Behavior Data Research Articles

Exploring the effects of dietary lysine and tryptophan on the social behaviour of pigs.

Negative social behaviours between pigs can cause stress, which can compromise welfare. There has been significant interest in exploring the effect of diet on negative social behaviours and the wider social behaviour repertoire of pigs. The aim of this study was to determine the effects of dietary Lysine (Lys) and Tryptophan (Trp) levels on the social behaviour of commercially housed pigs. A total of 2,293 PIC Camborough barrows with a mean starting weight of 11.87 ± 1.35kg were used in a randomised complete block design with a 2×3 factorial arrangement, with 16 replicates per treatment, of the following factors: 1) Standardised ileal digestible Lys levels: 100% Lys = diets with 100% PIC requirement at the midpoint of the growth phase (Lys 100) vs 80% Lys = diets with 80% PIC requirement at the midpoint of the growth phase (Lys 80); and 2) Trp to Lys ratio of 0.210, 0.185 or 0.160. Pigs were randomly allocated across the 6 treatments over 2 starting dates. Behaviour and lesion data were collected. There was an effect of Lys (P=0.032) on ear biting behaviour, with pigs on the Lys 80 treatments showing a higher level of ear biting behaviour. We also found an effect of the Lys score week interaction on the proportion of pens showing ear lesions (P<0.001) and an effect of the Lys Trp interaction (P=0.030) and the Lys score week interaction (P=0.0104) on the proportion of pens showing severe ear lesions. In conclusion, the lysine content of feed can affect the social behaviour of pigs, specifically ear biting, in commercial conditions.

Analysis of Student Learning Behavior on Online Education Platforms Based on Deep Learning

Artificial intelligence (AI) and deep learning (DL) techniques are increasingly used in education because of advancements in online learning platforms and their ongoing implementation. The existing methods suffer from low-processing efficiency, high prediction error, and increased memory requirements when faced with vast learning and student behavior data. Thus, based on DL, this research suggests a way to analyze student behavior in e-learning. Data on student behavior are gathered, and a learning behavior model for online learning is created. The proposed optimal DL approach aims to screen the collected behavior data using data preparation, analysis, and statistics. Additionally, the Pearson correlation coefficient (PCC) approach is employed to determine the degree of data similarity. The novelty of the research is followed by utilizing an optimized DL network, known as a deep neural network with red deer optimization (ODNN-RDO), to mine students’ behavior data in e-learning programs. Two datasets, metrics including accuracy, precision, and recall, together with error measures like relative error, the root mean square error (RMSE), and absolute error, are utilized to test the created models. The improved generated models achieved 98.15% accuracy and 0–0.04% error compared to the current methods. The optimization procedure subsequently optimizes the components to acquire the best outcomes regarding faculty and parent performance monitoring of students. With effective monitoring, this model maximizes the e-learning platform for planning student growth.

A Comprehensive Review on Machine Learning Approaches for Smartphone Addiction Prediction

Smartphone addiction has become a pressing public health concern, with far-reaching implications for mental health, social interactions, and overall well-being. This study proposes an unsupervised machine learning framework to predict smartphone addiction by analyzing user behavior data, such as screen time, application usage patterns, and psychological factors. The research leverages unsupervised learning techniques to identify patterns and commonalities in unlabelled data, aiming to classify individuals at risk of addiction without predefined categories. Using advanced machine learning libraries like TensorFlow and OpenCV, the system incorporates a robust pipeline involving data collection, preprocessing, and training/testing with high-performance models. The findings highlight excessive social media use and notification frequency as significant predictors, with applications ranging from personalized intervention programs and mental health monitoring to enhanced parental control systems. With a focus on scalable and adaptive methodologies, this framework not only addresses the challenges of smartphone addiction but also offers valuable insights for researchers, policymakers, and app developers to design effective solutions for healthier technology use in an increasingly digital society Key Words: Smartphone addiction, machine learning, user behavior analysis, mental health, unsupervised learning.

Accelerometer-based sedentary time and physical activity with incident and progressive tobacco smoking in 2503 children: A 13-year mediation and temporal longitudinal study.

Long-term accelerometer-based studies examining whether engaging in physical activity (PA) reduces the risk of smoking in children are lacking, and whether directly measured adiposity increases smoking risk is uncertain. From the Avon Longitudinal Study of Parents and Children (ALSPAC), UK birth cohort, 2503 children aged 11 years with complete smoking and at least one valid time-point movement behaviour data were followed up until age 24 years. ActiGraph assessed sedentary time (ST), light PA (LPA), and moderate-to-vigorous PA (MVPA) at ages 11, 15, and 24-year clinic visits and smoking-based questionnaires were administered, whilst fat mass was measured with dual-energy x-ray absorptiometry scanner. Data were analysed with generalized linear-mixed effect models with logit-links and structural equation models. Mean [SD] age at baseline was 11.71 [0.20] years and the prevalence of smoking at ages 13, 15, and 24 years were 1.5%, 13.5%, and 26.6%, respectively. Higher MVPA at age 11 years was significantly associated with lower odds of smoking at age 13 years (odds ratio, 0.991; [95% CI, 0.983-0.999], P=0.035). Cumulative MVPA during ages 11-24 years was significantly associated with lower odds of progressive smoking during growth from ages 13-24 years (0.992; [95% CI, 0.989-0.996], P<0.001). Cumulative ST and LPA had no significant associations with smoking. Higher fat mass suppressed (10% suppression) the inverse relationship between MVPA and smoking. In path analysis, higher MVPA at age 15 years temporally preceded reduced smoking at age 24 years. In the present study, engaging in MVPA from age 11 years may prevent 9 in 1000 children from starting smoking at the age of 13 years when smoking prevalence was 15 in 1000 children, i.e potentially preventing 60% of children from initiating smoking. However, the MVPA preventive approach diminished as children grew into young adults.

Neurobiological Mechanisms Link Bipolar Disorder to Cardiovascular Disease: A Retrospective Biobank Study of Adverse Event Risk and Contributory Mechanisms.

Individuals with bipolar disorder are at greater risk of developing cardiovascular disease. However, the mechanisms underlying this association remain poorly understood. This study aimed to (1) determine the risk of major adverse cardiovascular events (MACE) after adjusting for important confounders and (2) evaluate the neural, autonomic, and immune mechanisms underlying the link between bipolar disorder and cardiovascular disease. Leveraging the Mass General Brigham Biobank, bipolar disorder and incident MACE were identified using the International Classification of Disease (ICD) codes. Incident MACE events were assessed from enrollment to the date of data lock (December 2020); or to the 10-year period. Health behavior data were derived from optional surveys. Cox regression hazard models were applied. Of 118,827 Biobank participants, 6009 were diagnosed with bipolar disorder. Those with bipolar disorder (vs. without) demonstrated a higher risk of MACE after adjusting for cardiovascular risk factors (hazard ratio [95% confidence interval] = 1.29 [1.10-1.51], p = 0.002). The relationship remained significant over 10 years after adjustment for unhealthy lifestyle behaviors (1.29 [1.03, 1.61], p = 0.025). Furthermore, SNA, autonomic nervous system, and inflammatory markers each significantly associated with both bipolar disorder and MACE risk. Each of these measures mediated the association between bipolar disorder and MACE (accounting for 3.8%-17.8% of the relationship). This study demonstrates that bipolar disorder associates with heightened cardiovascular risk, even after accounting for cardiovascular risk. Moreover, the findings suggest that neurobiological pathways and perturbations in autonomic and inflammatory pathways may confer cardiovascular risk in bipolar disorder.

A ATUAÇÃO DO ESTADO NA PROTEÇÃO À DIGNIDADE DA PESSOA HUMANA NA VELHICE

The inversion of the age pyramid is one of the main problems to be solved by postmodern States, whether in terms of social security, wealth production or other social issues. Changes in the dynamics of consumption, production and communication, accelerated especially by access to the virtual world and the emergence of the most recent possibilities for using Artificial Intelligence, directly impact the way in which people interact and end up making the elderly population potential victims of acts by scammers and profiteers, considering that there is a lack of knowledge about the mechanisms of how virtual environments work. Since the protection of old age and human dignity are responsibilities of the State, the vulnerability of this segment of the population becomes a State problem, at the same time that actions to protect these people virtually may come up against issues related to individual freedom and choice, and the exercise of citizenship. Protecting the elderly online must encompass both technological education, together with the supervision and regulation of financial institutions, which should take greater care with their customers in this age group, and especially for large technology companies, which host websites and directly market social interaction platforms that are supposedly free, but which collect demographic and digital, political and consumer behavior data, because just as a car manufacturer is held responsible for any accident caused by a design error, large technology companies must be held objectively responsible for the vulnerability in the provision of their services.

A traffic order evaluation method integrating aggressive driving behavior and traffic flow data based on tensor factorization

Here’s a revised version of abstract, under 150 words: This study introduces the Traffic Disorder Index (TDI), a new metric for assessing traffic operational risk by combining aggressive driving behaviours and traffic flow data. The TDI quantifies traffic order and identifies weakly stable traffic conditions using Bayesian temporal tensor factorization and dimensionality reduction to simplify high-dimensional traffic data into a low-dimensional score. The TDI’s effectiveness is evaluated through its ability to detect traffic order levels, match abnormalities, and provide interpretability. The results show that 89.6% of identified abnormal events correlating to sections with medium or high disorder. Notably, areas of high traffic disorder tend to occur at points of interest, entrances, and exits, and analyses show that elevated TDI scores are associated with greater speed variance and fluctuations, particularly near intersections in urban centres. Compared to traditional methods, the TDI better captures spatiotemporal traffic characteristics and offers actionable insights for proactive traffic safety measures.

User Behavior Simulation with Large Language Model-based Agents

Simulating high quality user behavior data has always been a fundamental yet challenging problem in human-centered applications such as recommendation systems, social networks, among many others. The major difficulty of user behavior simulation originates from the intricate mechanism of human cognitive and decision processes. Recently, substantial evidence has suggested that by learning huge amounts of web knowledge, large language models (LLMs) can achieve human-like intelligence and generalization capabilities. Inspired by such capabilities, in this article, we take an initial step to study the potential of using LLMs for user behavior simulation in the recommendation domain. To make LLMs act like humans, we design profile, memory and action modules to equip them, building LLM-based agents to simulate real users. To enable interactions between different agents and observe their behavior patterns, we design a sandbox environment, where each agent can interact with the recommendation system, and different agents can converse with their friends via one-to-one chatting or one-to-many social broadcasting. In the experiments, we first demonstrate the believability of the agent-generated behaviors based on both subjective and objective evaluations. Then, to show the potential applications of our method, we simulate and study two social phenomena including (1) information cocoons and (2) user conformity behaviors. We find that controlling the personalization degree of recommendation algorithms and improving the heterogeneity of user social relations can be two effective strategies for alleviating the problem of information cocoon, and the conformity behaviors can be highly influenced by the amount of user social relations. To advance this direction, we have released our project at https://github.com/RUC-GSAI/YuLan-Rec .

Enhancing Student Academic Success Prediction Through Ensemble Learning and Image-Based Behavioral Data Transformation

Predicting student academic success is a significant task in the field of educational data analysis, offering insights for personalized learning interventions. However, the existing research faces challenges such as imbalanced datasets, inefficient feature transformation methods, and limited exploration data integration. This research introduces an innovative method for predicting student performance by transforming one-dimensional student online learning behavior data into two-dimensional images using four distinct text-to-image encoding methods: Pixel Representation (PR), Sine Wave Transformation (SWT), Recurrence Plot (RP), and Gramian Angular Field (GAF). We evaluated the transformed images using CNN and FCN individually as well as an ensemble network, EnCF. Additionally, traditional machine learning methods, such as Random Forest, Naive Bayes, AdaBoost, Decision Tree, SVM, Logistic Regression, Extra Trees, K-Nearest Neighbors, Gradient Boosting, and Stochastic Gradient Descent, were employed on the raw, untransformed data with the SMOTE method for comparison. The experimental results demonstrated that the Recurrence Plot (RP) method outperformed other transformation techniques when using CNN and achieved the highest classification accuracy of 0.9528 under the EnCF ensemble framework. Furthermore, the deep learning approaches consistently achieved better results than traditional machine learning, underscoring the advantages of image-based data transformation combined with advanced ensemble learning approaches.

A Deep Learning Recognition Method for Students’ Abnormal Behaviors in Smart Classroom Teaching Scenarios

In the wisdom classroom, the acquisition and processing of student behavior data is a complicated process. However, in the process of data acquisition, the equipment may be affected by noise interference factors such as illumination and occlusion, which leads to the deterioration of data quality. Therefore, a deep learning recognition method for students’ abnormal behaviors in the intelligent classroom teaching scene is proposed. By modifying the selection range of random points, five branches are constructed to extract local features, obscured features and overall features, and HOG features are used to describe the shape edges of local targets to effectively detect shape features. The improved YOLOv3 uses Darknet-47 as the feature extraction network to calculate the outliers. Based on the principle of maximum dependence and threshold value, the identification model of students’ abnormal behavior is designed. The experimental results show that the average false report rate of the design method is 5.65%, the crossover ratio is higher than 90%, the frame rate is maintained between 297.7 and 293.2 in all codes, and the recognition delay time is lower than 200[Formula: see text]ms.

Improving Combustion Analysis of Extruded Polystyrene via Custom Isolation Methodology

This study is dedicated to an in−depth analysis of the combustion characteristics of extruded polystyrene (XPS) as a building insulation material with the aim of accurately assessing its fire risk in the built environment. Innovatively, this research employed a cone calorimeter equipped with a self−designed insulating sample holder to conduct a systematic experimental study. Additionally, it performed a comprehensive analysis of the ignition characteristics, heat release rate, fire hazard, smoke release, and toxic gas emission of XPS materials. The experimental results revealed that the combustion behavior of XPS is influenced by multiple factors, including the content of flame retardants and external heat flux, which significantly affect the fire hazard of XPS. When the thermal radiation intensity escalates from 25 kW/m2 to 55 kW/m2, the peak heat release rate of XPS−B1 rises from 428 kW/m2 to 535 kW/m2, marking an increase of 25.00%. Conversely, the peak heat release rate of XPS−B2 surges from 348 kW/m2 to 579 kW/m2, reflecting a substantial increase of 66.38%. This research not only provides a solid theoretical foundation and detailed experimental data for the fire behavior of XPS materials but also holds significant practical importance for enhancing the fire safety of buildings. Overall, this research contributes to the scientific understanding of XPS insulation materials and supports the development of more effective fire prevention measures in construction.

Optimization Strategies in Consumer Choice Behavior for Personalized Recommendation Systems Based on Deep Reinforcement Learning

In domains such as e-commerce and media recommendations, personalized recommendation systems effectively alleviate the issue of information overload. However, existing systems still face challenges in multimodal data processing, data sparsity, and dynamic changes in user preferences. This paper proposes a Hierarchical Generative Reinforcement Learning Recommendation Optimization framework (HG-RLRO) that addresses these issues by integrating multimodal data, Generative Adversarial Networks (GAN), Inverse Reinforcement Learning (IRL), and Hierarchical Temporal Difference Learning (HTD). HG-RLRO employs a multi-agent architecture to handle textual and image data and utilizes GAN to generate simulated user behavior data to mitigate data sparsity. IRL dynamically infers user preferences across multiple time scales.

An Event Frequency‐Inverse Session Frequency Based Weighting Mechanism for Data Embedding Methodologies

ABSTRACTContextWith the rapid increase in data complexity and volume, analyzing and understanding large‐scale datasets has become a critical challenge, especially in domains like e‐commerce. This paper addresses this challenge by introducing a novel business process that utilizes an event frequency‐inverse session frequency (EF‐ISF) based weighting mechanism for embedding user navigational behavior in e‐commerce websites.ObjectiveThe primary research problem tackled in this study is the detection of patterns and distributions within user behavior data, which is often complicated by dynamic page content and session sparsity. We aim to enhance the interpretability and accuracy of user behavior analysis by integrating EF‐ISF with various embedding techniques, including Word2Vec, Node2Vec, DeepWalk, Struc2Vec, and LSTM Autoencoder.MethodologyOur approach is particularly effective in datasets where page refreshes lead to event repetitions, where sequential events with temporal dependencies are prevalent, and where event distributions across sessions are sparse. By applying this methodology across multiple e‐commerce datasets, we observe consistent performance in clustering quality, successfully detecting underlying patterns and groups within the data.ResultsA prototype implementation demonstrates the practical utility of the proposed method, with evaluation metrics confirming that EF‐ISF weighting not only enhances the embeddings but also facilitates a more nuanced and significant segmentation of users' browsing behaviors.ConclusionThese findings provide a valuable contribution to the field of e‐commerce data analytics, offering businesses a refined approach to optimizing user experience and engagement through a deeper understanding of user behavior patterns. This methodology can significantly improve the quality of insights derived from user behavior data, ultimately leading to better decision‐making and enhanced user satisfaction.

Analyzing Urban Parks for Older Adults’ Accessibility in Summer Using Gradient Boosting Decision Trees: A Case Study from Tianjin, China

With the acceleration of global aging, outdoor environments, especially urban green space’s planning and design, play a crucial role in not only promoting physical health but also significantly increasing the opportunities for social interactions for older adults. In recent years, the study of age-friendly outdoor environments has attracted increasing attention, with digital methods emerging as essential tools due to their precision and versatility. In this research, three parks in the Nankai District, Tianjin, are taken as the subject of a case study to explore the spatial factors that may exert influence on the behavior distribution of older adults in summery urban parks’ planning and design. With the behavior data of the older adults in the park collected using an Insta360 camera every hour (from 8 a.m. to 15 p.m.), the three parks are divided into a total of 49 areas for further analysis. Additionally, the visual indexes of the spatial syntax are analyzed with Depthmap 10, the sunlight conditions are analyzed with the Tangent model, and some other spatial factors, such as the green space ratio and the hard ground ratio, are calculated according to the semantic segmentation of the 360-degree panoramic view photo from the center of every area. SPSS and Gradient Boosting Decision Trees (GBDTs) are used to reveal not only the correlations between the sunlight conditions and the behavior distribution of behavior of the older adults, but also the importance ranking of spatial factors. Furthermore, some improvement strategies are proposed for spatial facility configuration, park furniture arrangement, rational hardscape planning, as well as greening and landscape design. By exploring how to improve the spatial planning and design of summery urban green space for older adults, this research provides guidance on the creation of urban green spaces in extremely hot weather that are not only visually appealing but also socially equitable and environmentally sustainable.

Identifying E-Commerce Fraud Through User Behavior Data: Observations and Insights

Abstract Traditional fraud detection approaches often use linking entities, such as device, email, and address, to identify fraudulent transactions and users. However, as fraud methods continue to evolve and escalate, the fraudsters can fabricate the involved entities and thus hide their real intent. To make fraud detection more robust, we incorporate user behaviors in the pipeline and consider biometric characteristics that are difficult to forge. In this work, we conduct a detailed study of how user behavior data can help identify and prevent fraudulent activity in e-commerce. We present Multi-Modal Behavioral Transformer (MMBT), where we combine both inner-page behavioral data, such as mouse trajectory, and inter-page behavioral data, such as page view sequences. We propose to construct mouse trajectory data as an image, treat each mouse position as a pixel in the image, convert the image into small patches, and hence transform the mouse trajectory into patch index sequences. Our experimental results on real-word data show that MMBT significantly outperforms baselines — the precision@recall = 0.1 increases by up to 7%. In addition, we have built an online pipeline to operationalize our model. In production, the 99th percentile latency is maintained below 500 milliseconds, allowing the platform to initiate rapid response measures and prevent potential losses.

Optimization of the Effect of Network Information Flow Advertising Based on User Behavior Data Encryption and Deep Learning

Existing network information flow advertising mostly relies on user behavior data, which is easy to cause user privacy leakage and low precision of advertising delivery. This article combines user behavior data encryption and deep learning technology to propose a method for optimizing the effect of network information flow advertising. The user behavior data is encrypted using Paillier homomorphic encryption technology. Then, the stacked auto encoder (SAE) is used to extract high-order nonlinear features, and the shallow features are automatically combined based on Field-aware Factorization Machines (FFM) to construct a SAE-FFM comprehensive model. At the same time, dropout is used to reduce model overfitting, and the advertising effect is optimized through two-stage pre-training and fine-tuning. The experimental results show that when the number of impressions is 10,000, the average advertising click through rate and conversion rate of the SAE-FFM model are 4.50% and 0.59% respectively, and the information leakage rate is 0.052%. The results show that the SAE-FFM model can greatly improve the click through rate and conversion rate of advertisements, prevent privacy leakage, and help the development of the information flow advertising industry.

Enhancing intelligent marketing systems: a multi-layer hypernetwork approach integrating evidence theory for influential node identification

PurposeThis study proposes an intelligent marketing system model based on a combination of multi-layer hypernetworks and evidence theory, aiming to address the shortcomings of traditional marketing models in accurately identifying key nodes. We propose a new method to improve the accuracy and response speed of intelligent marketing systems by combining evidence theory with multi-layer hypernetworks. We conducted an experiment using a certain car brand (SUV) as an example, which has a wide customer base in both domestic and international markets and has branches in multiple countries. By analyzing its sales data and user behavior, we evaluated the potential reduction in advertising costs and improvement in user satisfaction that may result from adopting this model.Design/methodology/approachThe proposed model begins with the development of a user interest model, which is subsequently converted into a user label model based on user behavior and a rating matrix. A multi-layer aggregation hypernetwork is then constructed to define the network’s topology. An identification framework is established using evidence theory, and the Dempster–Shafer (D-S) evidence combination method is applied to integrate local, positional and global network indicators. Simulation experiments are conducted to evaluate the model’s performance.FindingsThis study proposes an intelligent marketing system model that integrates multi-layer hypernetworks with Dempster–Shafer evidence theory to address the limitations of traditional marketing models in identifying influential nodes. The proposed model is tested in the automotive industry, specifically using sales and user behavior data from a well-known SUV brand operating globally. This industry provides a complex and competitive environment ideal for validating the model’s ability to improve marketing precision. The results demonstrate that the model significantly enhances the accuracy of key node identification, reduces advertising costs by 10–15% and improves customer satisfaction scores to over 90%. Furthermore, preliminary experiments in the retail and e-commerce sectors highlight the model’s adaptability and potential for broader application. By combining local, positional and global indicators, the model effectively optimizes marketing strategies, providing a novel framework for intelligent decision-making in diverse industries. This study selected a well-known SUV car brand as the experimental subject. This brand mainly sells SUV models and has a wide customer base worldwide. Its products are known for their high performance and reliability. The brand has millions of customers, and its main markets include North America, Europe and Asia. It has branches in multiple countries and has significant international influence. According to publicly available data, the brand’s annual revenue reaches billions of dollars.Originality/valueThe main contribution of the research is the proposal of a novel intelligent marketing optimization framework based on multi-layer hypernetworks and evidence theory, which can effectively solve the problems of data silos and information asymmetry faced in traditional marketing systems.

Predicting Seizures Episodes and High-Risk Events in Autism Through Adverse Behavioral Patterns.

Individuals with Autism Spectrum Disorder may display interfering behaviors that limit their inclusion in educational and community settings, negatively impacting their quality of life. These behaviors may also signal potential medical conditions or indicate upcoming high-risk behaviors. This study explores behavior patterns that precede high-risk, challenging behaviors or seizures the following day. We analyzed an existing dataset of behavior and seizure data from 331 children with profound ASD over nine years. We developed a deep learning-based algorithm designed to predict the likelihood of aggression, elopement, and self-injurious behavior (SIB) as three high-risk behavioral events, as well as seizure episodes as a high-risk medical event occurring the next day. The proposed model attained accuracies of 78.4%, 80.68%, 85.43%, and 69.95% for predicting the next-day occurrence of aggression, SIB, elopement, and seizure episodes, respectively. The results were proven significant for more than 95% of the population for all high-risk event predictions using permutation-based statistical tests. Our findings emphasize the potential of leveraging historical behavior data for the early detection of high-risk behavioral and medical events, paving the way for behavioral interventions and improved support in both social and educational environments.

MODELING CUSTOMER LIFETIME VALUE WITH MARKOV CHAIN IN THE INSURANCE INDUSTRY

In the competitive insurance industry, accurately predicting Customer Lifetime Value (CLV) is vital for sustaining long-term profitability and optimizing resource allocation. Traditional static models often fail to capture the dynamic and uncertain nature of customer behavior, which is influenced by factors such as life changes, economic conditions, and evolving product offerings. To address these limitations, this paper proposes an advanced modeling approach that integrates Markov Chains with survival analysis. Markov Chains are well-suited for modeling stochastic processes, where future states depend on current conditions, while survival analysis provides insights into event timing and likelihood for estimating the insurance premium. The proposed model combines these approaches to make a more complete and accurate prediction of CLV. This helps insurers make better decisions and improves the overall performance of their business. We employ the data of customer behavior from the insurance company in Bandung, Indonesia from 1994 to 2020. We found that CLV in the insurance industry is significantly affected by customer behavior.

Optimizing Learning Path Design in Mobile Learning Platforms for Online Courses

With the rapid advancement of information technology, online education, particularly vocational education, has become a vital avenue for enhancing skills and knowledge. Vocational education necessitates flexible and personalized learning path design to accommodate the diverse needs and behaviors of learners. Mobile learning platforms, as a pivotal form of modern online education, provide learners with the convenience of accessing educational resources anytime and anywhere. However, existing methods for optimizing learning paths exhibit notable limitations, primarily in accurately capturing learners’ dynamic behaviors and in providing personalized and intelligent path design. Therefore, how to optimize learning paths based on learners’ dynamic behavior data has become a research hotspot in academia and educational practice. At present, many studies focus on matching analysis based on students’ static characteristics with course content but overlook learners’ behavioral changes and dynamic needs during the learning process. Traditional recommendation algorithms and rule-based path design methods are difficult to cope with complex learning behaviors and diverse learner needs. This study addresses these limitations by proposing an optimization model for mobile learning path design based on multi-view prediction of dynamic student behaviors. The model extracts mobile interaction features from student groups, constructs a multiple mobile behavior collaborative encoder, and employs multiple task label prediction techniques to achieve personalized and intelligent optimization of learning paths. The results demonstrate that this approach significantly enhances the learning efficiency and experience of learners, offering novel insights and technological support for the path design of mobile learning platforms.