Navigating E-commerce Serendipity: Leveraging Innovator-Based Context Aware Collaborative Filtering for Product Recommendations

Navigating E-commerce Serendipity: Leveraging Innovator-Based Context Aware Collaborative Filtering for Product Recommendations

Koushik Reddy Chaganti Chinnala Balakrishna

Research Scholar Associate Professor

Dept. of Computer and Information Systems Department of CSE(AIML & Cyber Security) University of the Cumberlands Guru Nanak Institute of Technology(A)

KY USA Hyderabad, India

P.Naresh P.Rajyalakshmi

Asst.Professor Asst.Professor

Dept of Information Technology Department of CSE

Vignan Institute of Technology and Science (A) TKR College of Engineering and Technology Hyderabad Hyderabad

Abstract- This paper introduces a novel approach to enhancing product recommendations in e-commerce settings by leveraging innovator-based context-aware collaborative filtering. Traditional recommendation systems often struggle to capture the serendipitous nature of user preferences, leading to limited discovery of new and relevant products. In response, we propose a methodology that combines insights from innovators, who exhibit a propensity for exploring novel items, with contextual information such as time, location, and user behavior. By integrating these elements into a collaborative filtering framework, our approach aims to deliver personalized recommendations that not only align with users' existing preferences but also introduce unexpected and intriguing product suggestions. To evaluate the effectiveness of our proposed system, we conduct experiments using a comprehensive dataset from the e-commerce domain. Moreover, we present three distinct algorithmic paradigms: contextual pre-filtering, post-filtering, and modeling. These paradigms aim to integrate contextual information seamlessly into the recommendation process. We explore the potential synergies of amalgamating multiple context-aware recommendation techniques into a cohesive framework and offer insights into the benefits of such integration. Additionally, we offer a case study illustrating the implementation and efficacy of one such integrated approach in real-world scenarios. Results demonstrate significant improvements in recommendation accuracy, serendipity, and user engagement compared to traditional methods. Overall, our study highlights the potential of innovator-based context-aware collaborative filtering to

navigate e-commerce serendipity and enhance the shopping experience for users.

Keywords: Recommender System, Context aware Collaborative Filtering, Serendipity, Innovator-Based

1. INTRODUCTION

   Recommender Systems can be defined as programs which attempt to recommend the most suitable items (products or services) to particular users (individuals or businesses) by predicting a user's interest in an item based on related information about the items,the users and the interactions between items and users [1-4]. To provide genuine recommendations to a user so that the suggested items or products are offering the utmost satisfaction should be given the priority while designing any Recommender System [5-6]. There are plenty of Recommender Systems available in the literature.

   But the items offered as recommendations by the majority of the Recommender Systems do have the tendency to recommend popular or easily identifiable or routine items [7-8]. Because these offerings by the majority of the Recommender Systems lack the components of novelty and serendipity, such Recommender Systems end up facing the issues of Popularity Bias, ignorance of the Long Tail items and Matthew Effect etc [5-12]. Because of such shortfalls of the traditional Recommender Systems, the products which are popular in the catalog have the

   tendency to gain even more popularity and contribute to the ever expanding lengthy list of the Long Tail of Non-Popular Items, waiting to be recommended for the endless time, leading towards the starvation [5-10]. Traditional recommender systems usually use accuracy as the main measure to evaluate performance. However, improving accuracy does not mean improving user satisfaction [14], [15], [16]. For example, recommending items that the user already puts into cart can reach very high accuracy but does not make any sense. A good recommender system need not only accurately predict user shopping behaviors, but also broaden user horizon and discover their potential interests [17-21]

   The Emergence of Recommender Systems in the Early 90s: A Catalyst for Change

   The early 1990s witnessed a surge of interest in recommender systems. Automated collaborative filtering gained momentum, driven by pioneering projects like the GroupLens groups collaborative filtering for Usenet news and MITs Ringo for music artist recommendations. These initiatives laid the groundwork for collaborative filtering, a cornerstone of modern recommendation systems.

   The early 90s were marked by an awakening a realization that technology could be more than just functional; it could be intuitive, responsive, and deeply personalized. Recommender systems, driven by collaborative filtering, signaled a paradigm shift in how we engage with information and products. The stage was set for the grand narrative of personalized digital experiences.

   Industrys Recognition: Early Adoption of Recommender Systems

   At the forefront of this movement stood Amazon, an e- commerce giant that would redefine the rules of retail. As the company established its virtual presence, it quickly grasped the potential of recommender systems to transform the way customers discovered products. Amazons mission was not only to sell products but also to curate a personalized shopping experience that would keep customers engaged and satisfied.

   Central to Amazons approach was the concept of data- driven discovery. Instead of relying solely on customers to navigate through an overwhelming array of products, the company saw the value in leveraging data to present tailored recommendations. These recommendations would not only enhance the user experience but also facilitate upselling and cross-selling, thereby driving revenue.

   As the potential of recommender systems gained recognition, companies across various sectors began to explore their application. This recognition was

   underscored by the emergence of patents related to recommender technology. These patents, spanning industries ranging from technology to retail, showcased the widespread understanding of the transformative power of recommendation algorithms.

   Personalized Recommendation Systems: Unveiling Insights via Preference and Rating Data Collection

   In the rapidly evolving digital landscape, where information is abundant and choices are aplenty, recommender systems play a pivotal role in enhancing user experience by suggesting personalized content. From helping you discover your next favorite movie on Netflix to suggesting products on Amazon, these systems have become integral to our online interactions. In this article, we delve deep into the world of data collection for recommender systems, exploring how user preferences are captured and transformed into meaningful recommendations.

   • The Importance of Data Collection: Data, the fuel that powers these systems, is gathered through various user interactions and actions. Its essential not only for enhancing the recommendations but also for refining cross-selling strategies.

   • Objectives Gathering the Essence of User Preferences: At the heart of data collection lies the objective of capturing user preferences. Understanding what users like and how they engage with cotent is crucial for creating tailored recommendations. This data-driven approach bridges the gap between what users desire and what content providers offer.

   • Unveiling User Preference From Broad to Specific: User preference isnt a singular concept but rather a spectrum ranging from broad categories to specific item pairs. While some users might prefer action movies over romance, others might enjoy a combination of movies and popcorn. Collecting this diverse range of preferences helps create recommendations that cater to various tastes.

   • Navigating User Actions as Preference Indicators: Users express their preferences through a myriad of actions be it rating movies, writing reviews, making purchases, or simply clicking on links. These actions unveil insights into user inclinations and set the stage for personalized suggestions.

   • Explicit Ratings A Clear Expression of Preference: Explicit ratings are like virtual voices, allowing users to directly share their opinions. With options such as i) star ratings and

     ii) up/downvotes, users can convey their feelings about an item. The design of these interfaces,

   • including calibration, shapes how users perceive and communicate their preferences.

   • Crafting User Experience through Interface Design: Star ratings, whether with or without half- stars, are a popular choice across platforms like Netflix, GoodReads, and Amazon. Providing calibration tools aids users in comprehending the meaning behind each rating. Interface design influences the ease with which users communicate their preferences.

   • Implicit Data Uncovering Hidden Preferences: Not all preferences are explicitly communicated. Implicit data gathered from actions like clicks, purchases, reading/viewing time, and social interactions (follow/unfollow / likes/retweets), gives a glimpse into user interests that might otherwise remain unspoken.

   • Overcoming Challenges in Implicit Data: While implicit data can be a treasure trove of insights, it comes with challenges. These actions dont directly spell out preferences and can be influenced by factors other than liking an item. Preferences can also evolve, making this data a dynamic puzzle to decipher.

   • Unveiling Implicit Data Examples: Platforms like Pandora tailor music playlists based on song votes, while binary actions such as clicks indicate user interests. Understanding when an action was taken whether immediately after consumption or as a memory is crucial in interpreting implicit data. Examples of implicit datas prowess come to light.

   Here are some applications and a real-world example of a recommender system:

   E-commerce: Recommender systems are widely used in e-commerce platforms such as Amazon, eBay, and Alibaba to suggest products to customers based on their browsing history, purchase behavior, and preferences.

   Streaming Services: Platforms like Netflix, Spotify, and YouTube utilize recommender systems to recommend movies, TV shows, music, and videos to users based on their viewing or listening history, ratings, and preferences.

   Social Media: Social media platforms like Facebook, Instagram, and LinkedIn employ recommender systems to recommend friends, connections, posts, and content tailored to users' interests and activities.

   Online News and Content: News aggregators and content websites use recommender systems to personalize news articles, blog posts, and other content based on users' reading history, interests, and engagement.

   Travel and Hospitality: Booking websites such as Booking.com and Airbnb utilize recommender systems to suggest accommodations, flights, activities, and destinations based on users' preferences, search history, and behavior.

   Food and Dining: Food delivery platforms like Uber Eats and Grubhub leverage recommender systems to recommend restaurants, cuisines, and dishes based on users' location, past orders, and ratings.

   Job Search and Recruitment: Job search websites like LinkedIn and Indeed employ recommender systems to suggest job postings, companies, and career opportunities based on users' skills, experience, and preferences.

2. LITERATURE WORK

   A natural perspective for improving serendipity is to utilize side information like user profiles, content data, location information, etc. Murakami et al. compared the performance of improving serendipity between the Bayesian model and the keyword filtering method. The results reveal that keyword filtering can better balance accuracy and serendipity. Zhang et al. utilized music artist information by employing Latent Dirichlet Allocation technique and proposed two variants of item-based recommendation termed Community- Aware Auralist and Bubble-Aware Auralist, which can inject serendipity into music recommendation.

   Schedl et al. [12] proposed an algorithm which takes age, nation, style, and other factors into consideration while recommending music. The experimental results show that the additional information does improve user experience. Apart from side information, some efforts have been made for addressing the cold start problem from other perspectives [22], [23]. In [24], Wang et al. for the first time developed an active learning based framework for broadcast email prioritization, which exploits the collaborative filtering features, handles implicit feedback, and considers users time-sensitive responsiveness. The basic idea is to send the broadcast email to a small portion of users from the mailing list and then collect the time-sensitive feedbacks for predicting the priority of the email for the remaining users. This active learning framework is quite effective in addressing the completely cold start problem of broadcast email prioritization. Furthermore, a novel cross-domain recommendation framework was proposed for handling large numbers of mailing lists [25-29].

   Despite success in broadcast email prioritization, these methods are not directly applicable in e-commerce due to the reason that it is usually unsuitable to make a trial in a small portion of consumers for getting feedback.

   Zhou et al. (2010): This paper addresses the diversity- accuracy trade-off in recommender systems. It

   discusses methods for enhancing recommendation diversity without sacrificing accuracy, which is crucial for serendipitous discovery of products in e-commerce. Understanding this trade-off is essential for developing recommendation algorithms that balance both relevance and novelty.

   Ricci et al. (2015): "Recommender Systems Handbook" provides a comprehensive overview of recommendation techniques, including collaborative filtering, content-based filtering, and hybrid approaches. It serves as a valuable resource for understanding the foundations of recommendation systems and their application in e-commerce.

   Baltrunas & Ludwig (2012): This paper introduces contextual pre-filtering, a technique for incorporating contextual information into recommendation systems. Understanding how context influences user preferences is vital for enhancing serendipity in e-commerce recommendations, as contextual relevance can lead to more surprising and novel product suggestions.

   Amatriain et al. (2009): The study focuses on music recommendation and discovery in the long tail, addressing the challenge of recommending niche or less popular products. Serendipitous discovery often occurs in the long tail of products, where users encounter unexpected and unique items. Understanding the dynamics of the long tail is crucial for designing recommendation systems that promote serendipity.

   Chen et al. (2012): This paper discusses collaborative personalized tweet recommendation, exploring how collaborative filtering techniques can be applied to social media recommendations. Understading user interactions and behaviors on social platforms is essential for developing context-aware recommendation systems that leverage social context for serendipitous discovery.

   Said et al. (2014): The study evaluates a context-aware content-based music recommender, focusing on user- centric evaluation methodologies. Understanding how users perceive and interact with recommendations is crucial for assessing the effectiveness of recommendation algorithms in promoting serendipity and user satisfaction.

   Cantador et al. (2015): The paper explores the relationship between diversity in rank aggregation and context similarity for recommender systems. Understanding how to measure and incorporate diversity into recommendation algorithms is essential for promoting serendipitous discovery and enhancing user satisfaction in e-commerce settings.

   Shani & Gunawardana (2011): The paper provides insights into evaluating recommendation systems, focusing on user-centric evaluation metrics and methodologies. Understanding how to effectively evaluate recommendation algorithms is essential for

   assessing their impact on serendipitous discovery, user engagement, and overall satisfaction in e-commerce environments.

3. COLLABORATIVE FILTERING TYPES

   1. User-Based Collaborative Filtering:

      • Concept: User-based CF relies on the idea that users who have similar preferences in the past will continue to have similar preferences in the future. It identifies users with similar item ratings and recommends items that those similar users have liked.

      • Example: Suppose we have two users, Alice and Bob. If Alice has rated movies A, B, and C highly, and Bob has also rated movies A and B highly, the system might recommend movie C to Bob based on Alice's preferences.

   2. Item-Based Collaborative Filtering:

      • Concept: Item-based CF focuses on the similarity between items rather than users. It recommends items similar to those a user has already interacted with.

      • Example: If a user has watched and liked movies X and Y, the system identifies other movies that are similar to X and Y (based on user ratings or other features) and recommends them.

   3. Matrix Factorization Techniques:

      • Concept: Matrix factorization decomposes the user- item interaction matrix into latent factors (e.g., user preferences and item features). These latent factors capture underlying patterns and allow for personalized recommendations.

      • Example: Singular Value Decomposition (SVD) and Alternating Least Squares (ALS) are popular matrix factorization methods.

   4. Neighborhood Models:

      • Concept: Neighborhood models combine user-based and item-based approaches. They find a neighborhood of similar users or items and use their preferences to make recommendations.

      • Example: k-Nearest Neighbors (k-NN) identifies the k most similar users or items and aggregates their preferences.

   5. Hybrid Approaches:

      • Concept: Hybrid models combine CF with other techniques (e.g., content-based filtering, deep learning) to address limitations and enhance recommendation quality.

      • Example: Hybrid CF-CB models use both user-item interactions and item content features.

   Challenges and Limitations of Collaborative Filtering

   1. Data Sparsity:

      • Issue: Collaborative filtering relies on user-item interactions to make recommendations. However, in real-world scenarios, the data matrix representing

        these interactions is often sparse. Most users have rated only a small fraction of the available items, leading to missing values.

      • Insight: Sparse data hampers the effectiveness of CF algorithms, as they struggle to find meaningful patterns when faced with insufficient information.

      • Example: Imagine a movie recommendation system where a user has rated only a handful of films out of thousands available. The system must still provide relevant suggestions despite the sparsity.

   2. Cold Start Problem:

      • Issue: The cold start problem occurs when a new user or item enters the system. Since CF relies on historical interactions, it struggles to make accurate recommendations for these newcomers.

      • Insight: Without sufficient data, collaborative filtering algorithms cannot infer user preferences or item similarities effectively.

      • Example: A music streaming service adds a fresh artist to its catalog. Initially, there are no user ratings or interactions for this artist, making it challenging to recommend their songs.

   3. User and Item Biases:

      • Issue: CF assumes that users and items are equally distributed, but in reality, biases exist. Some users tend to rate more generously, while others are more critical. Similarly, certain items receive disproportionate attention.

      • Insight: Biases can skew recommendations, favoring popular items or overly active users.

      • Example: In a book recommendation system, bestsellers might dominate the recommendations due to their popularity, overshadowing lesser-known gems.

   4. Scalability:

      • Issue: As the number of users and items grows, the computational complexity of CF algorithms increases significantly.

      • Insight: Scalability challenges arise when computing similarity scores or updating recommendations in large-scale systems.

      • Example: An e-commerce platform with millions of products faces scalability issues when generating personalized product suggestions for each user.

   5. Lack of Diversity:

      • Issue: CF tends to recommend items similar to those a user has already interacted with. While this personalization is desirable, it may lead to a filter bubble, limiting exposure to diverse content.

      • Insight: Lack of diversity can result in monotony and hinder serendipitous discoveries.

      • Example: A movie streaming service consistently suggests films from the same genre, preventing users from exploring other categories.

4. CONTEXT-AWARE COLLABORATIVE

   FILTERING APPROACH

   The Context-Aware Collaborative Filtering Approach combines insights from innovator-based preferences and contextual information to enhance the relevance and serendipity of product recommendations in e- commerce settings. Here's a detailed description of the approach:

   Identification of Innovators: The approach begins with identifying innovators within the user community. Innovators are users who demonstrate a propensity for exploring novel and unique products. This identification process may involve analyzing user interaction patterns, purchase histories, and engagement with new or niche items.

   Contextual Information Integration: Contextual information, such as time, location, and user behavior, is integrated into the recommendation process. This contextual awareness enriches the relevance and timeliness of recommendations by considering the situational context in which users interact with the platform. For example, recommendations may vary based on the time of day, user location, or recent browsing history.

   Collaborative Filtering Framework: The collaborative filtering framework forms the backbone of the recommendation system. It leverages both innovator- based insights and contextual cues to generate personalized product recommendations. Collaborative filtering algorithms analyze user-item interactions and identify patterns of similarity among users or items to predict relevant recommendations for each user.

   Recommendation Generation: Recommendations are generated by combining innovator-driven preferences and contextual information within the collaborative filtering framework. The system considers the preferences of innovators as well as the contextual factors influencing user behavior to generate personalized and serendipitous recommendations. This process ensures that recommendations are both relevant to the user's interests and aligned with the current context.

   Evaluation Methodologies: Various evaluation methodologies are employed to assess the effectiveness and performance of the context-aware collaborative filtering approach. Evaluation metrics such as

   serendipity, relevance, accuracy, and user satisfaction are used to measure the quality of recommendations generated. User studies, A/B testing, and offline evaluation techniques may be utilized to validate the efficacy of the approach.

   The Context-Aware Collaborative Filtering (CACF) Approach framework integrates innovator-based insights and contextual information to enhance the relevance and serendipity of product recommendations in e-commerce. Here's a framework outlining its key components:

   Data Collection and Preprocessing:

   Gather user-item interaction data, including ratings, purchases, and browsing history. Collect contextual information such as time, location, device, and user demographics. Preprocess the data to handle missing values, normalize ratings, and encode contextual features.

   Innovator Identification: Analyze user interaction patterns to identify innovators who exhibit a propensity for exploring novel and unique products.

   Consider factors such as early adoption of new items, diversity in purchase behavior, and engagement with niche products.

   Contextual Feature Extraction:Extract contextual features from the data, including temporal, spatial, and user-specific attributes. Represent contextual information in a structured format suitable for collaborative filtering algorithms.

   Collaborative Filtering Algorithm Selection: Choose a collaborative filtering algorithm suitable for incorporating contextual information. Consider techniques such as neighborhood-based, matrix factorization, or deep learning-based collaborative filtering.

   Context-Aware Recommendation Generation: Incorporate innovator-based preferences and contextual features into the collaborative filtering algorithm. Generate personalized recommendations for each user by considering both historical interactions and current context.

   Evaluation and Validation: Evaluate the performance of the CACF approach using appropriate metrics such as serendipity, accuracy, diversity, and user satisfaction. Conduct offline experiments, A/B testing, or user studies to validate the effectiveness of the recommendations generated.

   Model Optimization and Refinement: Fine-tune the CACF model parameters to improve recommendation quality and performance. Incorporate user feedback and iterative experimentation to iteratively refine the model.

   Deployment and Integration: Deploy the CACF model within the e-commerce platform's recommendation engine.

   Integrate the recommendation system seamlessly into the user interface to provide personalized and context- aware product suggestions.

5. RESULTS AND ANALYSIS

   Collaborative filtering algorithms can be evaluated using various metrics to assess their performance. Here are some common evaluation metrics for collaborative filtering:

   Precision: Precision measures the proportion of recommended items that are relevant to the user. It is calculated as the number of relevant recommendations divided by the total number of recommendations provided.

   Recall: Recall measures the proportion of relevant items that are successfully recommended to the user. It is calculated as the number of relevant recommendations divided by the total number of relevant items.

   F1 Score: The F1 score is the harmonic mean of precision and recall, providing a single metric that balances both measures. It is calculated as:

   Coverage: Coverage measures the proportion of items in the catalog that are successfully recommended to users. It reflects the diversity of recommendations provided by the algorithm.

   Prediction Accuracy: Prediction accuracy measures the overall accuracy of the recommendations generated by the collaborative filtering algorithm. It can be evaluated using metrics such as Mean Absolute Error (MAE) or Root Mean Squared Error (RMSE) for rating prediction tasks.

   Datasets:

   Amazon Reviews: Amazon provides datasets containing user reviews and ratings for products across various categories, including electronics, books, and clothing. Contains 34,686,770 Amazon reviews from 6,643,669 users on 2,441,053 products, from the Stanford Network Analysis Project (SNAP). This subset contains 1,800,000 training samples and 200,000 testing samples in each polarity sentiment.

   Table 1: Precision,Recall and F1-score comparisons

   	Precission	Recall	F1-Score
   Existing	89.63	86.24	89.27
   Proposed	91.54	87.12	90.41

   Fig 1: comparison graph of Precision,Recall and F1- score

   Table 2: Predictive accuracy and coverage of items

   	Predictive accuracy		Coverage
   	MAE	RMSE
   Existing	53.78	61.25	82.35
   Proposed	53.98	62.14	89.39

   Fig 2: comparison graph of Predictive accuracy and coverage of items

6. CONCLUSION

Navigating E-commerce Serendipity: Leveraging Innovator-Based Context-Aware Collaborative Filtering for Product Recommendations" underscores the potential of combining innovator-driven insights with context-aware collaborative filtering in e- commerce settings. By harnessing the power of both innovative user preferences and contextual information such as time, location, and user behavior, this approach aims to unlock serendipitous product recommendations that resonate with users in diverse and dynamic situations. Through this fusion of innovation and contextual relevance, e-commerce platforms can strive to deliver more personalized, timely, and engaging shopping experiences, ultimately enhancing customer satisfaction and driving business success in an ever- evolving digital landscape. The recommendations generated by the proposed algorithm exhibit both relevance and novelty, making the system highly innovative and serendipitous. The synergy between relevance and novelty enhances the overall effectiveness of the system significantly. To assess the system's performance comprehensively, a novelty score evaluation metric was introduced and computed, resulting in an average novelty score of 87% for the Amazon Reviews Dataset dataset.

REFERENCES

1. Sunder Reddy, K. S. ., Lakshmi, P. R. ., Kumar, D. M. .,

   Naresh, P. ., Gholap, Y. N. ., & Gupta, K. G. . (2024). A Method for Unsupervised Ensemble Clustering to Examine Student Behavioral Patterns. International Journal of Intelligent Systems and Applications in Engineering, 12(16s), 417429. Retrieved from

   https://ijisae.org/index.php/IJISAE/rticle/view/4854.

2. Nagesh, C., Chaganti, K.R. , Chaganti, S. , Khaleelullah, S., Naresh, P. and Hussan, M. 2023. Leveraging Machine Learning based Ensemble Time Series Prediction Model for Rainfall Using SVM, KNN and Advanced ARIMA+ E- GARCH. International Journal on Recent and Innovation Trends in Computing and Communication. 11, 7s (Jul. 2023), 353358. DOI:https://doi.org/10.17762/ijritcc.v11i7s.7010.

3. S. Khaleelullah, P. Marry, P. Naresh, P. Srilatha, G. Sirisha and

   C. Nagesh, "A Framework for Design and Development of Message sharing using Open-Source Software," 2023 International Conference on Sustainable Computing and Data Communication Systems (ICSCDS), Erode, India, 2023, pp. 639-646, doi: 10.1109/ICSCDS56580.2023.10104679.

4. Balakrishnaa, Chinnala, and Tryambak Hirwarkarb. "Hybrid Broadcast Encryption and Group Key Agreement Protocol with Precise Cipher Texts." Turkish Journal of Computer and Mathematics Education Vol 12.5 (2021): 984-988.

5. Naresh, P., & Suguna, R. (2019). Association Rule Mining Algorithms on Large and Small Datasets: A Comparative Study. 2019 International Conference on Intelligent Computing and Control Systems (ICCS). DOI:10.1109/iccs45141.2019.9065836.

6. Naresh, P., & Suguna, R. (2021). Implementation of dynamic and fast mining algorithms on incremental datasets to discover qualitative rules. Applied Computer Science, 17(3), 82-91. https://doi.org/10.23743/acs-2021-23.

7. M. I. Thariq Hussan, D. Saidulu, P. T. Anitha, A. Manikandan and P. Naresh (2022), Object Detection and Recognition in Real Time Using Deep Learning for Visually Impaired People. IJEER 10(2), 80-86. DOI: 10.37391/IJEER.100205.

8. Suguna Ramadass and Shyamala Devi 2019 Prediction of Customer Attrition using Feature Extraction Techniques and its Performance Assessment through dissimilar Classifiers Springers book series Learning and Analytics in Intelligent Systems, Springer.

9. Balakrishna, C. ., Sapkal, A. ., Chowdary, B., Rajyalakshmi, P., Kumar, V. S. ., & Gupta, K. G. . (2023). Addressing the IoT Schemes for Securing the Modern Healthcare Systems with Block chain Neural Networks. International Journal on Recent and Innovation Trends in Computing and Communication, 11(7s), 347352. https://doi.org/10.17762/ijritcc.v11i7s.7009

10. Naresh P, Shekhar GN, Kumar MK, Rajyalakshmi P. Implementation of multi-node clusters in column oriented database using HDFS. Empirical Research Press Ltd. 2017; p. 186.

11. P.Naresh,et.al., Implementation of Map Reduce Based Clustering for Large Database in Cloud, Journal For Innovative Development in Pharmaceutical and Technical Science,vol.1,pp 1-4,2018.

12. B.M.G. Prasad, P. Naresh, V. Veeresh, Frequent Temporal Patterns Mining With Relative Intervals,International Refereed Journal of Engineering and Science ,Volume 4, Issue 6 (June 2015), PP.153-156.

13. K. R. Chaganti, U. S. Ramula, C. Sathyanarayana, R. Changala,

    N. Kirankumar and K. G. Gupta, "UI/UX Design for Online Learning Approach by Predictive Student Experience," 2023 7th International Conference on Electronics, Communication and Aerospace Technology (ICECA), Coimbatore, India, 2023,

    pp. 794-799, doi: 10.1109/ICECA58529.2023.10395866.

14. T. Aruna, P. Naresh, A. Rajeshwari, M. I. T. Hussan and K. G. Guptha, "Visualization and Prediction of Rainfall Using Deep Learning and Machine Learning Techniques," 2022 2nd International Conference on Technological Advancements in Computational Sciences (ICTACS), Tashkent, Uzbekistan, 2022, pp. 910-914, doi: 10.1109/ICTACS56270.2022.9988553.

15. Balakrishna, Ch & Hirwarkar, Tryambak. (2021). SYSTEMATIC REVIEW OF CONTRIBUTORY BROADCAST ENCRYPTION WITH EFFICIENT

    ENCRYPTION AND SHORT CIPHER TEXTS. Test

    Engineering and Management. 83. 26508 -26516.

16. V. Krishna, Y. D. Solomon Raju, C. V. Raghavendran, P. Naresh and A. Rajesh, "Identification of Nutritional Deficiencies in Crops Using Machine Learning and Image Processing Techniques," 2022 3rd International Conference on Intelligent Engineering and Management (ICIEM), London, United Kingdom, 2022, pp. 925-929, doi: 10.1109/ICIEM54221.2022.9853072.

17. P, N., & R Suguna. (2022). Enhancing the Performance of Association Rule Generation over Dynamic Data using Incremental Tree Structures. International Journal of Next- Generation Computing, 13(3). https://doi.org/10.47164/ijngc.v13i3.806.

18. B. Narsimha, Ch V Raghavendran, Pannangi Rajyalakshmi, G Kasi Reddy, M. Bhargavi and P. Naresh (2022), Cyber Defense in the Age of Artificial Intelligence and Machine Learning for Financial Fraud Detection Application. IJEER 10(2), 87-92. DOI: 10.37391/IJEER.100206.

19. Naresh, P., & Suguna, R. (2021). IPOC: An efficient approach for dynamic association rule generation using incremental data with updating supports. Indonesian Journal of Electrical Engineering and Computer Science, 24(2), 1084. https://doi.org/10.11591/ijeecs.v24.i2.pp1084-1090.

20. P.Rajyalakshmi et.al., Deep Learning Based NLP and LSTM Models for Sentiment Classification of Consumer Tweets

21. Ravi, C., Raghavendran, C. V. ., Satish, G. N. ., Reddy, K. V. ., Reddy, G. . K. ., & Balakrishna, C. . (2023). ANN and RSM based Modeling of Moringa Stenopetala Seed Oil Extraction: Process Optimization and Oil Characterization. International Journal on Recent and Innovation Trends in Computing and Communication, 11(7s), 329338. https://doi.org/10.17762/ijritcc.v11i7s.7007.

22. P. Naresh, S. V. N. Pavan, A. R. Mohammed, N. Chanti and M. Tharun, "Comparative Study of Machine Learning Algorithms for Fake Review Detection with Emphasis on SVM," 2023 International Conference on Sustainable Computing and Smart Systems (ICSCSS), Coimbatore, India, 2023, pp. 170-176, doi: 10.1109/ICSCSS57650.2023.10169190.

23. Hussan, M.I. & Reddy, G. & Anitha, P. & Kanagaraj, A. & Pannangi, Naresh. (2023). DDoS attack detection in IoT environment using optimized Elman recurrent neural networks based on chaotic bacterial colony optimization. Cluster Computing. 1-22. 10.1007/s10586-023-04187-4.

24. P. Naresh, P. Srinath, K. Akshit, M. S. S. Raju and P. VenkataTeja, "Decoding Network Anomalies using Supervised Machine Learning and Deep Learning Approaches," 2023 2nd International Conference on Automation, Computing and Renewable Systems (ICACRS), Pudukkottai, India, 2023, pp. 1598-1603, doi: 10.1109/ICACRS58579.2023.10404866.

25. Pannangi, Naresh & Ramadass, Suguna. (2019). Implementation of Improved Association Rule Mining Algorithms for Fast Mining with Efficient Tree Structures on Large Datasets. International Journal of Engineering and Advanced Technology. 9. 5136-5141. 10.35940/ijeat.B3876.129219.

26. C. B. . Koushik Reddy Chaganti, C. Sasikala, G.Sirisha, Shravani Amar, Ravindra Changala, Koppuravuri Gurnadha Gupta, Improved Plant Phenotyping System Employing Machine Learning Based Image Analysis, Int J Intell Syst Appl Eng, vol. 12, no. 3, pp. 24152421, Mar. 2024.

27. Rajyalakshmi, P., Chaganti, K. R., Chaganti, S., & Anusha, A. (2023). Deep learning based NLP and LSTM models for sentiment classification of consumer tweets. In Journal of Engineering Sciences (Vol. 14, Issue 05(S), pp. 126128) https//jespublication.com/specialissue/2023-V14I5016.pdf.

28. P. Rajyalakshmi, M. K. Kumar, G. U. Maheswari, and P. Naresh, Implementation of clustering algorithms for real time large datasets, International Journal of Innovative Technology and Exploring Engineering, vol. 8, no. 11, pp. 23032304,

    2019, doi: 10.35940/ijitee.c2570.0981119.

29. P. Rajyalakshmi, C. Balakrishna, E. Swarnalatha, B. S. Swapna Shanthi and K. Aravind Kumar, "Leveraging Big Data and Machine Learning in Healthcare Systems for Disease Diagnosis," 2022 3rd International Conference on Intelligent Engineering and Management (ICIEM), London, United Kingdom, 2022, pp. 930-934, doi: 10.1109/ICIEM54221.2022.9853149.