Recycling and the effectiveness of recycling programs in the United States present an ongoing crisis, with a growing volume of recyclable waste ending up in landfills each year. Existing recycling programs often face challenges, either due to their inconvenience or complexities. RecyLink is an innovative application that addresses this issue by streamlining the management of existing programs, ultimately aiming to reduce the amount of recyclable waste. RecyLink achieves this by utilizing local schools as convenient drop-off zones where students and their family members can easily deposit plastic bottles. A machine learning model is employed to scan and count the bottles as they are dropped off, maintaining a continuous tally that can be monitored by local recycling centers. While the program is still in its early stages and hasn't undergone extensive testing, initial data indicates promising results. RecyLink demonstrates the ability to recognize plastic bottles, determine their quantity accurately, and log this information appropriately. In essence, RecyLink represents a promising beginning in addressing the recycling challenges in the United States, and its success may extend to a global scale.

This research explores the intricate relationship between sentiment analysis, stock market dynamics, and Environmental, Social, and Governance (ESG) based investment analytics, harnessing sentiment as a predictive tool for stock price movements. Leveraging Twitter data, Natural Language Processing (NLP), TextBlob, and the scikit-learn RandomForestRegressor, in combination with machine learning algorithms, the study evaluates public sentiment's impact on stock prices, offering valuable insights to investors and risk managers. Moreover, the findings elucidate the potential to enhance ESG-based investment analytics by incorporating sentiment-derived insights into investment decision-making processes, which is particularly pertinent given the increasing market focus on sustainable investing. Experimental results unveil the potential of sentiment analysis in forecasting stock price changes and augmenting ESG investment strategies, underlining its utility as both a forecasting instrument and a risk management mechanism. However, the research also identifies challenges, including limitations of the Twitter API and the need for data refinement. Strategies to address these challenges are discussed, emphasizing the importance of diversifying data sources and enhancing data quality. This study advances our understanding of sentiment analysis in financial markets and its applicability to ESG-based investment analytics, offering data-driven guidance to navigate the complexities of the stock market landscape. Ultimately, it highlights the promising prospect of integrating social media sentiment analysis with machine learning for more informed stock market predictions, risk management, and sustainable investment strategy formulation.

This article explores a design method that satisfies the user's operational experience and emotional experience in order to solve the consumer's perceptual demand for products.By studying user behavior, different user needs are extracted; combined with context analysis, a user behavior map is constructed, so as to extract different contexts of product use, and then discover contact points that can be improved in user behavior.Finally, according to the Kano model, the demand attribute categories are reasonably divided, and the user needs are analyzed, and the main functional factors that affect the user's satisfaction with the home lighting products are obtained, so as to determine the design direction, and evoke a deeper emotional resonance with the shape, function and use method.According to the final design plan, it has a certain significance to the design method of traditional wood art lamps by improving the user experience and meeting the user's demand for product operation functions and emotional experience.

Having access to sufficient water usage data real-time is crucial not only to finding water leaks and preventing water bills (which can cost up to thousands of dollars) but also to tracking water usage and savings. However, to record a California household's water usage, a worker from the respective water district must personally check the water meter in order to update the data at a monthly frequency. This is because the current water infrastructure in California is outdated while also being expensive to replace. Additionally, while certain cities such as San Jose has Advanced Metering Systems for water installed, due to a lack of budget, it can be difficult for government agencies to develop a solution that can be implemented statewide, then nationwide, in an estimated 2-3 years from now. This paper proposes an IoT-based smart water monitor that utilizes computer vision to record water data, then compares the data's deviance from the predicted usage to check for water leaks. We applied our application to monitoring the water usage of my own household and conducted a qualitative evaluation of the approach. The results show that the water monitor is an effective and affordable way to water management.

As obesity becomes increasingly common worldwide [1], more people want to lose weight to improve their health and image. According to the Centers for Disease Control and Prevention (CDC), long-term changes in daily eating habits (such as regarding food/ nutrition type, calorie intake) are successful at keeping weights off [2]. Therefore, it would be helpful to have an artificial intelligence (AI) mobile program that identifies the types of food the user consumes and automatically calculates the total calories. This paper examines the development and optimization of an 11-categorical food classification model based on the Mobile-Net neural network using Python. Specifically, it classifies any food image as one of bread, dairy, dessert, egg product, fried food, meat, noodles, rice, seafood, soup, or fruit/vegetables. Methods of optimization include data preprocessing and learning rate and batch size adjustments. Experimental results show that scaling image inputs to standard size (Python Numpy resize) function), 300 training epochs, dynamic learning rate (start with 0.001 and *0.1 for every 30 epochs), and a batch size of 16 yields our best model of 83.44% accuracy.

With the increasing popularity of Artificial Intelligence (AI) and Machine Learning (ML), developing AI-based applications is in high demand in various industries. However, the AI development is still based on traditional programming frameworks and languages, which prevents domain experts from contributing to it without collaborating with developers. This research is to show how graphical software allows users from many domain (e.g., Doctors, Accountants, Advertisers) to build AI applications, train AI models without any prior knowledge of programming, and many of its unnecessary concepts. Using nodes and connectors as the primary graphical components, the application, GraphicalAI, is to show how graphics can be designed in a way to easily prototype any kinds of AI models. To enable domain experts to design AI models using the power of graphics and our human vision.

Despite the fact that the main objective of an Emergency Department (ED) is to treat seriously ill patients, quality of care degrades due to prolonged waiting time. This paper investigates overcrowding as the main bottleneck in healthcare system performance and addresses patients' dissatisfaction and low quality of EDs' services on wait time. In this paper, we present empirical study, data analytics and application towards answering whether there is a significant correlation between medical immediacy and the length of wait time. This research is based on a series of feature estimation on 27 features extracted from publicly available dataset from Disease Control and Prevention (CDC) to find the extent of their correlation to prolonged wait time. The results suggest that in 2015 more than 63% of EDs visits were related to non-urgent medical issues. Our findings indicate that overcrowding occurs due to the presence of very large number of patients with non-urgent medical problems.

This paper presents a solution for the predictions of flowering times concerning specific types of flowers. Since flower blooms are necessarily related to the local environment, the predictions (in months), are yielded by using machine learning to train a model considering the various environmental factors as variables. The environmental factors, which are temperature, precipitation, and the length of day, contribute to the chronological order of flowering periods. The predictions are accurate to a fraction of a month, and it can applied to control the flowering times by changing the values of the variables. The result provides an example of how data mining and machine learning presents itself to be a useful tool in the agricultural or environmental field.

As the development of economy and industry, air quality decreases as one of the exchanges of our achievements. Although air pollution has already been considered as a global and critical issue over the past decades, there has not been much innovation on the way people monitor and check the quality. Most of the air quality data today is provided by government or professional sensors set up in cities, which does not provide more detailed status in smaller geo locations with finer granularity, such as specific villages, schools, and shopping malls. In this project, we use machine learning to make a mathematical model which could be used to predict the air quality for small geo locations with accuracy and fine granularity. Through series of experiments and comparisons, the most accuracy mathematical model was found, which had a difference percentage less than 20

This paper presents a distributed, compressive multiple target localization and tracking system based on wireless fiber-optic sensors. This research aims to develop a novel, efficient, low data-throughput multiple target tracking platform. The platform is developed based on three main technologies: (1) multiplex sensing, (2) space encoding and (3) compressive localization. Multiplex sensing is adopted to enhance sensing efficiency. Space encoding can convert the location information of multi-target into a set of codes. Compressive localization further reduces the number of sensors and data-throughput. In this work, a graphical model is employed to model the variables and parameters of this tracking system, and tracking is implemented through an Expectation-Maximization (EM) procedure. The results demonstrated that the proposed system is efficient in multi-target tracking.

Machine Learning allows systems to learn and improve automatically from experiences without hand-coding. Thus, in recent years, many technology companies have been developing such application if Artificial Intelligence, from face recognition by Facebook, to the AlphaGo program by Google. The irrigation systems in the market nowadays mostly allow users to set them to a certain amount of water and at specific time intervals. However, there are usually more than one type of plants in a garden, and each species requires different amount of water. In order to resolve this issue, in this paper, we have developed an irrigation system, with the use of deep learning, that is able to adjust the amounts of water foe each type pf plant through plants recognition. There are two main parts of the solution, the software and the hardware. The prior is connected with cameras to undergo plant recognition, and utilizes database to find the suitable amount of water; the latter controls the amount of water that is able to flow out.

The review system promoted the prosperity of the E-commerce market, but it has also been improperly exploited by spammers who found it profitable to write fake reviews to mislead the consumers. Detecting spams accurately has been a challenging problem, since spammers alter their writing style and imitate normal users. The majority of the existing work tried to solve this problem by designing an indicator system using behavioral features, which often suffers from the lack of large-scale labeled datasets. This paper proposes a novel user relation graph model based on a bipartite graph built directly from the review data and introduces two novel algorithms called Finding Abnormal Dimensions by Kurtosis function (FADK) and Finding Abnormal Dimensions by Shapiro-Wilk test (FADSW) to find small groups of spammers in the large user-relation graph. FADK focus on each eigenvector and its neighborhood, which uses the kurtosis function as a crucial measure. In FADSW, the hypothesis test theory is introduced to solve this problem for the first time. This combination is a novel interdisciplinary research methodology. To get an unbiased performance evaluation, we evaluated our two algorithms using two different real-world datasets: 1) Amazon dataset from the US; and 2) JD.com dataset from China. While both algorithms showed relatively high distinguishing power, FADSW outperforms FADK on both the JD and the Amazon datasets.

Online reviews play a crucial role in helping consumers to make purchase decisions. However, a severe problem Internet Water Army (a large amount of paid posters who write inauthentic reviews) emerge in many E-commerce websites recently which dramatically undermines the value of user reviews. Although the word Internet Water Army originated from China, some other countries also suffered from this problem. Many organized underground paid poster groups found it extremely profitable to mislead the consumers by writing fake reviews. It had become more and more challenging to accurately detect the water army who could alter their writing style. In this paper, we design a comprehensive set of features to compare paid posters against normal users on different dimensions. Then we build an ensemble detection model of seven different algorithms. Our model has reached 0.726 in AUC measure and 0.683 in F1 measure on JD dataset, 0.926 in AUC measure and 0.871 in F1 measure on Amazon dataset, which outperforms previous studies. Our work provides some practical solutions and guidance to this severe problem for the whole E-commerce industry.

Within the context of the Internet-of-Things (IoT), the number of interconnected devices is increasing dramatically and allowing for access to physical data that was previously unimaginable. Physical data is rapidly changing which makes it important to keep networking connections active. Any drop in communication can lead to the loss of sensitive data. A redundant network connection is an attempt to utilize common networking solutions in order to decrease the likelihood of network downtime. It does this by adding a new level of abstraction to networking, allowing data to be sent over multiple networking solutions as if it were a single network, as well as an intelligent decision engine to determine the most optimized and reliable connection to use dynamically.

Crowdsourcing is an important computing technique that taps into the collective intelligence of the public at large to complete business-related tasks and solve many real-time problems. It is changing the way we work, hire, research, make and market. Many developing nations are trying to take advantage of crowdsourcing for information notification to make cost effective system, like real-time transit system, disaster notification system and other services which are available to the masses. However, many of them are still not able to completely benefit from it compared to developed nations. In this paper, we have identified a series of limitations of using crowdsourcing for information gathering and providing real-time notification in developing countries due to their unstable electronic communication infrastructure, their lack of contribution, lack of crowdsource (participating people), less exposure to English language, and unawareness of crowdsourcing. We proposed, and demonstrated, a solution to overcome these limitations by developing a prototype which uses SMS as a reliable method for providing real-time notification and information gathering. Our prototype uses prediction algorithms to fill the gaps in real-time notification. It also uses the prediction of a user's behavior to provide a better reward and motivational platform, as well as good usability.

Distractions have become an increasingly prevalent part of everyday life. These distractions, such as video games, social media, and streaming video services have a negative impact on productivity. Authors frequently suffer writers block due to an inability to focus. For this reason, distraction-free word processors have become quite popular. However, most of them have limited functionality and do not seek to eliminate many of the common distractions writers face. In this paper, we will discuss the semantics of good writing habits and present a custom designed writing system aimed at keeping the user on task. To encourage focus on a proper writing process, the software will have a required full screen display and limited text formatting options. Furthermore, the user will be able to take advantage of aids that encourage correct writing behavior.

Optimizing the deployment of software in a cloud environment is one approach for maximizing system Quality-of-Service (QoS) and minimizing total cost. A traditional challenge to this optimization is the large amount of benchmarking required to optimize even simplistic cloud systems. This paper introduces \(\hbox {C}^2\)RAM, an new approach to enable rapid, optimized deployment of software onto a cloud environment by substantially reducing the number of benchmarks required. \(\hbox {C}^2\)RAM continues to perform some benchmarking, and therefore its predictions of application QoS metrics, such as throughput and latency, are very accurate. Our results show a maximum difference of 1.06 % between \(\hbox {C}^2\)RAM predicted QoS and empirically measured QoS. Moreover, \(\hbox {C}^2\)RAM can be provided with QoS requirements for each software in the system, and will ensure that each requirement is met before presenting a deployment plan.

Deep Learning is an emerging field in Artificial Intelligence that uses biologically inspired neural networks to recognize patterns in the natural world. These neural networks have an amazing ability to process large amounts of data and learn from them. Recurrent Neural Networks (RNN) are used in applications involving natural language processing like text translations and text generation. This research evaluates the effectiveness of a RNN to be able to automatically generate programming code. Programming languages are different from natural languages in that they have unique structure and syntax. The goal for this research is to conduct experiments on a character RNN model with for three programming languages; Java, Python and C#, and evaluate the results by testing and analyzing the ability for the RNN to automatically produce code that is able to compile.

Industrial control systems (ICS), such as smart grid systems, are frequently composed of hundreds of devices distributed over a large geographic area. While mobile applications have been used with good success in managing ICSs, traditional methods of distributing applications (e.g., app stores) are not well suited to the task of discovering, distributing, and building human machine interfaces (HMIs) for ICS, as the highly individualized and often proprietary individual components of ICSs have vastly different interfaces leading to a need to download hundreds of applications. We propose the No Effort Rapid Development (NERD) middleware framework to address the challenges of in-field HMI discovery, provisioning, communication, and co-evolution with related ICSs. Middleware services offer the ability to simplify on-demand HMI distribution and operation of ICSs. NERD leverages existing ICS device-markers (e.g., QR-codes or RFID tags) or Bluetooth low-energy protocols for rapid cyber-physical discovery and provisioning of HMIs in the field. Device-markers and Bluetooth low-energy protocols have a very limited data capacity and transmission speed, and to achieve on-device storage of HMIs, we propose using a compact data-driven domain-specific language that emphasizes data sources and sinks between the HMI and IC.

Today's modern information era society imposes great necessity of various software applications which now have an increasingly important impact on human life. The need for software applications and developers is rapidly increasing. To make up that demands enterprises have to make more applications in shorter time frames. Workload of software companies is constantly increasing, as they not only have to develop new applications but also have to maintain existing software by promptly responding to the business changes and do so in an appropriate manner. In order to respond to the growing demands, this paper presents code generator tool which automatically generate production-ready source code based on the provided template. Use of template generator provides well-structured generated code, faster and cheaper application development and maintenance. In addition to the financial benefits, code generators preserve identical code structure and way of coding in every generated file, which makes next cycle in software life easier. In order to facilitate developers' tasks code generator presents concept of customize template library. Our goal was to produce a code generator which is simple to use, customize and which acts as the skeleton to downgrade the development time and expenses.