Graph representation learning—especially via graph neural networks (GNNs)—has demonstrated considerable promise in modeling intricate interaction systems, such as social networks and molecular structures. However, the deployment of GNN-based frameworks in industrial settings remains challenging due to the inherent complexity and noise in real-world graph data. This dissertation systematically addresses these challenges by advancing novel methodologies to improve the comprehensiveness and robustness of graph representation learning, with a dual focus on resolving data complexity and denoising across diverse graph-learning scenarios. In addressing graph data denoising, we design auxiliary self-supervised optimization objectives that disentangle noisy topological structures and misinformation while preserving the representational sufficiency of critical graph features. These tasks operate synergistically with primary learning objectives to enhance robustness against data corruption. The efficacy of these techniques is demonstrated through their application to real-world opioid prescription time series data for predicting potential opioid over-prescription. To mitigate data complexity, the study investigates two complementary approaches: (1) multimodal fusion, which employs attentive integration of graph data with features from other modalities, and (2) hierarchical substructure mining, which extracts semantic patterns at multiple granularities to enhance model generalization in demanding contexts. Finally, the dissertation explores the adaptability of graph data in a range of practical applications, including E-commerce demand forecasting and recommendations, to further enhance prediction and reasoning capabilities.

As education increasingly relies on data-driven methodologies, accurately predicting student performance is essential for implementing timely and effective interventions. The California Student Performance Dataset offers a distinctive basis for analyzing complex elements that affect educational results, such as student demographics, academic behaviours, and emotional health. This study presents the GNN-Transformer-InceptionNet (GNN-TINet) model to overcome the constraints of prior models that fail to effectively capture intricate interactions in multi-label contexts, where students may display numerous performance categories concurrently. The GNN-TINet utilizes InceptionNet, transformer architectures, and graph neural networks (GNN) to improve precision in multi-label student performance forecasting. Advanced preprocessing approaches, such as Contextual Frequency Encoding (CFI) and Contextual Adaptive Imputation (CAI), were used on a dataset of 97,000 occurrences. The model achieved exceptional outcomes, exceeding current standards with a Predictive Consistency Score (PCS) of 0.92 and an accuracy of 98.5%. Exploratory data analysis revealed significant relationships between GPA, homework completion, and parental involvement, emphasizing the complex nature of academic achievement. The results illustrate the GNN-TINet’s potential to identify at-risk pupils, providing a robust resource for educators and policymakers to improve learning outcomes. This study enhances educational data mining by enabling focused interventions that promote educational equality, tackling significant challenges in the domain.

Introduction This dataset contains the data described in the paper titled "A deep neural network approach to predicting clinical outcomes of neuroblastoma patients." by Tranchevent, Azuaje and Rajapakse. More precisely, this dataset contains the topological features extracted from graphs built from publicly available expression data (see details below). This dataset does not contain the original expression data, which are available elsewhere. We thank the scientists who did generate and share these data (please see below the relevant links and publications). Content File names start with the name of the publicly available dataset they are built on (among "Fischer", "Maris" and "Versteeg"). This name is followed by a tag representing whether they contain raw data ("raw", which means, in this case, the raw topological features) or TF formatted data ("TF", which stands for TensorFlow). This tag is then followed by a unique identifier representing a unique configuration. The configuration file "Global_configuration.tsv" contains details about these configurations such as which topological features are present and which clinical outcome is considered. The code associated to the same manuscript that uses these data is at https://gitlab.com/biomodlih/SingalunDeep. The procedure by which the raw data are transformed into the TensorFlow ready data is described in the paper. File format All files are TSV files that correspond to matrices with samples as rows and features as columns (or clinical data as columns for clinical data files). The data files contain various sets of topological features that were extracted from the sample graphs (or Patient Similarity Networks - PSN). The clinical files contain relevant clinical outcomes. The raw data files only contain the topological data. For instance, the file "Fischer_raw_2d0000_data_tsv" contains 24 values for each sample corresponding to the 12 centralities computed for both the microarray (Fischer-M) and RNA-seq (Fischer-R) datasets. The TensorFlow ready files do not contain the sample identifiers in the first column. However, they contain two extra columns at the end. The first extra column is the sample weights (for the classifiers and because we very often have a dominant class). The second extra column is the class labels (binary), based on the clinical outcome of interest. Dataset details The Fischer dataset is used to train, evaluate and validate the models, so the dataset is split into train / eval / valid files, which contains respectively 249, 125 and 124 rows (samples) of the original 498 samples. In contrast, the other two datasets (Maris and Versteeg) are smaller and are only used for validation (and therefore have no training or evaluation file). The Fischer dataset also has more data files because various configurations were tested (see manuscript). In contrast, the validation, using the Maris and Versteeg datasets is only done for a single configuration and there are therefore less files. For Fischer, a few configurations are listed in the global configuration file but there is no corresponding raw data. This is because these items are derived from concatenations of the original raw data (see global configuration file and manuscript for details). References This dataset is associated with Tranchevent L., Azuaje F.. Rajapakse J.C., A deep neural network approach to predicting clinical outcomes of neuroblastoma patients. If you use these data in your research, please do not forget to also cite the researchers who have generated the original expression datasets. Fischer dataset: Zhang W. et al., Comparison of RNA-seq and microarray-based models for clinical endpoint prediction. Genome Biology 16(1) (2015). doi:10.1186/s13059-015-0694-1 Wang C. et al., The concordance between RNA-seq and microarray data depends on chemical treatment and transcript abundance. Nat. Biotechnol. 32(9), 926–932. doi:10.1038/nbt.3001 Versteeg dataset: Molenaar J.J. et al., Sequencing of neuroblastoma identifies chromothripsis and defects in neuritogenesis genes. Nature 483(7391), 589–593. doi:10.1038/nature10910 Maris dataset: Wang Q. et al., Integrative genomics identifies distinct molecular classes of neuroblastoma and shows that multiple genes are targeted by regional alterations in DNA copy number. Cancer Res. 66(12), 6050–6062. doi:10.1158/0008-5472.CAN-05-4618 Project supported by the Fonds National de la Recherche (FNR), Luxembourg (SINGALUN project). This research was also partially supported by Tier-2 grant MOE2016-T2-1-029 by the Ministry of Education, Singapore.

Skills Graph Platform Market Outlook

According to our latest research, the global Skills Graph Platform market size is valued at USD 2.15 billion in 2024, with a robust growth trajectory expected over the coming years. The market is forecasted to reach USD 9.87 billion by 2033, expanding at a notable CAGR of 18.4% during the period from 2025 to 2033. This surge is primarily driven by the increasing demand for intelligent talent management, dynamic workforce planning, and the integration of advanced analytics and artificial intelligence into human capital management systems. As organizations worldwide seek to optimize talent acquisition, upskilling, and workforce agility, the adoption of Skills Graph Platforms is experiencing unprecedented momentum.

The primary growth factor fueling the Skills Graph Platform market is the accelerating digital transformation initiatives across enterprises and institutions. Organizations are increasingly recognizing the value of mapping, analyzing, and leveraging employee skills data at scale to drive business outcomes. The proliferation of remote and hybrid work models, coupled with the need to bridge skill gaps in a rapidly evolving labor market, is compelling businesses to invest in sophisticated platforms that provide real-time insights into workforce competencies. These platforms enable data-driven decisions for hiring, upskilling, and redeployment, which are critical in maintaining competitiveness amid technological disruption and shifting market demands. The rise of AI-powered talent analytics and the integration of machine learning algorithms further enhance the accuracy and predictive capabilities of these platforms, making them indispensable tools for future-ready enterprises.

Another significant driver is the increasing focus on personalized learning and development strategies within organizations and educational institutions. As the demand for continuous learning intensifies, Skills Graph Platforms play a pivotal role in identifying individual skill gaps, recommending tailored learning paths, and tracking progress. This level of personalization not only improves employee engagement and retention but also ensures that organizations can cultivate a highly skilled and adaptable workforce. Educational institutions, too, are leveraging these platforms to align curricula with industry requirements, thereby enhancing student employability and fostering stronger industry-academia collaborations. The integration of Skills Graph Platforms with existing Learning Management Systems (LMS) and Human Resource Information Systems (HRIS) is streamlining talent development processes and delivering measurable ROI.

The market is further bolstered by the growing emphasis on data-driven workforce planning and diversity, equity, and inclusion (DEI) initiatives. Skills Graph Platforms provide granular visibility into workforce demographics, skills distribution, and talent mobility, enabling organizations to design targeted interventions for workforce diversification and succession planning. Governments and regulatory bodies are also encouraging the adoption of such platforms to support national upskilling agendas and enhance labor market transparency. As businesses navigate complex regulatory environments and strive to meet ESG (Environmental, Social, and Governance) objectives, the ability to track and report on workforce competencies becomes a strategic imperative. This regulatory push, combined with the need for agile, skills-based workforce strategies, is expected to sustain market growth throughout the forecast period.

Regionally, North America continues to dominate the Skills Graph Platform market, owing to its mature HR technology ecosystem, high digital adoption rates, and a strong presence of leading solution providers. However, Asia Pacific is emerging as the fastest-growing region, driven by rapid industrialization, increasing investments in digital infrastructure, and a burgeoning demand for skilled talent. Europe also presents significant growth opportunities, particularly in sectors such as manufacturing, BFSI, and healthcare, where workforce agility and compliance are paramount. Latin America and the Middle East & Africa are witnessing steady adoption, supported by government-led digital skills initiatives and the expansion of multinational enterprises in these regions.

Component Analysis

The Skills Graph Platform market is segmented by component into Software and <

BackgroundResidents need to be trained across the boundaries of their own specialty to prepare them for collaborative practice. Intraprofessional learning (i.e. between individuals of different disciplines within the same profession) has received little attention in the postgraduate medical education literature, in contrast to the extensive literature on interprofessional learning between individuals of different professions. To address this gap, we performed a scoping review to investigate what and how residents learn from workplace-related intraprofessional activities, and what factors influence learning.MethodsThe PRISMA guidelines were used to conduct a scoping review of empirical studies on intraprofessional workplace learning in postgraduate medical education published between 1 January 2000 to 16 April 2020 in Pubmed, Embase, PsycINFO, ERIC and Web of Science.Inclusion criteria: focus on intraprofessional learning (i.e. the learning that occurs when two or more disciplines of the same profession engage), involves primary and/or secondary care postgraduate medical trainees, workplace learning: incidental and informal, intentional non-formal, and/or formal, contains empirical evidence from qualitative, quantitative or mixed methods studies, published in a peer-reviewed journal.Exclusion criteria: does not meet inclusion criteria of focus on intraprofessional learning, primary and/or secondary care postgraduate medical trainees and workplace learning, grey literature, reviews, commentaries, book, papers only describing curricula (no empirical data), publication before 2000, written in another language than English, unable to retrieve abstract or full-text paper.4330 records were screened, and finally 37 articles were included.The data was extracted using a data extraction chart. The collected data includes: article characteristics, study characteristics (study type, study design), participant characteristics (professions and specialties involved, sample size), characteristics of the intraprofessional workplace learning activity (type of workplace learning, description of the learning activity, learner role, supervision, duration and frequency), and learning outcomes.

Performance evaluation of proposed method GNN-TINet and existing methods.

In 2022, about 37.7 percent of the U.S. population who were aged 25 and above had graduated from college or another higher education institution, a slight decline from 37.9 the previous year. However, this is a significant increase from 1960, when only 7.7 percent of the U.S. population had graduated from college. Demographics Educational attainment varies by gender, location, race, and age throughout the United States. Asian-American and Pacific Islanders had the highest level of education, on average, while Massachusetts and the District of Colombia are areas home to the highest rates of residents with a bachelor’s degree or higher. However, education levels are correlated with wealth. While public education is free up until the 12th grade, the cost of university is out of reach for many Americans, making social mobility increasingly difficult. Earnings White Americans with a professional degree earned the most money on average, compared to other educational levels and races. However, regardless of educational attainment, males typically earned far more on average compared to females. Despite the decreasing wage gap over the years in the country, it remains an issue to this day. Not only is there a large wage gap between males and females, but there is also a large income gap linked to race as well.

Additional file 3 Full results of all models. Each model is described by its parameters and the corresponding balanced accuracy. Archive of XLSX files.

A. Difference in the proportions of the Likert Scale options by Brazil region for the first five questions. B. Difference in the proportions of the Likert Scale by Brazil region for the last five questions. (ZIP)

The data on the effect of using SOLO taxonomy on secondary mathematical competencies associated with quadratic equations was collected between May and June 2024 from three secondary schools in Mafinga district (Zambia). The sample involved 78 grade 11 students from three intact classes as experimental groups and 71 grade 11 students from three intact classes as comparison groups randomly selected. The researcher analyzed learning outcomes associated with quadratic equations in the Zambian secondary school mathematics syllabus and identified three expected competencies namely; algebraic graphing skills, algebraic reasoning skills and algebraic representation skills. With these competencies, four questions for testing students’ understanding of quadratic equations were formulated in line with the increasing complexity of the response structure reflected in the SOLO categories. Fifteen lessons were conducted to complete quadratic equations in five weeks. In each lesson, think-pair share, group and individual activities were implemented as formative assessments to monitor learning progress. Students from experimental groups completed formative assessments with formative feedback based on SOLO taxonomy that encouraged self-assessment and peer feedback. Students took the mathematics competence test within one-hour thirty minutes before and after learning quadratic equations. Students’ responses to quadratic tasks were analyzed using grading rubrics based on SOLO taxonomy that described the levels of understanding from pre-structural level (no understanding), uni-structural level and multi-structural level (surface understanding), relational level (deep understanding) to expended abstract level (conceptual understanding). The data set includes an Excel file named “Data set on students' competencies associated with quadratic equations based on SOLO taxonomy” with three sheets containing raw data for students before and after interventions: Graphing skills sheet, Reasoning skills sheet and Representation skills sheet. The levels of performance ranged from 0 = pre-structural level, 1 = uni-structural level, 2 = multi-structural level, 3 = relational level and 4 = uni-structural level. Students classified under relational and extended abstract level categories demonstrated higher order thinking skills while those under uni-structural and multi-structural levels demonstrated low order thinking skills. Students under pre-structural level demonstrated no understanding. Bar charts were used to present the proportion of students on each SOLO level in relation to algebraic graphing skills, algebraic reasoning skills and algebraic representation skills. SPSS v.25 was used to generate the outputs. Regardless of the levels of performance before interventions, the results showed that the proportion of students who achieved higher order thinking skills on expected mathematical competencies was higher for experimental groups than comparison groups after interventions.

Descriptive analysis for Likert scale variables.

A deep understanding of the relationship between the knowledge acquired and the graduation requirements is essential for students to precisely meet the graduation requirements and to become human resources with specific knowledge, skills and professionalism. In this paper, we define the ontology layer of the knowledge graph by deeply analyzing the relationship between graduation requirement, course and knowledge. Based on the implementation of the concept of Outcome Based Education, we use Knowledge extraction, fusion, reasoning techniques to construct a hierarchical knowledge graph with the main line of "knowledge-course-graduation requirements. In the process of knowledge extraction, in order to alleviate the huge labor overhead brought by traditional extraction methods, this paper adopts a transfer learning method to extract triadic knowledge using the multi-task framework EERJE, Finally, knowledge reasoning was also performed with the help of LLM to further expand the knowledge scope. The comprehensiveness, correctness and relatedness of the data were evaluated through the experiment, and the F1 value of the ternary group extraction was 87.76%, the accuracy rate of entity classification was 85.42%, the data coverage was more comprehensive, and the results showed that the data quality was better, and the knowledge graph constructed in this way can fully optimize the organization and management of teaching resources, help students intuitively and comprehensively grasp the correlation and difference between graduation requirements and various knowledge points, and let the Students can carry out personalized independent learning through the navigation mode of knowledge graph, strengthen their weak links, and complete the relevant graduation requirements, which effectively improves the degree of students’ graduation requirements achievement. This new paradigm of knowledge graph enabled teaching is of reference significance for engineering education majors to improve the degree of graduation requirements achievement.

List of factor specificity values.

Difference in the proportions of the Likert scale options by age.

Difference in the propositions of the Likert Scale options by medical specialty.

Types included in each main line structure and related descriptions.

Variables from the Measures of Maternal Stress (MOMS) study included in the graphical model.

Area under the ROC curve and optimism by EHR-continuity.

Attributes of women in the Measures of Maternal Stress (MOMS) study.

Patient characteristics of the study population based on EHR and claims data.