Abstract:Recently, due to the ubiquity and supremacy of E-recruitment platforms, job recommender systems have been largely studied. In this paper, we tackle the next job application problem, which has many practical applications. In particular, we propose to leverage next-item recommendation approaches to consider better the job seeker's career preference to discover the next relevant job postings (referred to jobs for short) they might apply for. Our proposed model, named Personalized-Attention Next-Application Prediction (PANAP), is composed of three modules. The first module learns job representations from textual content and metadata attributes in an unsupervised way. The second module learns job seeker representations. It includes a personalized-attention mechanism that can adapt the importance of each job in the learned career preference representation to the specific job seeker's profile. The attention mechanism also brings some interpretability to learned representations. Then, the third module models the Next-Application Prediction task as a top-K search process based on the similarity of representations. In addition, the geographic location is an essential factor that affects the preferences of job seekers in the recruitment domain. Therefore, we explore the influence of geographic location on the model performance from the perspective of negative sampling strategies. Experiments on the public CareerBuilder12 dataset show the interest in our approach.

What problem does this paper attempt to address?

### Problems Addressed by the Paper The paper aims to address the following issues: 1. **Dynamic and Ever-Changing Professional Interaction Relationships**: In the recruitment field, the interaction between jobs and individuals is dynamic and constantly changing. This interaction can be constructed through job application records or work experiences. The paper points out that although many conversation-based methods have been proposed in other fields, there is little research on conversation-based job recommendation in the recruitment field. Additionally, since most job seekers apply for only a few positions, the job-person interaction matrix is very sparse. 2. **Personalized Recommendation Problem**: Different job seekers may apply for the same position for different reasons. Therefore, the important information about a position is personalized for each job seeker. Existing general methods often fail to capture these specific details. Moreover, in the recruitment field, the personal background information of job seekers (such as geographic location) significantly impacts the recommendation results. 3. **Evolutionary Problem in the Recruitment Field**: The recruitment field itself is also constantly evolving, with new jobs, skills, and training emerging continuously. This makes it impractical to build static methods, as the data is constantly changing. Therefore, a method that can adapt to changes is needed to avoid retraining the model every time the data is updated. Additionally, due to the sensitivity and privacy issues of recruitment data, it is difficult to create an unbiased and continuously evolving labeled dataset, which limits the application of supervised learning methods. The paper proposes a hybrid model with a personalized attention mechanism (PANAP) to partially address the above issues. Specifically: - By modeling the next job application prediction problem as a sequential recommendation problem and comparing it with other conversation-based recommendation methods, the paper analyzes the dynamics of the recruitment field. - The model consists of three independent modules: the first module learns job representations from text content and metadata in an unsupervised manner; the second module includes a personalized attention mechanism to learn the representation of job seekers; the third module is used to predict the next position that job seekers might apply for. Experimental results show that the PANAP model and its variants outperform baseline models on multiple evaluation metrics, especially under the negative sampling strategy that considers geographic location factors.

Improving Next-Application Prediction with Deep Personalized-Attention Neural Network

Feature-Level Attentive Neural Model for Session-Based Recommendation

Job and Candidate Recommendation with Big Data Support: A Contextual Online Learning Approach

Self-Attention Mechanism Enhanced User Interests Modeling for Personalized Recommendation Services in Cyber-Physical-Social Systems.

Attentive Sequential Model Based on Graph Neural Network for Next Poi Recommendation.

An Enhanced Neural Network Approach to Person-Job Fit in Talent Recruitment

Job Prediction: From Deep Neural Network Models to Applications

Tree-based Contextual Learning for Online Job or Candidate Recommendation with Big Data Support in Professional Social Networks

DeepApp

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

Resume Activeness Prediction in Online Recruitment Scenarios

Enhancing Person-Job Fit for Talent Recruitment: An Ability-aware Neural Network Approach

A Context Awareness Hierarchical Attention Network for Next POI Recommendation in IoT Environment

NPA: Neural News Recommendation with Personalized Attention

Person-job fit estimation from candidate profile and related recruitment history with co-attention neural networks

A Next Location Predicting Approach Based on a Recurrent Neural Network and Self-attention.

DeepApp: Predicting Personalized Smartphone App Usage via Context-Aware Multi-Task Learning

A Job Recommendation Model Based on a Two-Layer Attention Mechanism

Personalized Next Point-of-Interest Recommendation via Latent Behavior Patterns Inference

Next-App Prediction by Fusing Semantic Information with Sequential Behavior

Knowledge-enhanced personalized hierarchical attention network for sequential recommendation