Exploring Data Mining Techniques And Applications For Intelligent Recruitment

In this competitive world,talents have become a key asset for organization growth and success.Therefore,talent recruitment has been set as the top priority in the strategic plan of many companies and organizations,which have made great efforts to develop the intelligent recruitment systems to help for attracting,identifying,and selecting suitable talents in an intelligent way.Recently,with the rapid development of online recruitment platforms,talent recruitment has gradually evolved from a seller's market with unequal information to a market where information is reciprocal in supply and demand.As a result,the competition in the recruitment market has become increasingly fierce,and thus has brought a series of new challenges to talent recruitment.Meanwhile,the rapid development of digitalized recruitment systems and online recruitment platforms also generate a large amount of recruitment data,which enable a new paradigm for the development of intelligent recruitment systems.Indeed,the study on intelligent recruitment has become a very important research direction in computer science,management science and other interdisciplinary fields.To this end,in this dissertation,we introduce the exploratory research efforts on three core scenarios of talent recruitment,including talent attraction,talent sourcing and talent assessments,in a systematic way.To be specific,by exploiting several data mining techniques and combining with the domain knowledge from human resource management,we propose the method for automatic job requirement generation with the consideration of the skill requirement,the method for measuring the matching degree between talents and jobs,the method for automatic skill-oriented question bank generation and intelligent question retrieval,and the method for personalized interview question recommendation by unitizing skill-graph.In particular,all of these studies are conducted on the platform of the Baidu Talent Intelligent system.The proposed methods have been deployed and validated on the realworld talent recruitment system.Specifically,the main research contributions of this dissertation can be summarized as follows.First,for the purpose of attracting talents,we propose an end-to-end capability-aware approach,namely Cajon,for automatic job requirement generation with the consideration of skill requirements.This research is based on real-world job description data extracted from large-scale recruitment data.Our approach can predict the skill requirements effectively for different types of jobs and assist the recruiter to facilitate the preparation of job posting Ads.To be specific,we first propose a novel capability-aware neural topic model to distill the various capability information from the larger-scale recruitment data.Also,an encoder-decoder recurrent neural network is designed for enabling the job requirement generation.In particular,the capability-aware attention mechanism and copy mechanism are proposed to guide the generation process to ensure the generated job requirement can comprehensively cover relevant and representative capability requirements for the job.Moreover,we propose a capability-aware policy gradient training algorithm to further enhance the rationality of the generated job requirement.Finally,extensive experiments on real-world recruitment data clearly show our Cajon framework can help to generate more effective job requirements and cover the reasonable capability requirement in an interpretable way.Second,in the talent sourcing scenario,we propose an ability-aware approach for Person-Job Fit task,namely TAPJFNN,which can effectively measure the matching degree between talents and job postings,and thus can improve the efficiency of talent sourcing.Specifically,we propose a word-level semantic representation for both job requirements and job seekers' experiences based on Recurrent Neural Network(RNN).Along this line,two hierarchical topic-based ability-aware attention strategies are designed to measure the different importance of job requirements for semantic representation,as well as measuring the different contribution of each job experience to a specific ability requirement.In addition,we design a refinement strategy for Persona-Job Fit prediction based on historical recruitment records.Furthermore,we introduce how to exploit our proposed framework for enabling two specific applications in talent recruitment,namely talent sourcing and job recommendation.Finally,extensive experiments on a large-scale real-world dataset clearly validate the effectiveness and interpretability of the TAPJFNN and its variants compared with several baselines.Third,in the talent assessment scenario,we propose an approach for automatic skill-oriented question bank generation and intelligent question retrieval.Also,we develop an intelligent interview assistance system,namely DuerQues,which can help interviewers design and select suitable questions in an efficient and effective way.To be specific,we first investigate how to automatically generate skill-oriented interview questions in a scalable way by learning the external knowledge from online knowledge sharing communities.Along this line,we develop a novel distantly supervised skill entity recognition method to identify the skill entities from the large-scale search query data and web page titles with the help of lower-cost human annotation.Also,we propose a neural generative model for generating skill-oriented interview questions.In particu-lar,we introduce a data-driven solution to generate the high-quality training instances and design a learning method for improving the performance of the question generation.Meanwhile,we exploit the click-through data from the query logs and design a recom-mender system for helping interviewers retrieval the suitable question more efficiency.Specifically,we design a graph-enhanced question recommendation algorithm which can recommend the suitable questions efficiently when an interviewer queries a set of skills.Finally,extensive experiments on the real-world datasets have clearly demon-strated the effectiveness of our DuerQues system in terms of the quality of generated skill-oriented questions and the performances of the question retrieval.Last but not least,we also propose an approach for personalized interview ques-tion recommendation by unitizing skill-graph,and develop a system,namely DuerQuiz,which can assess the job seekers more effectiveness.DuerQuiz is a fully deployed real-world system,where a knowledge graph of job skills,Skill-Graph,has been built for comprehensively modeling the relevant competencies that should be assessed in the job interview.Specifically,we first develop a novel skill entity extraction approach based on a bidirectional Long ShortTerm Memory(LSTM)with a Conditional Random Field(CRF)layer(LSTM-CRF)neural network enhanced with adapted gate mechanism.In particular,to improve the reliability of extracted skill entities,we design a label prop-agation method based on more than 10 billion click-through data from the large-scale query logs of a search engine.Furthermore,we discover the hypernym-hyponym re-lations between skill entities and construct the Skill-Graph by leveraging the classifier trained with extensive contextual features.Finally,we design a personalized question recommendation algorithm based on the Skill-Graph for improving the efficiency and effectiveness of job interview assessment.Finally,extensive experiments on a large-scale real-world dataset clearly validate the effectiveness of the proposed system.