A study in the International Journal of Computational Science and Engineering, introduces a new deep learning architecture called LightNet, designed to overcome the challenges of training deep learning models and revolutionize the agricultural sector. The study, focuses on plant disease management and fruit classification and addresses the normally high computational resource demands that have hindered the implementation of deep learning models on limited-resource devices for disease identification in images of leaves and fruit.

Deep learning, a powerful technique in artificial intelligence, has gained popularity across various applications. However, its resource-intensive nature has made it unsuitable for devices with limited computing power and storage capacity. Moreover, there is a dearth of efficient approaches for tackling real-world agricultural problems using deep learning.

Edna C. Too of the Department of Computer Science at Chuka University in Kenya has developed LightNet, a compact convolutional neural network (CNN) that uses two innovative strategies, skip connections and pruning. This increases efficiency considerably by allowing smoother information flow through the network while reducing unnecessary connections and parameters. The approach allows the system to outperform seemingly more powerful tools. For instance, it is half the size, double the efficiency, and three times faster than DenseNet.

The researchers evaluated LightNet using two real-world datasets: PlantsVillage, which focuses on plant disease detection, and Fruits-360, which involves fruit classification and grading. The results of the evaluation demonstrate just how well the system works for plant disease detection and fruit classification tasks. The potential is immense. By providing an efficient and accurate solution for deep learning in these real-world applications, LightNet offers a way for growers and suppliers to be better equipped to counter major problems across the sector. As it requires a lower-resource device there is the potential for it to be used in the field, as it were, at lower cost than other more resource-intense systems, ultimately improving crop management and food security.

Too, E.C. (2023) 'LightNet: pruned sparsed convolution neural network for image classification', Int. J. Computational Science and Engineering, Vol. 26, No. 3, pp.283–295.
DOI: 10.1504/IJCSE.2023.131508

Researchers have developed a novel artificial intelligence (AI) model that combines an algorithm based on the scouting and foraging behaviour of bee colonies with a fuzzy wavelet neural network to accurately predict road traffic accidents.

The artificial bee colony algorithm is a swarm intelligence algorithm that has been used to solve complex optimization problems in the past. Now, writing in the International Journal of Computing Science and Mathematics, Zhicheng Li of the Department of Urban Rail Transit and Information Engineering at Anhui Communications Vocational and Technical College in Hefei, China, has introduced self-adaptive mutation operations to overcome the algorithm's known limitations. The use of a fuzzy wavelet neural network reduces the time needed to solve a problem and improves improves its search skills for finding a solution.

The artificial bee colony algorithm consists of worker bees, onlooker bees, and scout bees. Worker bees explore solutions based on specific rules, while onlookers select promising solutions using information shared by the workers. The scouts introduce new random solutions to boost the diversity of possible solutions in processing the data. Through an iterative process, the algorithm converges toward an optimal or near-optimal solution to the problem, in this case the nature of road traffic accidents. The fuzzy wavelet neural network uses fuzzy logic and various statistical tools within a conventional neural network to handle uncertainty and imprecision within the data.

Li has carried out computer simulations with the system to see how well it might predict fatalities in road traffic accidents based on the various factors associated with a particular incident.

"Computer simulations show that this prediction method fully exploits the nonlinear approximation ability of the wavelet neural network model, effectively improves convergence speed and training efficiency, and reduces computational complexity," writes Li.

The work has the potential to improve our ability to anticipate and prevent lethal road traffic accidents by allowing limited resources to be more usefully assigned to proactive measures and road safety strategies. There are, in addition, implications for the arrival of driverless vehicles on our roads.

Li, Z. (2023) 'Traffic accident prediction based on an artificial bee colony algorithm and a self-adaptive fuzzy wavelet neural network', Int. J. Computing Science and Mathematics, Vol. 17, No. 3, pp.254–265.
DOI: 10.1504/IJCSM.2023.131464

In the information-overload era, authenticity is critical but elusive, while fake news, disinformation, and fraudulent reviews are common but not always easily spotted.

Research in the International Journal of Data Mining, Modelling and Management focuses on one particular aspect of this problem how to identify a fake review, specifically a fake movie review, using sentiment analysis techniques to discern meaning from a given review and determine whether it is genuine or not. The work has implications for movie buffs the world over who might then navigate the endless reviews with confidence. The results should also improve the credibility of the movie industry by helping to identify and remove such fraudulent reviews.

Isha Gupta and Neha Gupta of the Faculty of Computer Applications at the Manav Rachna International Institute of Research and Studies in Faridabad, India, and Indranath Chatterjee of the Department of Computer Engineering at Tongmyong University in Busan, South Korea, have analyzed vast amounts of text data to uncover the specific words that contribute to biases in reviews and their influence on overall viewer sentiment. The team used a "valence-aware" dictionary, one that understands the emotional tone or polarity conveyed by particular words or phrases. Valence can be of a positive, negative, or neutral nature.

The researchers were thus able to identify the influential words in a review associated with a specific genre whether the review was of a comedy, horror, action, drama, or thriller. By using a statistical method known as Pearson's correlation analysis, they could also identify influential features that distinguish each genre. This sheds light on the language used to describe different kinds of movies. Ultimately, the approach gives the team a quantitative assessment of the sentiment conveyed in a given movie review. Around one in five of the characteristic features of the reviews analysed were common across different genres, suggesting that "subtle changes in the feature set showing distinct discrimination among the words used for positive and negative reviews and also for each genre," the team writes. " there is a shallow degree of correlation present genre-wise."

The significance of this research extends beyond understanding viewer sentiments. The study's findings have important implications in the realm of identifying fake movie reviews. This approach to analyzing the language and sentiment expressed in a movie review, could allow service providers that host reviews to automatically assess the credibility and reliability of a given review and to flag or remove any from their system that is deemed to be fake or not credible in some way. Such a system would not represent censorship of genuine reviews, of course, but ensure that movie fans and industry professionals would have access to authentic information rather than fake reviews, which might otherwise influence movie choice and the consumer experience and at the bottom-line, industry profits, uptake of sequels and franchises, and overall commercial success.

Gupta, I., Chatterjee, I. and Gupta, N. (2023) 'Identification of relevant features influencing movie reviews using sentiment analysis', Int. J. Data Mining, Modelling and Management, Vol. 15, No. 2, pp.169–183.
DOI: 10.1504/IJDMMM.2023.131395

Research in the International Journal of Web Based Communities introduces a new and accurate approach to identifying abnormal users in social networks by examining several characteristics of user behaviour at once. By tapping into the APIs (Advanced Programming Interfaces) of various social networks, Jian Xie of the College of Education at Fuyang Normal University in Fuyang, China, was able to gather comprehensive data about users, including details about their accounts, the content they post, and the specific ways they behave. An analysis of this data allowed him to ascribe a set of attributes to users. By applying attribute reduction, he could then eliminate any redundant features and so build a targeted attribute feature set with which to analyse suspicious accounts.

Xie then used the assimilated data to train the XGBoost model, a powerful machine learning algorithm, to create a highly objective function that can quickly flag abnormal behaviour on a social network. Xie was able to identify abnormal users with 95 percent accuracy, sufficient to alert the system's administrators to any putative issues that could then be manually investigated and action taken to block malicious users, for instance. The error level achieved was low as was the speed with which abnormal users could be identified, within fractions of a second, in fact. Xie's approach is faster and more accurate than the previous methods he notes in his paper.

The findings have implications across social networking, where the identification of abnormal users, whether they are malicious third parties, trolls, spammers, bullies, misinformation accounts, fake accounts, hijacked usernames or bots, plays an important role in maintaining the safety of legitimate users and protecting the overall integrity of the digital community.

"This method has the characteristics of high feature extraction accuracy, low identification error rate, and low identification time of abnormal users in social networks," Xie concludes. He suggests that the approach could lay the foundations for developing powerful social network security policies.

Xie, J. (2023) 'An accurate identification method of abnormal users in social network based on multivariate characteristics', Int. J. Web Based Communities, Vol. 19, Nos. 2/3, pp.80–92.
DOI: 10.1504/IJWBC.2023.131386

Research in the International Journal of Global Warming has looked at the study, the carbon footprint of t-shirts made from different materials. The textile and clothing industry is vast and so has a significant impact on climate change through gathering resources, processing and manufacturing of products, and supply to the market. The team undertaking the work is based at Zhejiang Sci-Tech University in Hangzhou, Zhejiang, China, and examines the entire lifecycle of these popular garments whether made from the wholly natural fibre, cotton, the semi-synthetic material viscose, or entirely synthetic polyester. They consider the various stages from raw material extraction to end-of-life disposal.

Junran Liu, Yiqi Guo, Ying Zhang, and Laili Wang, Zhejiang Sci-Tech University worked with Lirong Sun of the Office for Social Responsibility of China National Textile and Apparel Council in Beijing, and Wei Bao of the College of Textile and Clothing at Qingdao University, in Shandong. The team unravelled the various costs in terms of resources and energy and found that the manufacturing of yarn of any time, used in fabric production, plays a significant role in the carbon footprint of a t-shirt. Yarn manufacture accounts for up to a half of the total emissions associated with these clothing products. Fabric manufacturing phase contributes to about 20 percent of the carbon footprint. The t-shirt's usage phase, which encompasses washing and drying, accounts for between 31 and 48 percent of the carbon footprint. In other words, production energy is the primary driver of carbon emissions when it comes to t-shirts.

The researchers point out that there is a positive impact of using plant-derived fibres, such as cotton and viscose, in offsetting greenhouse gas emissions through carbon sequestration by the source plants as they grow. Of course, at end-of-life, the materials must somehow find a secondary use in recycling or be landfilled rather than burnt, otherwise that stored carbon is released into the atmosphere once more.

The work has implications for manufacturers and consumers. Companies aiming to decrease the carbon footprint of their t-shirts should focus on making their production technologies more sustainable by increasing their use of renewable energy sources, for instance. Similarly, consumers should aim to use renewable energy to do their laundry as well as using detergents that allow them to wash their clothes effectively at low temperature. Moreover, care and repair might usefully extend the life of a t-shirt despite its superficial nature when compared to more formal or fashion clothing.

This kind of study underscores the urgency with which we need to work together to have the clothes we need but also to address the environmental impact of the fashion industry.

Liu, J., Sun, L., Guo, Y., Bao, W., Zhang, Y. and Wang, L. (2023) 'Carbon footprint of t-shirts made of cotton, polyester or viscose', Int. J. Global Warming, Vol. 30, No. 3, pp.271–281.
DOI: 10.1504/IJGW.2023.131404

Research in the International Journal of Services and Operations Management has investigated what influences the use of online food delivery services in India and the impact that is having on public health. The online food delivery sector in India has annual revenues of well over $7 billion. A significant proportion of this sector, about 85 percent, is restaurant-to-consumer services and the remainder, is services by direct providers.

The suspicion is that increasing reliance on food other than home-cooked food could be affecting lifestyle and health by side-stepping the traditional activities involved in preparing and eating a meal with family, friends, or even alone. Additionally, there are issues surrounding the quality of food delivered to one's door in terms of calorific content, how much of the food is considerably processed, and the presence of additives. Conversely, some food from suppliers may well be more nutritious and the time freed up from the drudge of cooking and clearing up may well allow people to be involved in other quality activities with those family and friends.

Arghya Rayof the International Management Institute Kolkata, Pradip Kumar Bala of the Indian Institute of Management Ranchi, India, and Rashmi Jain of the Feliciano School of Business in Montclair, New Jersey, USA, explain that the advent of online food delivery services has had a significant effect on the food industry and our lifestyles in recent years. The team used a mixed-method approach, starting with preliminary qualitative interviews of online food delivery service users and followed by a quantitative survey of almost 300 Generation Y users in India. Generation Y is commonly defined as people born between around 1981 to 1996.

The team found that emotional values, conditional values, monetary values, and health consciousness all had a significant impact on consumers' intention to use online food delivery services. Moreover, continued use leads to ongoing changes in food consumption patterns. The researchers point out that their work may not necessarily extrapolate to other countries, especially those with very different food standards.

The team suggests that suppliers should improve standards given the ever-increasing use of online food delivery services if there is not to be a long-term and potentially detrimental impact on consumers. The findings also point to differences based on gender, age, and location, such data might allow those same providers to better tailor what they offer consumers to particular demographics.

Ray, A., Bala, P.K. and Jain, R. (2023) 'Assessing the factors influencing continued use of online-food-delivery services and the impact on health: a multi-group analysis', Int. J. Services and Operations Management, Vol. 45, No. 1, pp.1–36.
DOI: 10.1504/IJSOM.2023.131281

Research in the International Journal of Wireless and Mobile Computing addresses a significant safety issue facing power supply companies – the presence of bird nests on power line towers and other infrastructure.

Haopeng Yang and Enrang Zheng of the School of Electrical and Control Engineering at Shaanxi University of Science and Technology, and Yichen Wang and Junge Shen of the Unmanned System Research Institute at the Northwestern Polytechnical University all in Xi'an, Shaanxi, China, have developed a real-time detection system capable of swiftly identifying bird nests on transmission towers.

At first glance, one might imagine that the presence of a nest on a power transmission tower, or pylon, would be harmless, but there are serious issues with damage and the potential for avian activity to "trip" safety cutouts on power systems, leading to outages for consumers. This is particularly true of substantial nests built high up on pylons by raptors, storks, and other large species.

Unfortunately, the detection of such small objects as bird nests and the passing of the information back to a control centre have represented an ongoing challenge due to their small size and the potential data loss during detection. The team's new system, uses an algorithm that can identify and so detect bird nests at different scales allowing for rapid risk identification. An unmanned aerial vehicle (UAV), often referred to as a drone, fitted with a camera can patrol the towers, record and analyse images using an onboard computer running the team's algorithm and report back to the controllers with information that flags specific towers with a nest problem. The team's algorithm readily overcomes the problem of the background scenery in an image of a pylon being checked detecting only the presence of nests.

The research team says they have achieved an average accuracy rating of 90.05%. This high level of performance meets the demands of the State Grid for high-precision and real-time line maintenance inspections. The automated detection system precludes the need for costly regular manual inspections.

Yang, H., Zheng, E., Wang, Y. and Shen, J. (2023) 'Real-time detection system of bird nests on power transmission lines based on lightweight network', Int. J. Wireless and Mobile Computing, Vol. 24, Nos. 3/4, pp.217–225.
DOI: 10.1504/IJWMC.2023.131295IJ

The Indian retail industry contributes more than 10 percent to the country's gross domestic product (GDP). A new study looks at how impulsive and non-rational consumer behaviour in the Indian clothing sector potentially leading to personal debt can have a detrimental effect on the economy as a whole. Moreover, the work points to how promoting more rational purchasing decisions could be better for consumers as well as leading to a more sustainable and responsible industry.

Of course, marketers work to promote products and they will use emotional means to do so that will often trigger a non-rational response from a would-be customer. Conversely, who, in a free society, is to tell a customer what they do and don't need when it comes to clothes shopping. After all, people buy clothes for obvious practical reasons, but also for self-expression, for enjoyment, and many other non-practical reasons.

Komal Malik and Manoj Joshi of the Amity Business School at Amity University Uttar Pradesh, Lucknow Campus, used an experiential research design to survey and capture non-rational behaviour shopping behaviour among Indian consumers. "Non-rationality can be referred as the influence of emotional factors rather than tangible gains and losses associated with a choice," the authors write. In addition, in their paper in the International Journal of Business and Globalisation, they reviewed the existing literature to provide context. The team's analysis considered factors such as brand loyalty, gift and special occasion purchases, social affinity, lifestyle choice, the feel-good factor, offers and discounts, changing fashion, personality.

The researchers found that consumer behaviour was driven by rational as well as non-rational factors, but it was the latter, associated with impulse purchases that was often associated with consumers spending on credit. If such consumers do not have the funds or disposable income to back their purchases then repeated impulse buys, has the potential to lead to greater debt. Understanding and addressing this behaviour is crucial for both marketers and policymakers, the research suggests.

Malik, K. and Joshi, M. (2023) 'I saw it, I bought it! The irrational buying behaviour in retail sector', Int. J. Business and Globalisation, Vol. 34, No. 1, pp.17–27.
DOI: 10.1504/IJBG.2023.131268

Currently, there is no established method for automatically assessing the level of FAIRness (Findability, Accessibility, Interoperability, and Reusability) of semantic resources. The term "semantic resources" refers to various types of data, information, or knowledge artefacts that are represented in a structured and standardized way. These resources can include ontologies (technical, structured glossaries), vocabularies, data sets, and other relevant knowledge. An example is the AgroPortal semantic resource repository, an online platform for storing and organizing semantic resources related to the domain of agri-food and environment.

Writing in the International Journal of Metadata, Semantics and Ontologies, a team from France has used the AgroPortal as a case study to help them develop a metadata-based automatic assessment methodology for such resources, which they call Ontology FAIRness Evaluator (O'FAIRe).

Emna Amdouni, Syphax Bouazzouni, and Clement Jonquet of the University of Montpellier explain that making digital scientific data openly available remains an important challenge for the scientific community and funding agencies. The FAIR movement arose in 2014 to help address this challenge and has been largely embraced. However, FAIR, as many observers have pointed out, is only representing specifications for digital objects, or entities, rather than being a standardised or technically based system. There has thus been a need for a way to independently assess how well an entity adheres to the principles of FAIR.

In this context, the team's proposal is aligned with existing initiatives and consists of 61 questions, primarily based on metadata descriptions, and using ontology libraries or repositories to ensure unified metadata for FAIRness assessment. The team implemented O'FAIRe in AgroPortal and successfully conducted a preliminary FAIRness analysis of 149 semantic resources in the agri-food/environment domain. The proposal should allow FAIR digital entities to be assessed objectively pushing us towards a more encompassing system in which entities and resources can be read and used competently equally well by humans and computers without barriers and problems arising because of inconsistencies across and within domains.

The researchers conclude that their work addresses many of the scientific and technical challenges regarding the implementation of the 15 FAIR principles for ontologies and semantic resources. The team writes that their work might now "guide the semantic community to put the FAIR principles into practice and enable them to qualify the degree of FAIRness of their semantic resource."

Amdouni, E., Bouazzouni, S. and Jonquet, C. (2022) 'O'FAIRe makes you an offer: metadata-based automatic FAIRness assessment for ontologies and semantic resources', Int. J. Metadata Semantics and Ontologies, Vol. 16, No. 1, pp.16–46.
DOI: 10.1504/IJMSO.2022.131133

In the ever-changing digital landscape, bloggers have risen to prominence as influencers, playing an important role in helping consumers pick and choose the products and services on which they want to spend their time and money. However, even with the many disparate social media apps that distract consumers from "traditional" blogs, there remains a huge number who have influence across many different spheres and represent a useful resource for marketers.

The problem remains how to identify and classify the many, many bloggers for best impact in a marketing campaign. A research study in the International Journal of Internet Marketing and Advertising shows how a comprehensive framework can classify consumer bloggers based on their unique content creation approach.

Beatrice Ietto and Federica Pascucci of the Università Politecnica delle Marche in Ancona, Italy, have drawn on social-practice theory to construct their classification framework. In this theory, content creation is viewed as habitual behavior shaped by socio-cultural contexts. The team has focused on an extensive netnographic analysis of Australian music bloggers to offer new insights into the critical factors that influence a blogger's content creation approach.

The work shows that blogger in this niche create content primarily driven by their subjective evaluation of four key dimensions: personal influences, audience influences, community influences, and commercial influences. These dimensions play a pivotal role in shaping the blogger's content creation strategies and determining the nature of their engagement with their readership.

With the details of these insights to hand, the team created a multidimensional framework for the classification of bloggers as "passionate", "hype followers", "sophisticated and sub-cultural", "celebratory and overly positive, and the "professionals". The framework could offer marketing practitioners a useful resource for identifying and collaborating with the most appropriate bloggers that mesh well with their promotional strategies. The framework goes beyond the simplistic metrics of site "hits" and "reach" and looks at how the blog functions and how that would seamlessly work with a marketing campaign.

Ietto, B. and Pascucci, F. (2023) 'Classifying bloggers based on content creation approaches: implications for influencers marketing strategies', Int. J. Internet Marketing and Advertising, Vol. 18, No. 4, pp.335–358.
DOI: 10.1504/IJIMA.2023.131255