
Worldwide advertising companies are increasingly able to advertise and promote products through social media that has generated a large volume of information and this data can be analyzed by using data binding systems, data automation systems, and data analytics. These provide an opportunity to attackers to create ad-click frauds that are suspicious activities that are done by generating automated bots or misguiding users by encouraging prohibited clicks on a part of pictures in a social media advertisement. These kinds of fraudulent operations modify the performance of a model, misleading advertising companies, and that will create big financial losses all over the world. In this paper we developed a new novel Hybrid Deep Learning Technique (HDLT), in this model we used three methods that are Extreme Gradient Boosting (XG Boost), Recurrent Neural Networks (RNN), Convolution Neural Networks (CNN) to identify and protect from fraudulent ad clicks. The model uses robust classification capabilities of XG Boost technique, RNN temporal sequence modeling capabilities, and binds the CNN spatial capability to detect and analyze patterns of fraudulent clicks on social media advertisements. Experiments are conducted by using two types of datasets, one is proprietary advertising logs and other one is Talking Data Ad-Tracking Fraud Dataset. With help of all these three techniques proposed method shows 99.34
This research grants a holistic scientific metric evaluation of Artificial Intelligence (AI) research in e-commerce. It is based on 851 Scopus-indexed articles published from 2000 to 2025. Bibliometric tools such as VOSviewer and Biblioshiny were utilized to highlight various aspects of the literature, such as publishing trends, sources of influence, leading authors, and collaborative networks. The results show that research on the use of AI in e-commerce is booming globally, which is mainly driven by India, China, and the United States. Besides that, business, computer science, and social science are the main areas of contribution. The co-occurrence and thematic analyses reveal the main subjects of the research, e.g. personalization, digital trust, intelligent systems, consumer behaviour, and decision support. However, conceptual fragmentation is the main characteristic of the field, with only a few instances of focusing on sustainability, ethics, and cross-cultural aspects. Besides discussing theoretical concepts and providing practical advice, the paper also points to the future directions of research which are developing responsible and integrated AI applications in the sphere of digital commerce.
India is a land of traditional medicinal systems, and the Marathwada region blooms with ethnobotanical knowledge. However, most of this knowledge remains untouched in descriptive text and other scattered sources, which limits its use directly in computational systems. In this case study we present an analysis of 21 medicinal plants using Exploratory Data Analysis (EDA) for extracting and analyzing significant therapeutic information. We used frequency analysis, co-occurrence matrices, and clustering techniques (PCA and K-means). Although the dataset was limited in size, over 180 ailments and 200 disease categories were identified, which resulted in more than 3800 unique structured plant-ailment and plant-disease associations. We also uncovered the dominant therapeutic domains like digestive, immunity, respiratory, skin-health, and some plant-specific roles. This study demonstrates that even the smallest structured dataset can be explored for therapeutic functions and EDA helps in knowledge discovery by laying the groundwork for digital herbal databases, intelligent recommendation systems and AI-ready healthcare applications.
This paper introduces an AI-assisted precision farming system, developed as a Streamlit web application, combining machine learning models, real-time environmental inputs, and voice interfaces for sustainable and data-driven agriculture. The system uses a Random Forest Classifier for crop suggestions, trained on an Excel database with soil nutrient values, pH, and weather factors. A Random Forest Regressor estimates crop yield based on encoded categorical variables (state, crop, season), year, and area of cultivation. Leaf disease detection using a CNN (Convolutional Neural Network) enables early diagnosis of diseases from images uploaded by farmers. A smart crop calendar from Excel data offers planting and harvesting sessions, irrigation practices, and fertiliser recommendations for proper crop yielding, and real-time weather tracking through the OpenWeatherMap API recommendations according to current and future weather conditions. A voice interface, along with an AI chatbot by OpenAI GPT-3.5, provides relevant answers to agricultural questions of farmers. The system provides an interactive interface for rural farmers. The system enhanced accuracy for crop identification, yield prediction, and system usability, with the potential to incorporate drone image analysis and NDVI for crop analysis.
The circulation of counterfeit medicines is an international problem and occurs in areas of poor control. False medicines are not only an effective risk threat to the population of a particular country, but they also undermine the credibility of healthcare infrastructures. The real-life issues concern the failure to track medicines efficiently through the supply chain, the incompetence of transactions to be transparent, and the possibility of determining fake goods before they are made available to the patients. Moreover, the fact that everything is done manually and that there is no cohesive system to allow verifying the authenticity interferes with workflows, increases delays, and puts the risk of fake medicines entering the market further. Such problems are further exacerbated by the lack of a unified platform where all the entities in the place (manufacturers, distributors, healthcare providers, patients) can interact in a safe and transparent way. To address the problems, this project intends to offer a blockchain-based medicine supply chain management system. The QR codes with a unique digital identifier of each batch of medicines are assigned in a secure blockchain ledger. This makes the process have full traceability, whereby stakeholders will be able to track the movement of medicines in real-time through the production and delivery process. Smart contracts robotize the checking mechanism, mark false medicines and keep the platforms in line with the regimes. Eliminating manual interventions and storing the information on an unchangeable blockchain, the system increases transparency, decreases fraud and minimizes human mistakes. This paper addresses the authenticity, traceability and transparency that can provide a secure way to fight fake medicines and restore confidence in the healthcare system. It forms a current model of medicine tracking, besides providing a better, quicker, and more secure supply of real medicines to patient