WO2023286087A1

WO2023286087A1 - Providing personalized recommendations based on users behavior over an e-commerce platform

Info

Publication number: WO2023286087A1
Application number: PCT/IN2022/050640
Authority: WO
Inventors: Kamal Kumar; Vineet MAHALE; Ashish PAGOTE; Jnanesh PRABHU; Sayandeep BHOWMIK
Original assignee: Tata Unistore Limited
Priority date: 2021-07-15
Filing date: 2022-07-15
Publication date: 2023-01-19

Abstract

The present invention describes a system and a method for providing personalized recommendations to users accessing an e-commerce platform. One or more data models are trained on data associated with browsing history of a user accessing the e-commerce platform, for providing recommendations for making purchases over the e-commerce platform. One or more recommendations specific to the user may be generated using the one or more data models. A notification corresponding to the one or more recommendations may be provided to the user. The notification may enable the user to browse the e-commerce platform in respect of the one or more recommendations. Activities of the user may be tracked over the e-commerce platform. The repository may be iteratively updated based on the activities of the user over the e-commerce platform, to provide personalized recommendations for making purchases over the e-commerce platform.

Description

“PROVIDING PERSONALIZED RECOMMENDATIONS BASED ON USERS BEHAVIOR OVER AN E-COMMERCE PLATFORM”

FIELD OF INVENTION

The present invention generally relates to a marketing approach. Specifically, the present invention relates to a method for performing marketing based on behavioral economics.

BACKGROUND OF THE INVENTION

The subject matter discussed in the background section should not be assumed to be prior art merely as a result of its mention in the background section. Similarly, a problem mentioned in the background section or associated with the subject matter of the background section should not be assumed to have been previously recognized in the prior art. The subject matter in the background section merely represents different approaches, which in and of themselves may also correspond to implementations of the claimed technology.

Nudging customers for engaging them is an important element in e-commerce industry. Marketing stack of e-commerce platforms often provide recommendations related to purchase of different products to the users. This approach is extensively used for boosting user growth, maintaining user engagement, and ensuring repeated visits of the users on the e-commerce platforms. If the right set of products are recommended, the users are highly likely to re-visit the e-commerce platform, find products that are relevant to them, and possibly move towards a purchase. This also enables user association, which is pivotal for an e-commerce business to sustain in the longer term. However, to move the focus significantly on purchases, businesses believe that the faster they can guide a user in finding products, the higher the chances of purchase becomes. For minimizing the user effort in product discovery, organizations often focus on rushing users through a purchase life cycle initiating a sense of urgency by sending frequent push notifications based on users’ recently viewed products, and combining them with offers and limited time deals. While this approach helps the organizations to achieve user growth and sales in the shorter term, it does not necessarily build user engagement on the digital platforms and users often become infrequent shoppers or purely deal seekers.

Overwhelming customers with push notifications of viewed products being available with offers and creating a fear of missing out the offers in their mind may generate a lethargic outlook towards the purchase. Contiguous communication along the similar lines open the risk of the user uninstalling/blocking the online application for avoiding future notifications, thereby losing a customer to imprecise over- targeting.

Existing technologies mainly rely on trigger-based notification like nudging users to checkout items left in cart or rule-based ones like geospatial targeting which is not adept at identifying recent user preferences or changes in user preferences over time thereby performance peaks to saturation after some time which can often lead to a gap in customer shopping experience.

Therefore, there is a need for a system and a method that can predict the user preferences with high degree of accuracy and provide relevant suggestions for shopping over e-commerce platforms.

OBJECTS OF THE INVENTION

A general objective of the invention is to provide personalized recommendations to users based on their behavior over e-commerce platforms via marketing channels.

Another objective of the invention is to accurately predict user preferences related to purchase of online products. Another objective is to improve the engagement of users on the e-commerce platform via marketing channels.

Another objective of the invention is to provide push notifications related to the personalized recommendations.

Another objective of the invention is to reach out to customers at an appropriate time through a push notification.

Yet another objective of the invention is to improve click through rates and efficiency of e-commerce platforms.

SUMMARY OF THE INVENTION

This summary is provided to introduce aspects related to provide a marketing approach based on behavioural science, and the aspects are further described below in the detailed description. This summary is not intended to identify essential features of the claimed subject matter nor is it intended for use in determining or limiting the scope of the claimed subject matter.

In one embodiment, personalized recommendations may be provided to users accessing an e-commerce platform. Data associated with browsing history of a user accessing the e-commerce platform may be acquired in a repository. The data associated with the browsing history comprises one or more of click stream data, product metadata, transaction data, and user’s metadata. One or more data models may be trained for providing recommendations for making purchases over the e-commerce platform. The one or more data models may be trained on the data associated with the browsing history. One or more recommendations specific to the user may be generated using the one or more data models. A notification corresponding to the one or more recommendations may be provided to the user. The notification may enable the user to browse the e-commerce platform in respect of the one or more recommendations. Activities of the user may be tracked over the e-commerce platform. The repository may be iteratively updated based on the activities of the user over the e-commerce platform, to provide personalized recommendations for making purchases over the e- commerce platform.

In one aspect, the data associated with the browsing history comprises one or more of click stream data, product metadata, transaction data, and the user’ s metadata.

In an aspect, the one or more data models are tested on one or more of contextual factors, individual factors, and social factors. The contextual factors include one or more of affinities to different products available on the e-commerce platform, regularity of interest shown by the user, interactions in newly recommended products on the e- commerce platform, and response of the user to the notification. The individual factors include the user’s psyche, the user’s consistency, speed of decision making of the user, and demographics. The social factors include social norm and proof, trust, and liking.

In an aspect, the one or more recommendations are generated based on one or more techniques of machine learning and deep learning.

In an aspect, the notification corresponding to the one or more recommendations is provided in form of a push notification.

In an embodiment, a system for providing personalized recommendations to users accessing an e-commerce platform may comprise a server; and a repository storing programmed instructions executable by the processor. Data associated with browsing history of a user accessing the e-commerce platform may be acquired in a repository. The data associated with the browsing history comprises one or more of click stream data, product metadata, transaction data, and user’ s metadata. One or more data models may be trained on the data associated with the browsing history for providing recommendations for making purchases over the e-commerce platform. One or more recommendations specific to the user may be generated using the one or more data models. A notification corresponding to the one or more recommendations may be provided to the user. The notification may enable the user to browse the e-commerce platform in respect of the one or more recommendations. Activities of the user may be tracked over the e-commerce platform. The repository may be iteratively updated based on the activities of the user over the e-commerce platform, to provide personalized recommendations for making purchases over the e-commerce platform.

Other aspects and advantages of the invention will become apparent from the following description, taken in conjunction with the accompanying drawings, illustrating by way of example the principles of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings constitute a part of the description and are used to provide a further understanding of the present invention.

Fig. 1 illustrates a network connection diagram of a system for providing personalized recommendations based on users’ behavior over an e-commerce platform, in accordance with an embodiment of the present invention.

Figs. 2A-2D cumulatively illustrate a process flow for providing personalized recommendations based on users’ behavior over an e-commerce platform, in accordance with an embodiment of the present invention.

Fig. 3 illustrates a flow chart of a method for providing personalized recommendations based on users’ behavior over an e-commerce platform, in accordance with an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The detailed description set forth below in connection with the appended drawings is intended as a description of various embodiments of the present invention and is not intended to represent the only embodiments in which the present invention may be practiced. Each embodiment described in this disclosure is provided merely as an example or illustration of the present invention, and should not necessarily be construed as preferred or advantageous over other embodiments. The detailed description includes specific details for the purpose of providing a thorough understanding of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced without these specific details.

Fig. 1 depicts network connection diagram of a system 100 for providing personalized recommendations based on users behavior over an e-commerce platform, in accordance with an embodiment of the present invention. The system 100 is connected with a plurality of client devices 102-1 to 102-n through a communication network 104. The plurality of client devices 102-1 to 102-n may include any electronic device that could be used to access an e-commerce software application, such as a smartphone 102-2, tablet 102-4, laptop 102-6, and a desktop 102-n.

The communication network 104 may be a wired and/or a wireless network. The communication network 104, if wireless, may be implemented using communication techniques such as Visible Light Communication (VLC), Worldwide Interoperability for Microwave Access (WiMAX), Long Term Evolution (LTE), Wireless Local Area Network (WLAN), Infrared (IR) communication, Public Switched Telephone Network (PSTN), Radio waves, and other communication techniques known in the art.

The system 100 may be a processing device capable of storing and processing data related to different activities of the users over the e-commerce software application. For example, the system 100 may be a server configured locally at a site or over a network cloud. By processing the data related to different activities of the users, the system 100 would determine and provide personalized recommendations to each user. This would improve the users’ online experience and ultimately lead to increase in product sales. The detailed methodology of determining the personalized recommendations is described henceforth with reference to Fig. 2. Figs. 2A-2D cumulatively illustrate a process flow for providing personalized recommendations based on users’ behavior over an e-commerce platform, in accordance with an embodiment of the present invention. At step 202 illustrated in Fig. 2A, basic information entered on the e-commerce platform by the user is stored. Data through various modes like browsing history on e-commerce platform is captured at step 204 illustrated in Fig. 2A. Such data may include click stream data, product metadata, transaction data, and customer’s metadata, and may be provided by source systems residing in cloud as an infrastructure. At step 206 illustrated in Fig. 2A, the data may be stored, for example in Simple Storage Service (S3™) which is a cloud based data storage service offered by Amazon Web Services™.

At step 208 illustrated in Fig. 2A, the data may be provided to a tech stack including Application Programming Interface (API) libraries, analytics engine, and machine learning platforms. In one implementation, the tech stack may include, but not limited to Python™, Amazon SageMaker™, Keras™, and Spark™. In different implementations, these Platforms/Services may be replaced by equivalent platforms / services providing the same capabilities. At step 210 illustrated in Fig. 2B, data engineering operations may be carried out on the data. The data engineering operations may include, but not limited to, feature re-engineering, data aggregation, feature extraction, data standardization, and variable selection.

In one implementation, the data engineering operations may be performed for creation of variables, transformation of the variables, and embedding. These variables may be defined based on exploratory research of user journeys and how sequentially the users change their behaviours. Most of the variables were either present in the form of percentage changes, log transformations, or scaled values of absolute numbers. Embedding was used to represent product attributes, marketing campaign attributes, and customer attributes basis persona.

High variance could be generally observed in values of features including product views, session durations, and gap between consecutive sessions. For processing of such features, different kind of transformations such as log and smoothening techniques could be utilized, such as through application of Kalman filters.

At step 212 illustrated in Fig. 2B, development and evaluation of one or more data models may be performed to predict user propensity of purchase within a specific category and to understand specific user interests leveraged to derive messaging aspects. Development of the one or more data models may include one or more operations such as dimensionality reduction, model selection, and hyperparameter tuning. Evaluation of the one or more data models may include operations such as Key Performance Indicator (KPI) tracking and model re-training.

In one implementation, dimensionality reduction included Principal Component Analysis (PCA), specifically for use on embeddings. Categorical encodings were avoided because embeddings provided better accuracies even when used in data models such as XGboost and Light Gradient Boosting Machine (LGBM). Hyperparameter tuning was performed differently based on types of the data models. For boosted trees, different values of leaves, max depth, and regularisation coefficients was tried, given the imbalanced nature of classes. Parameter selection was performed based on Bayesian search after an initial experiment of Grid search. For sequential data models, deeper networks were trained with multiple layers, such as with 256,128 layers. Dropouts, rmsprop, Adagrad were included to help with accuracy improvements. Different learning rates were tried, and finally a learning rate of 0.07 was finalized. During training of the data models, trees were trained separately over shorter sessions and sequential models were trained for longer sequence of sessions. This helped in identifying and segregating users who are fast shoppers and the users who take time and have exploratory nature.

Post development and evaluation of the one or more data models, tenets of a recommendation framework may be tested, at step 214 illustrated in Fig. 2C. The tenets may include, but not limited to, contextual factors, individual factors, and social factors. The contextual factors may include affinities to different products, regularity of interest shown, interactions in newly recommended products, and response to messaging. The individual factors may include customer psyche, consistency, fast and slow decision making, and demographics. The social factors may include social norm and proof, trust, and liking.

In one implementation, a split and target approach was utilized. Different category purchases may involve different purchase journeys, as it was observed from user browsing journey research. Behavioural economics tenets were specific and curated at a category level. After identifying high propensity users and their interested categories, Collaborative Filtering (CF) techniques were used to arrive at recommendations. At first, products were filtered out based on values of key business metrics, such as %PC1, conversion rates, Gross merchandise value (GMV) share, and median rank of product during a Business As Usual (BAU) week on our Product Line Profitability (PLP). In case the values of the key business metrics was determined to be very low, corresponding products were filtered out. Thereafter, users are selected based on purchase propensity scores. A CF data model is only used for users that need to be engaged more on variety of products refined to their tastes. If the propensity scores are very high, the users are engaged based on their most interested type of products or brands, which are directly derived from their high intent touch points. Such process if performed by curating the messages in notifications.

Using a focused set on the former user cohort, the process is made computationally efficient and cheaper by training user and product embeddings. PCA is also used to reduce the dimensionality of these embeddings. User embeddings include BE traits specific to categories and users if when sufficient browsing and purchase history is available. For example, in electronic category purchases where more complex product attributes are involved, BE traits such as Bounded rationality leveraging metrics from user's clicks, search and filters are defined. Product attributes essential to users may be selectively identified and used in product embeddings. All the traits, when tuned in a CF technique, help in narrowing down of a target set of products that are most relevant to every user. At step 216 illustrated in Fig. 2C, a behaviour economics intelligence layer may interpret the user journey, and determine user propensity and recommendations. The behaviour economics intelligence layer may function as a final adaptive layer that tunes itself based on learning obtained at several levels i.e. the different steps described previously. In one implementation, the behaviour economics intelligence layer may utilize techniques including machine learning and deep learning models to formulate product recommendations. The product recommendations may be sent to a campaign management tool, at step 218 illustrated in Fig. 2D. The campaign management tool may perform campaign management operations to generate notifications corresponding to the product recommendations.

At step 220 illustrated in Fig. 2D, the notification corresponding to the product recommendations may be provided to a user/customer. In one implementation, the notification may be provided to the user as a push notification on the e-commerce platform/ software application. At step 222 illustrated in Fig. 2D, the user may click on the push notification based on which the user may be landed on a webpage displaying the relevant products (product recommendations determined for him). Thereupon, the user’s activities related to purchase and browsing of the recommended or other products may be tracked, and through a feedback loop, data associated with the user’s activities performed in response to the notification may be stored and process further, in the manner described through the steps 204 to 222. During such process, the activities performed by the user on the e-commerce platform, for example browsing and purchasing products in absence of notifications are continuously tracked, and corresponding data is retrieved from step 202 for processing.

Fig. 3 illustrates a flow chart 300 showing a method for providing personalized recommendations to users accessing an e-commerce platform, in accordance with an embodiment of the present invention. In this regard, each block may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the drawings. For example, two blocks shown in succession in Fig. 3 may in fact be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. Any process descriptions or blocks in flow charts should be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps in the process, and alternate implementations are included within the scope of the example embodiments in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved.

At step 302, data associated with browsing history of a user accessing the e-commerce platform may be acquired in a repository. Such data may include click stream data, product metadata, transaction data, and customer’s metadata, and may be provided by source systems residing in cloud as an infrastructure.

At step 304, one or more data models may be trained on the data associated with the browsing history. The one or more data models may be trained for providing recommendations for making purchases over the e-commerce platform.

At step 306, one or more recommendations specific to the user may be generated using the one or more data models. The one or more recommendations are generated based on one or more techniques of machine learning and deep learning.

At step 308, a notification corresponding to the one or more recommendations may be provided to the user. The notification may enable the user to browse the e-commerce platform in respect of the one or more recommendations. At step 310, activities of the user over the e-commerce platform may be tracked. For example, the user’s activities related to purchase and browsing of the recommended or other products may be tracked.

At step 312, the repository may be iteratively updated based on the activities of the user over the e-commerce platform. Thus, personalized recommendations may be provided to the user for making purchases over the e-commerce platform.

The present disclosure proposes a solution that leverages users’ inconspicuous touchpoints during browsing journey to understand the user psyche and decipher context to make an informed assessment of user’s purchase life cycle i.e. the users in discovery phase, evaluation phase, or final pre-purchase lap. These touchpoints were identified from an extensive research into the customer psychographics, analysing how user’s purchase life cycle progresses through every phase of purchase, how user makes his decisions during discovery and evaluation, and what intent indicators could be derived during his evolving purchase life cycle. Such information is fed into deep learning models layering browsing preferences of individual users arriving at the relevant assortment of products which are capable of continuously learning and evolving through touchpoints gathered by user’s engagement with platform. Through this gained knowledge of the users, the solution helps in reaching the users at the most relevant time, the context of which may be derived but not limited to factors that influence user behaviour such as browsing or location or temporal factors.

The present disclosure identifies and removes barriers to purchase and nudges people to check-out products which are ideal choices according to their needs. This approach focuses on understanding the customer’s need, perceive the signals through various touchpoints left by the customer, identifying customer psyche and where the customer is currently placed in his/her purchase life cycle, understanding the context of purchase. The present disclosure not only curate the product recommendations on the basis of the above said indicators but also to understand the feedback of the customer to this consideration set - for refining and curating the list further. The present disclosure gives more emphasis is on the customer and his preferences thereby providing customers an enriching shopping experience. The proposed solution offered through present disclosure may be used both as a plug-able module or product-as-a-service.

Understand, advise, and adapt to the dynamic behaviour of users by redefining the usage of push notifications campaigns remains the core of present disclosure. Push notifications are not just a business driver but an opportunity to interact with user in order to create an evolving product funnel as per user preferences. The engine is able to engage in this communication as it can decipher users' subtle interests in every stage of their purchase life cycle using a hybrid of behavioural economics, big data and advanced data sciences.

A plug and play product combines behavioural sciences with data science to mine through massive clickstream data and deciphers inconspicuous user preferences to recommend relevant set of products. Proposed system takes clickstream and transactional data as input and outputs curated target lists as recommendations, which are integrated with campaign management (CM) tools for launching innovative & personalized campaigns.

The present disclosure enhances the ability of push notifiers to guide and enable users towards an enriching experience by providing a decision engine, which uses machine learning techniques to find insightful patterns in their browsing and prescribes the next step to step driven largely by their own psyche. It also expands the ability by reaching out to the users with the relevant message at the appropriate time, which may be identified but not limited to factors that could influence users - browsing, weather, demographics etc.

The hybrid module analyses the touchpoints left by the user/consumer, making itself aware of the behaviour that a person showcases and refines its recommendations. This shift from functional rule-based triggers to insightful behaviour-based recommendations was instrumental in enhancing user engagement. Relevant end user touch points in form of push notifications have higher engagement with end users as evidently measured by high click-through-rates, open rates, time spent and product views. This was driven by better relevance in marketed products, which were derived through user preferences and purchase intent indicators. In one test observation, the behavioural science driven campaigns also led to an increase in 6% of the marketplace annual revenue. In one such observation, it was identified that the proposed system was able to engage users better on the platform through visits from the innovation-led campaign, enabling them to view multiple pages, and having higher interactions. This is evident in measurable metrics such as 5% increase in time spent per visitor and 23% increase in product views made by customer per session. The boosted engagement led users to make confirmed choices on purchase and lead to quicker purchases in the final session. In one implementation, the improvement in attributes achieved by employing the present invention were as depicted below in table 1.

Table 1 Present disclosure sets up a conversation that will help user address their preferences and responds to them even though they change over time. Hence, the present disclosure offers higher technical and economic significance through a personalized and behavioural approach in marketing. The present disclosure has adopted an approach of keeping the customer at the center. The present disclosure not only provided a personalized user journey to the end user but also ensures a multi-fold improvement in business metrics such as click through rates and efficiency rates. The present disclosure explains the way and the need of customer psyche to be read. Using the behavioral sciences and data science, the user preferences will be predicted with high accuracy. Hence the proposed invention may augment organizations in providing personalized experience to the customers and increase their revenues.

An interface may be used to provide input or fetch output from the system. The interface may be implemented as a Command Line Interface (CLI), Graphical User Interface (GUI). Further, Application Programming Interfaces (APIs) may also be used for remotely interacting with edge systems and cloud servers.

A processor may include one or more general purpose processors (e.g., INTEL® or Advanced Micro Devices® (AMD) microprocessors) and/or one or more special purpose processors (e.g., digital signal processors or Xilinx® System On Chip (SOC) Field Programmable Gate Array (FPGA) processor), MIPS/ARM-class processor, a microprocessor, a digital signal processor, an application specific integrated circuit, a microcontroller, a state machine, or any type of programmable logic array.

A repository may include, but is no limited to, non-transitory machine-readable storage devices such as hard drives, magnetic tape, floppy diskettes, optical disks, Compact Disc Read-Only Memories (CD-ROMs), and magneto-optical disks, semiconductor memories, such as ROMs, Random Access Memories (RAMs), Programmable Read- Only Memories (PROMs), Erasable PROMs (EPROMs), Electrically Erasable PROMs (EEPROMs), flash memory, magnetic or optical cards, or other type of media/machine- readable medium suitable for storing electronic instructions. Module or unit or engine as used herein, such as the data collection unit, data pre processing unit, semantic engine, and the communication unit are intended to encompass any collection or set of program instructions executable over network cloud so as to perform required task by the software. The term “software” as used herein is intended to encompass such instructions stored in storage medium such as RAM, a hard disk, optical disk, or so forth, and is also intended to encompass so-called “firmware” that is software stored on a ROM or so forth. Such software may be organized in various ways, and may include software components organized as libraries, Internet-based programs stored on a remote server or so forth, source code, interpretive code, object code, directly executable code, and so forth. It is contemplated that the software may invoke system-level code or calls to other software residing on server or other location to perform certain functions.

The terms “or” and “and/or” as used herein are to be interpreted as inclusive or meaning any one or any combination. Therefore, “A, B or C” or “A, B and/or C” mean “any of the following: A; B; C; A and B; A and C; B and C; A, B and C.” An exception to this definition will occur only when a combination of elements, functions, steps or acts are in some way inherently mutually exclusive.

Any combination of the above features and functionalities may be used in accordance with one or more embodiments. In the foregoing specification, embodiments have been described with reference to numerous specific details that may vary from implementation to implementation. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense. The sole and exclusive indicator of the scope of the invention, and what is intended by the applicants to be the scope of the invention, is the literal and equivalent scope of the set as claimed in claims that issue from this application, in the specific form in which such claims issue, including any subsequent correction.

Claims

CLAIMS:

1. A method for providing personalized recommendations to users accessing an e- commerce platform, the method comprising: acquiring (302), in a repository, data associated with browsing history of a user accessing the e-commerce platform; training (304), on the data associated with the browsing history, one or more data models for providing recommendations for making purchases over the e-commerce platform; generating (306), using the one or more data models, one or more recommendations specific to the user; providing (308), to the user, a notification corresponding to the one or more recommendations, wherein the notification enables the user to browse the e-commerce platform in respect of the one or more recommendations; tracking (310) activities of the user over the e-commerce platform; and iteratively updating (312) the repository based on the activities of the user over the e-commerce platform, to provide personalized recommendations for making purchases over the e-commerce platform.

2. The method as claimed in claim 1, wherein the data associated with the browsing history comprises one or more of click stream data, product metadata, transaction data, and the user’s metadata.

3. The method as claimed in claim 1, wherein the one or more data models are tested on one or more of contextual factors, individual factors, and social factors.

4. The method as claimed in claim 3, wherein the contextual factors include one or more of affinities to different products available on the e-commerce platform, regularity of interest shown by the user, interactions in newly recommended products on the e-commerce platform, and response of the user to the notification, the individual factors include the user’s psyche, the user’s consistency, speed of decision making of the user, and demographics, and the social factors include social norm and proof, trust, and liking.

5. The method as claimed in claim 1, wherein the one or more recommendations are generated based on one or more techniques of machine learning and deep learning.

6. The method as claimed in claim 1, wherein the notification corresponding to the one or more recommendations is provided in form of a push notification.

7. A system (100) for providing personalized recommendations to users accessing an e-commerce platform, the system comprising: a server; and a repository storing programmed instructions executable by the processor, wherein the processor executes the programmed instructions to: acquire, in a repository, data associated with browsing history of a user accessing the e-commerce platform; train, on the data associated with the browsing history, one or more data models for providing recommendations for making purchases over the e-commerce platform; generate, using the one or more data models, one or more recommendations specific to the user; provide, to the user, a notification corresponding to the one or more recommendations, wherein the notification enables the user to browse the e-commerce platform in respect of the one or more recommendations; track activities of the user over the e-commerce platform; and iteratively update the repository based on the activities of the user over the e-commerce platform, to provide personalized recommendations for making purchases over the e-commerce platform.

8. The system (100) as claimed in claim 7, wherein the data associated with the browsing history comprises one or more of click stream data, product metadata, transaction data, and user’s metadata.

9. The system (100) as claimed in claim 7, wherein the one or more data models are tested on one or more of contextual factors, individual factors, and social factors.

10. The system (100) as claimed in claim 9, wherein the contextual factors include one or more of affinities to different products available on the e-commerce platform, regularity of interest shown by the user, interactions in newly recommended products on the e-commerce platform, and response of the user to the notification, wherein the individual factors include the user’s psyche, the user’s consistency, speed of decision making of the user, and demographics, and the social factors include social norm and proof, trust, and liking.

11. The system (100) as claimed in claim 7, wherein the one or more recommendations are generated based on one or more techniques of machine learning and deep learning.

12. The system (100) as claimed in claim 7, wherein the notification corresponding to the one or more recommendations is provided in form of a push notification.