means the product has been returned less by the customers, implying that the customers are satisfied with the product and hence the product is of good quality. The returns index is determined by the returns index unit [108] based on the equation (2) given below.

Return index of the listing = — y g—■— ——g— ... (2)
Here, recency weighted returns of the product listing are the returns of the product listing that are weighted on the basis of how many products have been returned by the customers recently. For example, the product listing has been frequently returned by the customers, then it is a problem, but if the product listing was returned earlier and there are no returns of the products recently, it may imply that the seller has improved upon the quality of the product. Also for this purpose of generating returns indices of various products, the processing unit [102] first generates a median recency weighted returns of various products in a cohort which comprises products similar to the product listing of which the returns index is being calculated. The processing unit [102] sends this information to the returns index unit [108] which uses this information received from the processing unit [102] to calculate a returns index for the product listing.

Further, the processing unit [102] generates a set of one or more listing quality scores (LQS scores) for each of the one or more non-mature listings based on the set of seller ratings, the set of rating indices retrieved from the rating index unit [106], and the set of return indices retrieved from the returns index unit [108]. These listing quality scores represent the actual quality of the product listings that are shown to the users. The quality of a product may be good or bad and may be measured using various factors such as attributes of performance or the core functionality of the product, features, build quality of the product, compliance to standards, durability, serviceability, etc. In several cases, the quality of the product may also depend on the seller who is selling it. Similarly, the services of a seller might also be good or bad and may be measured using product availability, trustworthiness, fulfilment guarantees, efficiency of return process, comparative pricing, etc. The listing quality score takes into account the combination of all such factors through the seller quality score, rating index and return index of a product listing. The LQS is used as a product quality indicator for individual listings. Accordingly, the listing quality score can be represented as given in equation (3) below:
12*Rating index+5* Return index+2*SOS score

~ ' 12+5+2 .... (3)

As used in the equation (3) above, "rating index" is the ratio of the average ratings of a product listing to the median average rating of the cohort of products belonging to the same category/vertical and/or having similar specifictaions and/or selling in a similar price range of the product for which the rating index is computed, as given in equation (1). For example, a product "Somsunk Mobile A7" is newly listed by a seller for which the system [100] of the present disclosure has to predict listing quality score. For this product, the cohort may comprise all the mobile phones of 'Somsunk' brand. Alternatively, a cohort for this purpose may comprise all mobile phones of all brands that are or have previously been listed on the digital platform, or the cohort may comprise the portable electronic devices that may be used for communication and/or are similar to mobile phones such as tablets, etc. Also, in an implementation, a cohort may comprise similar products of similar price range, for example, all mobile phones of 'Somsunk' brand within a price range of 'Rs. 10,000-15,000'. A person skilled in the art would appreciate that the above example of mobile phone brand and price are only exemplary and do not restrict the invention in any possible manner. Thus, the rating index is calculated for a product listing based on the product listings in a particular category. For example, a product listing has a rating of 3.6, and maps to a cohort with a median rating of 3.9. This product listing will have a rating Index of 0.92 (i.e., 3.6/3.9). Further, the "returns index" is the inverse of the ratio of the return rate of a product listing to the median average rating of the cohort of similar products, for example, products belonging to the same category/vertical of the product and/or having similar specifications, and/or products belonging to the same category/vertical of the product and falling within a particular price range, for which the returns index is computed, as given in equation (2). And, SQS is determined based on inputs such as seller cancellation (i.e., cancellation of an order done by the seller after receiving an order from the user) and ready to deliver (RTD) breaches (i.e.,
breach of a promise made by the seller to deliver a product on time to the customer or to warehouse for further delivery logistics or to the pick-up delivery person responsible to deliver the product from the seller to the customer - in either case, the product may not reach the customer on time) and similar seller metrics in a category. LQS for each listing is calculated as a linear combination of the above three components. Also, weights for each of the above 3 components are pre-defined weights determined to maximise the correlation of final LQS with net promoter score (NPS) which is one of the actual scores as rated by the users. Further, the processing unit [102] determines a set of listing bands comprising a listing band for each of the one or more non-mature listings and the one or more mature listings. Thus, the continuous-valued LQS score is further divided into 10 LQS bands {0, 1, 2,...,9}. The bands categorise product listings based on their quality. For example, the LQS band 0 (or band 0) may contain the best quality product listings, and band 9 may contain the worst quality product listings. Also, the bands are created based on cutoffs on the listing score at a category level. In other words, the set of listing bands is determined based on one or more predefined rules. In an exemplary implementation, the listing band cutoffs are determined to maintain a fixed percentage of delivered units in each band and ensure high quality listings to be discovered. For example, the cutoff for band 0 is calculated in such a way that 33% of delivered units will fall in there. This ensures that there will be enough products to show in the results page from a particular category for most of the queries without having to display less relevant but higher quality products from other categories. In an exemplary implementation, the rule for listing band cutoffs may be applied as:

• LQS Band 0 comprises listings contributing to top 30% of units

• LQS Band 1 comprises listings contributing to further 25% of units

• LQS Band 2 comprises listings contributing to further 20% of units

• LQS Band 3 comprises listings contributing to further 10% of units

• LQS Band 4 - 9 comprise listings contributing to bottom 15% of units
Further, a user interface unit [110] coupled with the processing unit [102] is configured to provide quality product listings on the digital platform based on the set of listing bands. As used herein, the user interface unit [110] includes an output device in the form of a display, such as a liquid crystal display (LCD), cathode ray tube (CRT) monitors, light emitting diode (LED) screens, etc. and/or one or more input devices such as touchpads or touchscreens. The display may be a part of a portable electronic device such as smartphones, tablets, mobile phones, wearable devices, etc. They also include monitors or LED/LCD screens, television screens, etc. that may not be portable. The display is typically configured to provide visual information such as text and graphics. An input device is typically configured to perform operations such as issuing commands, selecting, and moving a cursor or selector in an electronic device. Also, "providing quality product listings" includes determining the product listings for displaying to the users and/or displaying the quality product listings to users on the digital platform.

Referring to Figure 2A, which illustrates an architecture of the rating index unit [106] forming part of a system [100] for providing quality product listings to users on a digital platform, in accordance with exemplary embodiments of the present disclosure. As shown in Figure 2A, the rating index unit [106] comprises a pre-trained first prediction unit [202] and a first smoothing unit [204]. The pre-trained first prediction unit [202] is configured to generate a set of predicted ratings. This set of predicted ratings is generated prior to the processing unit [102] retrieving the set of rating indices from the rating index unit [106].

Also, various metrics such as Mean Absolute Error (MAE) are used for measuring accuracy of the predicted ratings. In an implementation, the prediction unit [202] does not generate accurate predicted ratings due to some reason such as in some exceptional conditions, for example, a system downtime, or the listing level data of the mature listings is not available. In such case, a
rolled-up approach is implemented by the first prediction unit [202]. In this approach, the first prediction unit [202] checks the availability of the listing level data of mature listings. If the listing level data is not available, then the first prediction unit [202] is configured to check the data of higher level such as product level data, and if that is also not available, then brand and/or category level data, and so on. For example, a new product listing of smartphone "xsmartphone 8" is created. And, the first prediction unit [202] does not have any data of "xsmartphone 8". In this case, the first prediction unit [202] will check the data available for "xsmartphone 7", "xsmartphone 6" and then all other smartphones from this brand and then all smartphones to predict the ratings.

In an implementation, the first prediction unit [202] is a pre-trained unit trained based on a machine learning model. For generating the predicted ratings for the non-mature product listings, the first prediction unit [202] undergoes pretraining or learning process. In this process, the first prediction unit [202] retrieves a first set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of average ratings for the one or more mature listings from the memory unit [108]. Further, the first prediction unit [202] comprises a first set of one or more learning functions comprising a learning function for each of the one or more categories of products. The first prediction unit [202] includes learning functions or learning models for various categories of products known as verticals. These learning functions are trained using various input features. The input features may be categorised as product attributes such as brand, material, size, weight, memory size, etc., as seller attributes such as seller score, cancellation score, delivery breaches, etc., as listing attributes such as selling price, maximum rerail price, etc. among others. For example, for a vertical/category of products, say, "shoes", input features such as "material", "weight", "brand", "selling price" are valid input features, and "memory size"
would be an invalid input feature as shoes don't have a memory size such as that in a mobile phone. Therefore, for a particular vertical, the associated learning functions would be trained using the values of data in some valid input features that are associated with that vertical.

In the process of training the first prediction unit [202], the set of mature product listings is provided to the first prediction unit [202]. The set of mature product listings comprises data related to the rating of the product listings such as the value of the input features, and rating of the product and seller provided by the old users, etc. Based on the available data, the first prediction unit [202] maps average ratings for the mature listings with each input attribute/feature in the first set of one or more input attributes based on the one or more learning functions. Also, the first set of one or more input attributes refers to the input attributes that are related to and/or used by the first prediction unit [202]. Also, the first prediction unit [202] is trained to update the learning functions based on the data provided in various time intervals. Each learning function is trained based on a set of mature product listings in a particular category of products.

The first prediction unit [202] is also trained to identify the learning function to be applied on a particular product. For example, a non-mature product listing belonging to the "shoes" category is supplied to the first prediction unit [202]. So, the first prediction unit [202] itself determines the learning function to be applied on the shoes category. Accordingly, the learning function is capable of predicting a rating to the non-mature product listing of the shoe using the input attributes of the non-mature product listing as the learning function has already learnt based on the same and/or similar input attributes of the "shoes" category/vertical. For example, a learning function is based on input features/attributes such as 'seller rating = 4', 'brand = proma', 'colour = red', and 'material = mesh'. And for such input features, the learning function is trained to generate a 'predicted rating = 4'. Now, say a different seller with a 'seller rating = 2' creates a new product listing of the similar shoe of 'brand = proma', 'colour =
blue', and 'material = mesh'. Assuming in this example that colour of the shoe may not affect the rating of product, and since the seller rating in this case for non-mature product listing is lower than the mature product listing, the first prediction unit [202] may generate a 'predicted rating = 3' for this non-mature product listing. A person skilled in the art would appreciate that the values and attributes provided in the above example are exemplary and for understanding purposes only, and do not restrict the invention in any possible manner.

However, the above predicted ratings maybe noisy and unstable, as the ratings are predicted based on all the data that is available, for example, the ratings available for the mature product listings as well as the limited data available for non-mature product listings. Therefore, the average rating predictions from the learning models are used and the available values are used to smooth out the predicted ratings. This enables the first prediction unit [202] to efficiently make use of all available data for making the predictions. Thus, the first smoothing unit [204] is configured to generate a set of smooth ratings based on the set of predicted ratings, a total number of ratings, an average rating, and a hyperparameter. The smoothening function for rating, in an implementation, may be represented as given in equation (4) below:

Ik*predicted rating 1

[ + num_ratings * auerage_rating ]

Smoothened_Rating =

k + mtm_ratings .... (4)

As used in the equation (4) above, "k" refers to a hyperparameter, which is used to reduce the mean absolute error in the results, "num_ratings" refers to the total number of ratings in the vertical, and "average_rating" refers to the median average rating of the vertical. Also, this first smoothing unit [204] helps in normalising the data in case where the number of data points (i.e., ratings) is less. For example, there are a number of predicted ratings and a number of actual ratings that is less than the required threshold of data points to be used as actual data of a product listing. In this case, the first smoothing unit [204] is
configured to use the available ratings along with the predicted ratings, and also decide what weightage is to be given to the predicted ratings and the actual ratings available. The smoothened ratings are then sent from the first smoothing unit [204] to the rating index unit [106]. The rating index unit [106] is further configured to determine the set of rating indices for the one or more non-mature listings based on the set of smooth ratings. The rating index is the ratio of the average ratings of a listing to the median average rating of the cohort, where a cohort is defined as product vertical x price bucket, that is, the group of products belonging to a same category/vertical of products having similar prices or falling within a particular price range. Also, the ratings indices consider behaviour of the users, for example, different users have different behaviour towards different price points on the same product. In an implementation, the rating index of a product listing is determined as (Rating of the listing / Median rating of cohort).. Finally, these rating indices are used as an input to calculate the LQS score for the non-mature product listing. Thus, the set of rating indices comprises a rating index for each non-mature product listing in a category based on an average rating of a set of mature product listings in the category.

Now, referring to Figure 2B which illustrates an architecture of a returns index unit [108] forming part of a system for providing quality product listings to users on a digital platform, in accordance with exemplary embodiments of the present disclosure. As shown in Figure 2B, the returns index unit [108] comprises a pre-trained second prediction unit [212] and a second smoothing unit [214]. The pre-trained second prediction unit [212] is configured to generate a set of predicted returns. This set of predicted returns is generated prior to the processing unit [102] retrieving the set of rating indices from the returns index unit [108]. Also, the set of predicted returns may be generated simultaneously along with generating the set of predicted ratings by the first prediction unit [202], with or without a time difference. In other words, the steps of generating predicted ratings and predicted returns or the steps of generating rating indices
or return indices are not sequential in nature and may happen simultaneously in the system [100].

Further, in an implementation, the second prediction unit [212] is a pretrained unit trained based on a machine learning model. For generating the predicted returns for the non-mature product listings, the second prediction unit [212] undergoes pre-training or learning process. In this process, the second prediction unit [212] retrieves a second set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of average ratings for the one or more mature listings from the memory unit [108]. Further, the second prediction unit [212] comprises a second set of one or more learning functions comprising a learning function for each of the one or more categories of products. The second prediction unit [212] includes learning functions or learning models for various categories of products/ verticals. These learning functions are trained using various input features. The input features may be categorised as product attributes such as brand, material, size, weight, memory size, etc., as seller attributes such as seller score, cancellation score, delivery breaches, etc., as listing attributes such as selling price, maximum rerail price, etc. among others. For example, for a vertical/category of products, say, "shoes", input features such as "material", "weight", "brand", "selling price" are valid input features, and "memory size" would be an invalid input feature as shoes don't have a memory size such as that in a mobile phone. Therefore, for a particular vertical, the associated learning functions would be trained using the values of data in some valid input features that are associated with that vertical.

In the process of training the second prediction unit [212], the set of mature product listings is provided to the second prediction unit [212]. The set of mature product listings comprises data related to the rating of the product listings such as the value of the input features, and rating of the product and seller provided by the old users, etc. Based on the available data, the second
prediction unit [212] maps return ratios for the mature listings with each input attribute/feature in the second set of one or more input attributes based on the one or more learning functions. Also, the second set of one or more input attributes refers to the input attributes that are related to and/or used by the second prediction unit [202]. Also, the second prediction unit [212] is trained to update the associated learning functions based on the data provided in various time intervals. Each learning function is trained based on a set of mature product listings in a particular category of products.

The second prediction unit [212] is also trained to identify the learning function to be applied on a particular product. For example, a non-mature product listing belonging to the "shoes" category is supplied to the first prediction unit [202]. So, the second prediction unit [212] itself figures out the learning function to be applied on the shoes category. Accordingly, the learning function is capable of predicting a return to the non-mature product listing of the shoe using the input attributes of the non-mature product listing as the learning function has already learnt based on the same and/or similar input attributes of the "shoes" category/vertical. However, these predicted returns maybe noisy and unstable, as the returns are predicted based on all the data that is available, for example, the number of returns of the same product made earlier, total number of orders made by users, and other data available for the mature product listings as well as the limited data available for non-mature product listings. Therefore, the average returns ratio predictions from the learning models are used and the available values are used to smooth out the predicted ratings. This enables the second prediction unit [212] to efficiently make use of all available data for making the predictions. Thus, the second smoothing unit [214] is configured to generate a set of smooth returns based on the set of predicted returns, a total number of delivered units of products, an average returns ratio and a hyperparameter. The smoothening function for returns, in an implementation, may be represented as given in equation (5) below:
j k * predictedjreturns
1 + num_delivered_units * average_returns_ratio J

Smoothened_Returns = - — -

k + num_delivered_unit$ .... (5)

As used in the equation (5) above, "k" refers to a hyperparameter, which is used to reduce the mean absolute error in the results, "num_delivered_units" refers to the total number of delivered units in the vertical, and "average_returns_ratio" refers to the median average returns ratio of the products in the vertical. Also, the second smoothing unit [214] helps in normalising the data in case where the number of data points (i.e., returns) is less. For example, there are a number of predicted returns and a number of actual returns that is less than the required threshold of data points to be used as actual data of a product listing. In this case, the second smoothing unit [214] is configured to use the available returns along with the predicted returns, and also decide what weightage is to be given to the predicted returns and the actual returns data available. The smoothened ratings are then sent from second smoothing unit [214] to returns index unit [108]. The returns index unit [108] is further configured to determine the set of return indices for the one or more non-mature listings based on the set of smooth ratings from the second smoothing unit [214]. The returns index is the inverse of the ratio of the return rate of a listing to the median return rate of the cohort. Thus, the set of return indices comprises a return index for each non-mature product listing in a category based on a returns ratio of a set of mature product listings in the category.

Now referring to Figure 3, an exemplary method flow diagram depicting a method for providing quality product listings to users on a digital platform is shown. The method starts at step 302 and goes to step 304. At step 304, the processing unit [102] retrieves a set of seller ratings, a set of one or more mature listings and a set of one or more non-mature listings from a memory unit [104], a set of rating indices from a rating index unit [106], and a set of return indices from a returns index unit [108]. In an implementation, the processing unit [102]retrieves the data of a pre-defined period of time, say, of last 365 days, etc.

The set of seller ratings comprises ratings provided by users to the sellers that are selling on the digital platform, also called as seller quality score or SQS score. A seller on the platform can be an old seller or a new seller. An old seller is the one who has been associated with the digital platform by listings its products on the platform, and is therefore expected to have received a number of reviews and ratings from the users on the platform. These reviews/ratings may be present in the form of an SQS score and saved in the memory unit [104].

Similarly, a product listing can be an old product listing that has been previously sold and/or is being sold on the digital platform for quite some time and therefore has rating and reviews from users on the platform. Further, this product listing can be a mature listing or a non-mature listing. A mature listing refers to the product listing that has been ordered and/or returned and/or reviewed by the customers at least a minimum threshold number of times. And, a non-mature listing refers to the product listing that has not been ordered and/or rated by the customers at least a minimum threshold number of times.

At step 306, the processing unit [102] generates a set of one or more listing quality scores (LQS scores) for each of the one or more non-mature listings based on the set of seller ratings, the set of rating indices retrieved from the rating index unit [106], and the set of return indices retrieved from the returns index unit [108]. These listing quality scores represent the actual quality of the product listings that are shown to the users. The quality of a product may be good or bad and may be measured using various factors such as attributes of performance or the core functionality of the product, features, build quality of the product, compliance to standards, durability, serviceability, etc. In several cases, it may also depend on the seller who is selling it. Similarly, the services of a seller might also be good or bad and may be measured using product availability, trustworthiness, fulfilment guarantees, efficiency of return process, comparative pricing, etc. The listing quality score takes into account the combination of all
such factors through the seller quality score, rating index and return index of a product listing. The LQS is used as a product quality indicator for individual listings. Accordingly, the listing quality score can be represented as given in equation (1) above in this disclosure.

Further at step 308, the processing unit [102] determines a set of listing bands comprising a listing band for each of the one or more non-mature listings and the one or more mature listings. Thus, the continuous-valued LQS score is further divided into 10 LQS bands {0, 1, 2,...,9}. The bands categorise product listings based on their quality. For example, the LQS band 0 (or band 0) may contain the best quality product listings, and band 9 may contain the worst quality product listings. Also, the bands are created based on cutoffs on the listing score at a category level. In other words, the set of listing bands is determined based on one or more pre-defined rules. In an exemplary implementation, the listing band cutoffs are determined to maintain a fixed percentage of delivered units in each band and ensure high quality listings to be discovered.

At step 310, a user interface unit [110] coupled with the processing unit [102] is configured to provide quality product listings on the digital platform based on the set of listing bands. As used herein, the user interface unit [110] includes an output device in the form of a display, such as a liquid crystal display (LCD), cathode ray tube (CRT) monitors, light emitting diode (LED) screens, etc. and/or one or more input devices such as touchpads or touchscreens. The display may be a part of a portable electronic device such as smartphones, tablets, mobile phones, wearable devices, etc. They also include monitors or LED/LCD screens, television screens, etc. that may not be portable. The display is typically configured to provide visual information such as text and graphics. An input device is typically configured to perform operations such as issuing commands, selecting, and moving a cursor or selector in an electronic device. Also, "providing quality product listings" includes determining the product listings for displaying to the users and/or displaying the quality product listings to
users on the digital platform. And the process ends at step 312.

The method prior to retrieving, by the processing unit [102], the set of rating indices from the rating index unit [106], further comprises generating a set of predicted ratings by the pre-trained first prediction unit [202].

In an implementation, where the prediction unit [202] does not generate accurate predicted ratings due to some reason, for example, the listing level data of the mature listings is not available. In such case, a rolled-up approach is implemented by the first prediction unit [202]. In this approach, the first prediction unit [202] checks the availability of the listing level data of mature listings. If the listing level data is not available, then the first prediction unit [202] is configured to check the data of higher level such as product level data, and if that is also not available, then brand or category level data, and so on.

In an implementation, the first prediction unit [202] is a pre-trained unit trained based on a machine learning model. For generating the predicted ratings for the non-mature product listings, the method comprises the first prediction unit [202] undergoing a pre-training or learning process. In this process, the first prediction unit [202] retrieves a first set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of average ratings for the one or more mature listings from the memory unit [108]. Further, the first prediction unit [202] comprises a first set of one or more learning functions comprising a learning function for each of the one or more categories of products. The first prediction unit [202] includes learning functions or learning models for various categories of products known as verticals. These learning functions are trained using various input features. The input features may be categorised as product attributes such as brand, material, size, weight, memory size, etc., as seller attributes such as seller score, cancellation score, delivery breaches, demographic information of seller such as location of the seller, business volume of the seller, etc., as listing attributes such as selling price, maximum rerail price, etc. among others. For
example, for a vertical/category of products, say, "shoes", input features such as "material", "weight", "brand", "selling price" are valid input features, and "memory size" would be an invalid input feature as shoes don't have a memory size such as that in a mobile phone. Therefore, for a particular vertical, the associated learning functions would be trained using the values of data in some valid input features that are associated with that vertical.

In the process of training the first prediction unit [202], the set of mature product listings is provided to the first prediction unit [202]. The set of mature product listings comprises data related to the rating of the product listings such as the value of the input features, and rating of the product and seller provided by the old users, etc. Based on the available data, the first prediction unit [202] maps average ratings for the mature listings with each input attribute/feature in the first set of one or more input attributes based on the one or more learning functions. Also, the first set of one or more input attributes refers to the input attributes that are related to and/or used by the first prediction unit [202]. Also, the first prediction unit [202] is trained to update the learning functions based on the data provided in various time intervals. Each learning function is trained based on a set of mature product listings in a particular category of products.

The first prediction unit [202] is also trained to identify the learning function to be applied on a particular product. For example, a non-mature product listing belonging to the "shoes" category is supplied to the first prediction unit [202]. So, the first prediction unit [202] itself figures out the learning function to be applied on the shoes category. Accordingly, the learning function is capable of predicting a rating to the non-mature product listing of the shoe using the input attributes of the non-mature product listing as the learning function has already learnt based on the same and/or similar input attributes of the "shoes" category/vertical.

However, the above predicted ratings maybe noisy and unstable, as the ratings are predicted based on all the data that is available, for example, the
ratings available for the mature product listings as well as the limited data available for non-mature product listings. Therefore, the average rating predictions from the learning models are used and the available values are used to smooth out the predicted ratings. This enables the first prediction unit [202] to efficiently make use of all available data for making the predictions. Thus, the first smoothing unit [204] is configured to generate a set of smooth ratings based on the set of predicted ratings, a total number of ratings, an average rating, and a hyperparameter. The smoothening function for rating, in an implementation, may be represented as given in equation (2) above. Also, the first smoothing unit [204] helps in normalising the data in case where the number of data points is less. For example, there are a number of predicted ratings and a number of actual ratings that is less than the required threshold of data points (i.e., ratings) to be used as actual data of a product listing. In this case, the first smoothing unit [204] may use the actual available ratings along with the predicted ratings, and also decide what weightage is to be given to the predicted ratings and the actual ratings available. The rating index unit [106] further determines the set of rating indices for the one or more non-mature listings based on the set of smooth ratings. Also, the ratings indices consider behaviour of the users, for example, different users have different behaviour towards different price points on the same product. In an implementation, the rating index of a product listing is determined as (Rating of the listing / Median rating of cohort), where a cohort is defined as (analytical vertical x price bucket). Finally, these rating indices are used as an input to calculate the LQS score for the non-mature product listing.

Further, the method prior to the processing unit [102] retrieving the set of rating indices from the returns index unit [108] comprises generating a set of predicted returns. The returns index unit [108] comprises a pre-trained second prediction unit [212] and a second smoothing unit [214]. The pre-trained second prediction unit [212] is configured to generate a set of predicted returns. Also, the set of predicted returns may be generated simultaneously along with
generating the set of predicted ratings by the first prediction unit [202], with or without a time difference. In other words, the steps of generating predicted ratings and predicted returns or the steps of generating rating indices or return indices are not sequential in nature and may happen simultaneously in the system [100]. Further, in an implementation, the second prediction unit [212] is a pre-trained unit trained based on a machine learning model. For generating the predicted returns for the non-mature product listings, the second prediction unit [212] undergoes pre-training or learning process. In this process, the second prediction unit [212] retrieves a second set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of average ratings for the one or more mature listings from the memory unit [108]. Further, the second prediction unit [212] comprises a second set of one or more learning functions comprising a learning function for each of the one or more categories of products. The second prediction unit [212] includes learning functions or learning models for various categories of products/ verticals. These learning functions are trained using various input features. The input features may be categorised as product attributes such as brand, material, size, weight, memory size, etc., as seller attributes such as seller score, cancellation score, delivery breaches, etc., as listing attributes such as selling price, maximum rerail price, etc. among others. For example, for a vertical/category of products, say, "shoes", input features such as "material", "weight", "brand", "selling price" are valid input features, and "memory size" would be an invalid input feature as shoes don't have a memory size such as that in a mobile phone. Therefore, for a particular vertical, the associated learning functions would be trained using the values of data in some valid input features that are associated with that vertical.

In the process of training the second prediction unit [212], the set of mature product listings is provided to the second prediction unit [212]. The set of mature product listings comprises data related to the rating of the product
listings such as the value of the input features, and rating of the product and seller provided by the old users, etc. Based on the available data, the second prediction unit [212] maps return ratios for the mature listings with each input attribute/feature in the second set of one or more input attributes based on the one or more learning functions. Also, the second set of one or more input attributes refers to the input attributes that are related to and/or used by the second prediction unit [212]. Also, the second prediction unit [212] is trained to update the associated learning functions based on the data provided in various time intervals. Each learning function is trained based on a set of mature product listings in a particular category of products.

The second prediction unit [212] is also trained to identify the learning function to be applied on a particular product. For example, a non-mature product listing belonging to the "shoes" category is supplied to the first prediction unit [202]. So, the second prediction unit [212] itself figures out the learning function to be applied on the shoes category. Accordingly, the learning function is capable of predicting a return to the non-mature product listing of the shoe using the input attributes of the non-mature product listing as the learning function has already learnt based on the same and/or similar input attributes of the "shoes" category/vertical. However, these predicted returns maybe noisy and unstable, as the returns are predicted based on all the data that is available, for example, the number of returns of the same product made earlier, total number of orders made by users, and other data available for the mature product listings as well as the limited data available for non-mature product listings. Therefore, the average returns ratio predictions from the learning models are used and the available values are used to smooth out the predicted ratings. This enables the second prediction unit [212] to efficiently make use of all available data for making the predictions. Thus, the second smoothing unit [214] is configured to generate a set of smooth returns based on the set of predicted returns, a total number of delivered units of products, an average
returns ratio and a hyperparameter. The smoothening function for returns, in an implementation, may be represented as given in equation (3) above. Also, the second smoothing unit [214] helps in normalising the data in case where the number of data points (i.e., returns) is less. For example, there are a number of predicted returns and a number of actual returns that is less than the required threshold of data points to be used as actual data of a product listing. In this case, the second smoothing unit [214] is configured to use the available returns along with the predicted returns, and also decide what weightage is to be given to the predicted returns and the actual returns data available. The smoothened ratings are then sent from second smoothing unit [214] to returns index unit [108]. The returns index unit [108] is further configured to determine the set of return indices for the one or more non-mature listings based on the set of smooth ratings from the second smoothing unit [214].

It is evident from the above disclosure, that the solution provided by the disclosure is technically advanced as compared to the prior known solutions. The existing methods using the survey based approaches only provide ratings to the product listings that are mature and enough data is available with the system to generate ratings for the listings. Since the existing approaches did not implement the prediction units for predicting the scores and also did not generate the listing quality scores based on seller quality scores, rating indices, and return indices as the present disclosure, they were not able to solve the problem of cold start which is present in a new product listing. The present disclosure enables a person skilled in the art to solve the cold start problem in product listings. Also, the present disclosure enables a person skilled in the art to obtain a method and system that overcomes the technical limitation of obtaining a more accurate rating of product listing as it is based on higher number of inputs. By implementing the above disclosure, a person skilled in the art is able to provide new product listings that are of good quality to the users at priority irrespective of the fact that sufficient data is not available to calculate the actual rating of the
product listing on the digital platform.

While considerable emphasis has been placed herein on the disclosed embodiments, it will be appreciated that many embodiments can be made and that many changes can be made to the embodiments without departing from the principles of the present disclosure. These and other changes in the embodiments of the present disclosure will be apparent to those skilled in the art, whereby it is to be understood that the foregoing descriptive matter to be implemented is illustrative and non-limiting.
We claim:

1. A method for providing quality product listings to users on a digital

platform, the method comprising:

- retrieving, by a processing unit [102], a set of seller ratings, a set of one or more mature listings and a set of one or more non-mature listings from a memory unit [104], a set of rating indices from a rating index unit [106], and a set of return indices from a returns index unit [108];

- generating, by the processing unit [102], a set of one or more listing quality scores for each of the one or more non-mature listings based on the set of seller ratings, the set of rating indices, and the set of return indices;

- determining, by the processing unit [102], a set of listing bands comprising a listing band for each of the one or more non-mature listings and the one or more mature listings, wherein the set of listing bands is determined based on one or more pre-defined rules; and

- providing, by a user interface unit [110], quality product listings on the digital platform based on the set of listing bands.

2. The method as claimed in claim 1, wherein the method prior to

retrieving, by the processing unit [102], the set of rating indices from the

rating index unit [106], further comprises:

- generating, by a pre-trained first prediction unit [202], a set of predicted ratings;

- generating, by a first smoothing unit [204], a set of smooth ratings based on the set of predicted ratings, a total number of ratings, an average rating, and a hyperparameter; and

- determining, by the rating index unit [106], the set of rating indices for the one or more non-mature listings based on the set of smooth ratings.
3. The method as claimed in claim 1, wherein the method prior to retrieving, by the processing unit [102], the set of return indices from the return index unit [108] further comprises:

- generating, by a pre-trained second prediction unit [212], a set of predicted returns;

- generating, by a second smoothing unit [214], a set of smooth returns based on the set of predicted returns, a total number of delivered units, an average returns ratio, and a hyperparameter; and

- determining, by the returns index unit [108], the set of return indices for the one or more non-mature listings based on the set of smooth returns.

4. The method as claimed in claim 2, wherein the method prior to the generating, by the pre-trained first prediction unit [202], the set of predicted ratings, further comprises pre-training of the first prediction unit [202] comprising:

- retrieving, by the first prediction unit [106] from the memory unit [108], a first set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of average ratings for the one or more mature listings;

- generating, by the rating index unit [106], a first set of one or more learning functions comprising a learning function for each of the one or more categories of products; and

- mapping, by the rating index unit [106], average ratings for the one or more mature listings with each input attribute in the first set of one or more input attributes based on the one or more learning functions.

5. The method as claimed in claim 2, wherein the method prior to the generating, by the pre-trained second prediction unit [212], the set of predicted returns, further comprises pre-training of the second prediction
unit [212] comprising:

- retrieving, by the second prediction unit [108] from the memory unit [108], a second set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of return ratios for the one or more mature listings;

- generating, by the returns index unit [108], a second set of one or more learning functions comprising a learning function for each of the one or more categories of products; and

- mapping, by the returns index unit [108], return ratios for the one or more mature listings with each input attribute in the second set of one or more input attributes based on the one or more learning functions.

6. The method as claimed in claim 1, wherein the set of rating indices comprises a rating index for each non-mature product listing in a category based on an average rating of a set of mature product listings in the category.

7. The method as claimed in claim 1, wherein the set of returns indices comprises a returns index for each non-mature product listing in a category based on an average return ratio of a set of mature product listings in the category.

8. A system for providing quality product listings to users on a digital platform, the system comprising:

- a processing unit [102] configured to:

o retrieve a set of seller ratings, a set of one or more mature listings, and a set of one or more non-mature listings from a memory unit [104], a set of rating indices from a rating index unit [106], and a set of return indices from a returns index unit [108];
o generate a set of one or more listing quality scores for each of the one or more non-mature listings based on the set of seller ratings, the set of rating indices retrieved from the rating index unit [106], and the set of return indices retrieved from the returns index unit [108];

o determine a set of listing bands comprising a listing band for each of the one or more non-mature listings and the one or more mature listings, wherein the set of listing bands is determined based on one or more pre-defined rules; and

- a user interface unit [110] coupled with the processing unit [102], the user interface unit [102] configured to:

o provide quality product listings on the digital platform based on the set of listing bands.

9. The system as claimed in claim 8, further comprising:

- a pre-trained first prediction unit [202] configured to generate a set of predicted ratings, wherein the set of predicted ratings is generated prior to the processing unit [102] retrieving the set of rating indices from the rating index unit [106]; and

- a first smoothing unit [204] configured to generate a set of smooth ratings based on the set of predicted ratings, a total number of ratings, an average rating, and a hyperparameter.

10. The system as claimed in claim 9, wherein the rating index unit [106] is further configured to determine the set of rating indices for the one or more non-mature listings based on the set of smooth ratings.

11. The system as claimed in claim 8, further comprising:

- a pre-trained second prediction unit [212] configured to generate a set of predicted returns, wherein the set of predicted returns is generated prior to the processing unit [102] retrieving the set of return indices from the returns index unit [108]; and
- a second smoothing unit [214] configured to generate a set of smooth returns based on the set of predicted returns, a total number of delivered units, an average return ratio, and a hyperparameter.

12. The system as claimed in claim 11, wherein the returns index unit [108] is further configured to determine the set of return indices for the one or more non-mature listings based on the set of smooth ratings.

13. The system as claimed in claim 9, wherein for pretraining the first prediction unit [202], the first prediction unit [202] is further configured to:

- retrieve, from the memory unit [108], a first set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of average ratings for the one or more mature listings;

- generate a first set of one or more learning functions comprising a learning function for each of the one or more categories of products; and

- map average ratings for the one or more mature listings with each input attribute in the first set of one or more input attributes based on the one or more learning functions.

14. The system as claimed in claim 8, wherein for pretraining the second prediction unit [212], the second prediction unit [212] is further configured to:

- retrieve, from the memory unit [108], a second set of one or more input attributes, a set of one or more categories comprising a category associated with each of the one or more mature listings, and a set of returns ratios for the one or more mature listings;

- generate a second set of one or more learning functions comprising a learning function for each of the one or more categories of products; and
- map returns ratios for the one or more mature listings with each input attribute in the second set of one or more input attributes based on the one or more learning functions.

15. The system as claimed in claim 8, wherein the set of rating indices comprises a rating index for each non-mature product listing in a category based on an average rating of a set of mature product listings in the category.

16. The system as claimed in claim 8, wherein the set of return indices comprises a return index for each non-mature product listing in a category based on a returns ratio of a set of mature product listings in the category.