Abstract:It is always challenging task for students to select right universities. For students, graduate job placement is the most important component of university quality. However, existing university evaluation methods predominantly depend on either subjective criteria, such as the perceived quality of the learning environment and academic prestige, or on factors like faculty excellence, which may not provide a precise indication of graduate job placement. Indeed, there is still a lack of a data-driven approach to accurately measure university quality based on the employment situation of graduates. Moreover, the inherently unsupervised nature of university evaluation, compounded by the absence of a reasonable ground truth, necessitates the development of a reliable supervised methodology to precisely quantify university quality. Our basic assumption is that highly influential companies would attract graduates from high-ranking universities. To this end, in this paper, we formulate university evaluation problem into the graduate flow prediction problem, and propose an Influence based Graph Autoencoder (IGAE) method to learn the representation of universities based on the employment of their graduates. Specifically, we first build a talent transition graph based on the massive resume information. This graph reveals the flow of talent between institutions. Then, considering the asymmetric and heterogeneous properties of talent flow, an unidirectional aggregation process with a heterogeneous attention mechanism is designed to encode the nodes in the directed graph and preserve the influence terms at the same time. Afterwards, a novel dual self-attention module is exploited to capture the dynamic pattern of institutions to forecast future employment. Furthermore, we design an influence based decoder to predict the existence of talent flows and estimate the frequency of employment, which can be learnt in a joint learning framework. Finally, we conduct extensive experiments on a real-world dataset for performance evaluation. The experimental results clearly validate the effectiveness of our approach compared to the state-of-the-art baselines, and we provide a case study on university influence analysis.

Career Mobility Analysis With Uncertainty-Aware Graph Autoencoders: A Job Title Transition Perspective

A Multi-Channel Multi-Tower GNN Model for Job Transfer Prediction Based on Academic Social Network.

Attentive Heterogeneous Graph Embedding for Job Mobility Prediction

Task-Oriented Genetic Activation for Large-Scale Complex Heterogeneous Graph Embedding.

Spatial-Temporal Graph Discriminant AutoEncoder for Traffic Congestion Forecasting

Towards Unified Representation Learning for Career Mobility Analysis with Trajectory Hypergraph

University Evaluation Through Graduate Employment Prediction: an Influence Based Graph Autoencoder Approach

A Hierarchical Career-Path-Aware Neural Network for Job Mobility Prediction

Job2Vec: Job Title Benchmarking with Collective Multi-View Representation Learning

Cyclic Graph Attentive Match Encoder (CGAME): A Novel Neural Network for OD Estimation

Being Automated or Not? Risk Identification of Occupations with Graph Neural Networks

Pre-DyGAE: Pre-training Enhanced Dynamic Graph Autoencoder for Occupational Skill Demand Forecasting

Remote Work Optimization with Robust Multi-channel Graph Neural Networks

Deep Occupational Representations Based on Career Mobility

Understanding stock market instability via graph auto-encoders

Extracting Job Title Hierarchy from Career Trajectories: A Bayesian Perspective

What Is My Next Job: Predicting The Company Size And Position In Career Changes

Uncertainty-aware Human Mobility Modeling and Anomaly Detection

Automation and occupational mobility: A data-driven network model

T-GAE: Transferable Graph Autoencoder for Network Alignment

JobComposer: Career Path Optimization via Multicriteria Utility Learning