Search results

Effectively predicting research and development project portfolio benefit (R&D PPB) could assist organizations in monitoring the execution of research and development project portfolio (R&D PP). However, due to the uncertainty and complexity of R&D PPB, current research remains lacking a valid R&D PPB prediction tool. Therefore, an R&D PPB prediction model is proposed via a backpropagation neural network (BPNN).

The R&D PPB prediction model is constructed via a refined immune genetic algorithm coupling backpropagation neural network (RIGA-BPNN). Firstly, considering the characteristics of R&D PP, benefit evaluation criteria are identified. Secondly, the benefit criteria values are derived as input variables to the model via trapezoidal fuzzy numbers, and then the R&D PPB value is determined as the output variable through the CRITIC method. Thirdly, a refined immune genetic algorithm (RIGA) is designed to optimize BPNN by enhancing polyfitness, crossover and mutation probabilities. Lastly, the R&D PPB prediction model is constructed via the RIGA-BPNN, followed by training and testing.

The accuracy of the R&D PPB prediction model stands at 99.26%. In addition, the comparative experiment results indicate that the proposed model surpasses BPNN and the immune genetic algorithm coupling backpropagation neural network (IGA-BPNN) in both convergence speed and accuracy, showcasing superior performance in R&D PPB prediction. This study enriches the R&D PPB predicting methodology by providing managers with an effective benefits management tool.

The research implications of this study encompass three aspects. First, this study provides a profound insight into R&D PPB prediction and enriches the research in PP fields. Secondly, during the construction of the R&D PPB prediction model, the utilization of the composite system synergy model for quantifying synergy contributes to a comprehensive understanding of intricate interactions among benefits. Lastly, in this research, a RIGA is proposed for optimizing the BPNN to efficiently predict R&D PPB.

This study carries threefold implications for the practice of R&D PPM. To begin with, the approach proposed serves as an effective tool for managers to predict R&D PPB. Then, the model excels in efficiency and flexibility. Furthermore, the proposed model could be used to tackle additional challenges in R&D PPM, such as gauging the potential risk level of R&D PP.

Effective predicting of R&D PPB enables organizations to allocate their limited resources more strategically, ensuring optimal use of capital, manpower and time. By accurately predicting benefit, an organization can prioritize high-potential initiatives, thereby improving innovation efficiency and reducing the risk of failed investments. This approach not only strengthens market competitiveness but also positions organizations to adapt more effectively to changing market conditions, fostering long-term growth and sustainability in a competitive business environment.

Incorporating the characteristics of R&D PP and quantifying the synergy between benefits, this study facilitates a more insightful R&D PPB prediction. Additionally, improvements to the polyfitness, crossover and mutation probabilities of IGA are made, and the aforementioned RIGA is applied to optimize the BPNN. It significantly enhances the prediction accuracy and convergence speed of the neural network, improving the effectiveness of the R&D PPB prediction model.

This study/paper aims to understand the public perceptions of AI through mass media discourse. In the past few years, significant progress has been made in the field of artificial intelligence (AI). The benefits of AI are obvious, but there is still huge uncertainty and controversy over the public perception of AI. How does the mass media conceptualize AI?

In this paper, the authors analyze the evolution of AI covered by five major news media outlets in the past 30 years from 7 dimensions: scientific subject, keyword, country, institution, people, topic and opinion polarity.

First of all, different subjects are competing for and dividing up the right to speak of AI, leading to the gradual fragmentation of the concept of AI. Second, reporting on AI often includes reference to commercial institutions and scientists, showing a successful integration of science and business. Moreover, the result of topic modeling shows that news media mainly defines AI from three perspectives: an imagination, a commercial product and a field of scientific research. Finally, negative reports have focused on various issues relating to AI ethics.

The results can help bridge various conversations surrounding AI and promote richer discussions, increase the participation of scientists, businesses, governments and the public and provide more perspectives on the functions, prospects and pitfalls of AI.

The salinity of the oilfield produced water has a significant effect on steel corrosion. The purpose of this paper is to study the influence of salinity on corrosion behavior of X60 steel and it also provides basic for material selection of gas wells with high salinity.

The weight loss experiment was carried out on steel with high temperature and high pressure autoclave. The surface morphology and composition of corrosion scales were studied by means of scanning electron microscopy, energy dispersive spectroscopy and X-ray diffractometry.

The results show that as salinity increases, the corrosion rate of X60 steel will gradually experience a rapid decline stage and then a slow decline stage. X60 steel is mainly exhibiting uniform corrosion in the first rapid decline stage and pitting corrosion in the second slow decline stage. The increase in salinity reduces gas solubility, which, in turn, changes the morphology and density of the corrosion scales of X60 steel. At low salinity, loose iron oxides generated on the surface of the steel, which poorly protects the substrate. At high salinity, surface of the steel gradually forms protective films. Chloride ions in the saline solution mainly affect the structure of the corrosion scales and initiate pitting corrosion. The increased chloride ions lead to more pitting pits on the surface of steel. The recrystallization of FeCO₃ in pitting pits causes the corrosion scales to bulge.

The investigation determined the critical concentration of pitting corrosion and uniform corrosion of X60 steel, and the new corrosion mechanism model was presented.

Social media platforms (SMPs) are pivotal in information dissemination and molding public opinion. Various platforms exhibit differences and characteristics. It is necessary to compare and analyze the roles played by different platforms in the evolution of public events.

This study develops a framework to evaluate the role of SMPs at different stages of public events. To calculate some of these indicators, the GPT-AP-TextRank topic model is constructed. The study further analyzes the correlation between indicators at different stages and SMP’s role and compares SMP’s different roles among the four stages.

The results reveal significant disparities in the role of different SMPs during public events. Weibo demonstrates notable performance during the outbreak, spread and recession stages of the event, exhibiting a strong influence on public event evolution. Bilibili, Douban, Zhihu and Baidu Tieba show relatively ordinary roles. In addition, compared to the spread stage, SMPs exhibit a stronger ability to influence event redirection in the initial stage, which is different from the original assumption of the study.

The findings expose the powerful roles of SMPs in event evolution, providing valuable insights for enhancing public event governance.

This study proposes an evaluation method for SMPs’ role and introduces a novel GPT-AP-TextRank topic generation model for the indicator calculation.

The type 120 emergency valve is an essential braking component of railway freight trains, but corresponding diaphragms consisting of natural rubber (NR) and chloroprene rubber (CR) exhibit insufficient aging resistance and low-temperature resistance, respectively. In order to develop type 120 emergency valve rubber diaphragms with long-life and high-performance, low-temperatureresistant CR and NR were processed.

The physical properties of the low-temperature-resistant CR and NR were tested by low-temperature stretching, dynamic mechanical analysis, differential scanning calorimetry and thermogravimetric analysis. Single-valve and single-vehicle tests of type 120 emergency valves were carried out for emergency diaphragms consisting of NR and CR.

The low-temperature-resistant CR and NR exhibited excellent physical properties. The elasticity and low-temperature resistance of NR were superior to those of CR, whereas the mechanical properties of the two rubbers were similar in the temperature range of 0 °C–150 °C. The NR and CR emergency diaphragms met the requirements of the single-valve test. In the low-temperature single-vehicle test, only the low-temperature sensitivity test of the NR emergency diaphragm met the requirements.

The innovation of this study is that it provides valuable data and experience for future development of type 120 valve rubber diaphragms.

This paper reviews the pros and cons of different parametric modeling methods, which can provide a theoretical reference for parametric reconstruction of 3D human body models for virtual fitting.

In this study, we briefly analyze the mainstream datasets of models of the human body used in the area to provide a foundation for parametric methods of such reconstruction. We then analyze and compare parametric methods of reconstruction based on their use of the following forms of input data: point cloud data, image contours, sizes of features and points representing the joints. Finally, we summarize the advantages and problems of each method as well as the current challenges to the use of parametric modeling in virtual fitting and the opportunities provided by it.

Considering the aspects of integrity and accurate of representations of the shape and posture of the body, and the efficiency of the calculation of the requisite parameters, the reconstruction method of human body by integrating orthogonal image contour morphological features, multifeature size constraints and joint point positioning can better represent human body shape, posture and personalized feature size and has higher research value.

This article obtains a research thinking for reconstructing a 3D model for virtual fitting that is based on three kinds of data, which is helpful for establishing personalized and high-precision human body models.

Pipeline robots are often used in pipeline non-destructive testing. Given the need for long-range in-pipe inspections, this study aims to develop a wireless in-pipe inspection robot for image acquisition.

In this paper, an in-pipe robot with a new mechanical system is proposed. This system combines a three-arm load-bearing structure with spring sleeves and a half-umbrella diametric change structure, which can ensure the stability of the camera when acquiring images while maintaining the robot’s flexibility. In addition, data were transmitted wirelessly via a system that uses a 433 MHz ultra-high frequency and wireless local-area network–based image transmission system. Software and practical tests were conducted to verify the robot’s design. A preliminary examination of the robot’s cruising range was also conducted.

The feasibility of the robot was demonstrated using CATIA V5 and MSC ADAMS software. The simulation results showed that the centre of mass of the robot remained in a stable position and that it could function in a simulated pipeline network. In the practical test, the prototype functioned stably, correctly executed remote instructions and transmitted in near real-time its location, battery voltage and the captured images. Additionally, the tests demonstrated that the robot could successfully pass through the bends in a 200-mm-wide pipe at any angle between 0° and 90°. In actual wireless network conditions, the electrical system functioned for 44.7 consecutive minutes.

A wheeled wireless robot adopts a new mechanical system. For inspections of plastic pipelines, the robot can adapt to pipes with diameters of 150–210 mm and has the potential for practical applications.

Drawing on the philosophy of retail service quality, this study aims to empirically investigate how salespeople’s personal interaction affects customers’ positive word-of-mouth (WOM) intention through the lens of affective customer experience and consumers’ cross-cultural attitudes toward domestic or foreign products.

The hypothesis was tested with a two-step structural equation model using survey data obtained from 529 shopping center customers in China.

The positive impact of salespeople’s personal interaction on customers’ positive WOM intention is fully mediated by affective customer experience. In addition, consumer ethnocentrism strengthens the positive impact of salespeople’s personal interaction on affective customer experience, whereas this study fails to find the moderating effect of foreign product affinity.

To increase customers’ positive WOM intention, store managers need to encourage their frontline sales personnel to personally interact with customers to support customers’ problem-solving. Also, frontline salespeople should pay more attention to consumers’ cross-cultural attitudes such as consumer ethnocentrism when interacting with their customers.

By linking affective customer experience and consumers’ cross-cultural attitudes of ethnocentrism and affinity toward domestic or foreign products, this study further extends the knowledge of retail service quality on the relationship between salespeople’s personal interaction and customers’ positive WOM intention.

When recommending products to consumers, it is important to be able to accurately predict how consumers will rate them. However, existing collaborative filtering models are difficult to achieve a balance between rating prediction accuracy and complexity. Therefore, the purpose of this paper is to propose an accurate and effective model to predict users’ ratings of products for the accurate recommendation of products to users.

First, we introduce an attention mechanism that dynamically assigns weights to user preferences, highlighting key interaction information and enhancing the model’s understanding of user behavior. Second, a fold embedding strategy is employed to segment user interaction data, increasing the information density of each subset while reducing the complexity of the attention mechanism. Finally, a masking strategy is integrated to mitigate overfitting by concealing portions of user-item interactions, thereby improving the model’s generalization ability.

The experimental results demonstrate that the proposed model significantly minimizes prediction error across five real-world datasets. On average, the evaluation metrics root mean square error (RMSE) and mean absolute error (MAE) are reduced by 9.11 and 13.3%, respectively. Additionally, the Friedman test results confirm that these improvements are statistically significant. Consequently, the proposed model more accurately captures the intrinsic correlation between users and products, leading to a substantial reduction in prediction error.

We propose a novel collaborative filtering model to learn the user-item interaction matrix effectively. Additionally, we introduce a fold embedding strategy to reduce the computational resource consumption of the attention mechanism. Finally, we implement a masking strategy to encourage the model to focus on key features and patterns, thereby mitigating overfitting.

This paper aims to find the best method to predict the friction coefficient of textured 45# steel by comparing different machine learning algorithms and analytical calculations.

Five machine learning algorithms, including K-nearest neighbor, random forest, support vector machine (SVM), gradient boosting decision tree (GBDT) and artificial neural network (ANN), are applied to predict friction coefficient of textured 45# steel surface under oil lubrication. The superiority of machine learning is verified by comparing it with analytical calculations and experimental results.

The results show that machine learning methods can accurately predict friction coefficient between interfaces compared to analytical calculations, in which SVM, GBDT and ANN methods show close prediction performance. When texture and working parameters both change, sliding speed plays the most important role, indicating that working parameters have more significant influence on friction coefficient than texture parameters.

This study can reduce the experimental cost and time of textured 45# steel, and provide a reference for the widespread application of machine learning in the friction field in the future.