{"id":990,"date":"2021-12-15T08:03:17","date_gmt":"2021-12-15T07:03:17","guid":{"rendered":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/nouveausite\/?page_id=990"},"modified":"2024-03-29T08:23:24","modified_gmt":"2024-03-29T07:23:24","slug":"these-yann-duperis","status":"publish","type":"page","link":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/these-yann-duperis\/","title":{"rendered":"These Yann Duperis"},"content":{"rendered":"

\"\"Yann Duperis<\/span><\/strong>, \u00ab Recommandation de candidats par apprentissage profond bas\u00e9e sur des similarit\u00e9s s\u00e9mantiques entre phrases-cl\u00e9s \u00bb. Th\u00e8se soutenue le vendredi 23 f\u00e9vrier 2024 \u00e0 Marseille.
\n<\/span><\/p>\n

Manuscrit<\/a><\/span><\/p>\n

\n
\n
\n

R\u00e9sum\u00e9:<\/span><\/strong><\/p>\n

La recherche automatis\u00e9e d\u2019un candidat id\u00e9al lors d\u2019un recrutement se heurte \u00e0 la difficult\u00e9 \u00e0 mod\u00e9liser correctement les besoins exprim\u00e9s en langage naturel dans une offre d\u2019emploi et les capacit\u00e9s d\u2019un individu d\u00e9clar\u00e9es dans un CV, notamment pour le recrutement d\u2019experts qui n\u00e9cessite une connaissance pr\u00e9cise du secteur d\u2019activit\u00e9 concern\u00e9. Historiquement, les m\u00e9thodes de recommandation de candidats ou de poste ont utilis\u00e9 des approches classiques de Recherche d\u2019Information bas\u00e9es sur les termes utilis\u00e9s dans les CV et les offres d\u2019emploi. Des m\u00e9thodes bas\u00e9es sur une mod\u00e9lisation symbolique explicite des connaissances li\u00e9es aux domaines d\u2019activit\u00e9s couverts par des activit\u00e9s de recrutement ont aussi \u00e9t\u00e9 propos\u00e9es. Plus r\u00e9cemment, l\u2019arriv\u00e9e de l\u2019apprentissage profond et de l\u2019apprentissage par transfert pour le traitement automatis\u00e9 du langage naturel a offert de nouvelles possibilit\u00e9s.
\nDans cette th\u00e8se, nous pr\u00e9sentons des travaux de recherche visant \u00e0 d\u00e9velopper une m\u00e9thode de recommandation de candidats explicable, bas\u00e9e sur le contenu et pertinente pour le recrutement de profils experts. Notre premi\u00e8re contribution est une m\u00e9thode de pr\u00e9diction de comp\u00e9tences \u00e0 partir des termes d\u2019un profil. La m\u00e9thode propos\u00e9e utilise l\u2019apprentissage profond et offre des performances \u00e9lev\u00e9es pour certaines comp\u00e9tences rares et sp\u00e9cialis\u00e9es. Notre seconde contribution est une m\u00e9thode de recommandation de candidats reposant sur l\u2019exploitation de noms de m\u00e9tiers et de comp\u00e9tences mentionn\u00e9s \u00e0 la fois dans une offre d\u2019emploi et dans un profil, les phrases-cl\u00e9s. Cette m\u00e9thode de recommandation de candidats surpasse une m\u00e9thode de r\u00e9f\u00e9rence et a \u00e9t\u00e9 valid\u00e9e par des recruteurs professionnels au travers de l\u2019usage d\u2019un syst\u00e8me de recommandation impl\u00e9mentant cette m\u00e9thode.
\nMots cl\u00e9s<\/strong> : Syst\u00e8me de recommandation de candidats, Ad\u00e9quation candidat-poste, Apprentissage profond, Apprentissage par transfert, Phrases-cl\u00e9s.<\/span><\/p>\n

Abstract:<\/em><\/span><\/strong><\/p>\n

The automated research of an ideal candidate for a position is hindered by the difficulty to adequately model the need for competency expressed in natural language in a job offer and the abilities of an individual declared in a resume, especially for the recruitment of experts which requires accurate knowledge about the considered activity sector.
\nHistorically, candidate or job recommendation methods have used classical approaches of Information Retrieval based on common terms in both resumes and job offers. Recommendation methods based on explicit symbolic modelling of the knowledge related to the activity sector in which the recruitment is taking place have also been proposed. Recently, the advent of both deep learning and transfer learning for natural language processing offered new possibilities.
\nIn this thesis, we present the research work undertaken to develop a candidate recommendation method which is explainable, content-based, and relevant for the recruitment of expert professionals.
\nOur first contribution is a method to predict an individual\u2019s skills based on the terms used in their resume. The proposed method uses deep learning and delivers high performance for some rare and specialised skills.
\nOur second contribution is a candidate recommendation method which evaluates the adequacy of a candidate for a job based on job titles and skills mentioned in both a resume and a job offer, the keyphrases. The proposed method uses deep learning techniques to match semantically similar keyphrases. This recommendation method outperforms a reference method and has been validated by professional recruiters using a recommendation system implementing this method.
\nKeywords :<\/strong> Candidate recommender system, Person-job fit, Deep Learning, Transfer Learning, Keyphrases.<\/em><\/span><\/p>\n<\/div>\n<\/div>\n

\"\"<\/p>\n

\"\"<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

Yann Duperis, \u00ab Recommandation de candidats par apprentissage profond bas\u00e9e sur des similarit\u00e9s s\u00e9mantiques entre phrases-cl\u00e9s \u00bb. Th\u00e8se soutenue le vendredi 23 f\u00e9vrier 2024 \u00e0 Marseille. Manuscrit R\u00e9sum\u00e9: La recherche automatis\u00e9e d\u2019un candidat id\u00e9al lors d\u2019un recrutement se heurte \u00e0 la difficult\u00e9 \u00e0 mod\u00e9liser correctement les besoins exprim\u00e9s en langage naturel dans une offre d\u2019emploi … <\/p>\n

Continuer la lecture \u00ab\u00a0These Yann Duperis\u00a0\u00bb<\/span><\/a><\/p>\n","protected":false},"author":2,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"_crdt_document":"","footnotes":""},"class_list":["post-990","page","type-page","status-publish","hentry","entry"],"_links":{"self":[{"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/pages\/990","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/comments?post=990"}],"version-history":[{"count":15,"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/pages\/990\/revisions"}],"predecessor-version":[{"id":1718,"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/pages\/990\/revisions\/1718"}],"wp:attachment":[{"href":"https:\/\/pageperso.lis-lab.fr\/bernard.espinasse\/index.php\/wp-json\/wp\/v2\/media?parent=990"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}