Gilbert Saporta - A Semi-Supervised Recommender System to Predict Online Job Offer Performance

Job vacancies are more and more published exclusively on internet . In order to predict job posting performance on a job board website , a Ph.D. sponsored by Multiposting company is prepared by J.Séguéla. An intelligent tool is developed, which recommends the best job boards according to the job offer. The system is based on the analysis of both structured and unstructured (textual) characteristics of previous job offers. We present here a semi-supervised recommender system predicting the ranking of job boards with respect to job posting returns and compare it with other supervised and non supervised approaches.

Keywords : Statistical methodology includes Latent Semantic Indexing, PLS regression and k-nn prediction.