pr\u00e9sentent<\/p>\n\n\n\n
Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en :<\/p>\n\n\n\n
le VENDREDI 12 D\u00e9CEMBRE 2025 \u00e0 14h00<\/p>\n\n\n\n
\u00e0<\/p>\n\n\n\n
Amphith\u00e9\u00e2tre 6
T\u00e9l\u00e9com SudParis, 19 place Marguerite Perey, 91120 Palaiseau<\/p>\n\n\n\n
Membres du jury :<\/strong><\/p>\n\n\n\n Mme Francesca BUGIOTTI, <\/strong>Ma\u00eetresse de conf\u00e9rences, CentraleSup\u00e9lec, FRANCE – Rapporteur<\/p>\n\n\n\n M. Zoltan MIKLOS<\/strong>, Professeur, Universit\u00e9 de Rennes, FRANCE – Rapporteur<\/p>\n\n\n\n Mme Sylvie CALABRETTO<\/strong>, Professeure, INSA Lyon, FRANCE – Examinateur<\/p>\n\n\n\n M. Raphael TRONCY<\/strong>, Ma\u00eetre de conf\u00e9rences, Eurecom, FRANCE – Examinateur Mme Amel BOUZEGHOUB<\/strong>, Professeure, T\u00e9l\u00e9com SudParis, FRANCE – Directeur de these R\u00e9sum\u00e9 :<\/strong><\/p>\n\n\n\n Dans cette th\u00e8se, nous proposons de nouvelles m\u00e9thodes de recommandation de candidats \u00e0 l\u2019embauche, exploitant les capacit\u00e9s s\u00e9mantiques des graphes h\u00e9t\u00e9rog\u00e8nes. Tout d\u2019abord, nous construisons un graphe h\u00e9t\u00e9rog\u00e8ne \u00e0 partir de donn\u00e9es issues d\u2019un syst\u00e8me de suivi des candidats (Applicant Tracking System, ATS), devenu la norme industrielle pour g\u00e9rer les processus de recrutement. Nous enrichissons ensuite la s\u00e9mantique de ce graphe \u00e0 l\u2019aide de deux bases de connaissances externes : ESCO (la classification europ\u00e9enne des comp\u00e9tences, des connaissances et des professions) et Wikidata. Nous nommons le graphe r\u00e9sultant Job Tracking History (JTH). Ensuite, nous d\u00e9veloppons un premier syst\u00e8me de recommandation bas\u00e9 sur les graphes pour r\u00e9pondre au probl\u00e8me du d\u00e9marrage \u00e0 froid dans les recommandations candidat-postes. Les interactions utilisateur-objet dans les donn\u00e9es de recrutement sont par nature \u00e9parses, car les individus ne changent pas fr\u00e9quemment d\u2019emploi. Pour surmonter cette difficult\u00e9, nous concevons une m\u00e9thode de r\u00e9seau de neurones graphiques qui inf\u00e8re des recommandations \u00e0 partir du JTH, compensant le manque d\u2019interactions candidats-postes par des relations s\u00e9mantiques. Nous appelons cette premi\u00e8re m\u00e9thode RecruiterGCN. Enfin, sur la base de nos premiers r\u00e9sultats, nous constatons que l\u2019int\u00e9gration de la temporalit\u00e9 est essentielle \u00e0 notre approche, les candidatures et les offres d\u2019emploi ayant une disponibilit\u00e9 limit\u00e9e dans le temps. Pour relever ce d\u00e9fi, nous proposons TIMBRE (Temporal Integrated Model for Better REcommendations), une approche introduisant une nouvelle repr\u00e9sentation du temps dans un graphe h\u00e9t\u00e9rog\u00e8ne, gr\u00e2ce \u00e0 l\u2019ajout de nouveaux types de n\u0153uds et \u00e0 une nouvelle m\u00e9thode d\u2019\u00e9chantillonnage de graphes. Nous d\u00e9montrons que les m\u00e9triques traditionnelles utilis\u00e9es dans les syst\u00e8mes de recommandation temporels bas\u00e9s sur les graphes ne sont pas corr\u00e9l\u00e9es avec celles \u00e9valuant les capacit\u00e9s de classement. Sur ces derni\u00e8res, TIMBRE surpasse largement les grands mod\u00e8les de langage dans la recommandation candidat-postes.<\/p>\n\n\n\n In this thesis, we provide new methods for job candidate recommendation, harnessing the semantic capabilities of heterogeneous graphs. First, we build a heterogeneous graph from data extracted from an Applicant Tracking System (ATS), which has become the industry standard for handling the recruitment process. We then enrich the graph semantics with two external knowledge bases: ESCO, the European Classification for Skills and Knowledge, and Wikidata. We named the resulting graph Job Tracking History (JTH). Secondly, we built a first graph-based recommender system to address the cold start problem in job-candidate recommendations. User-item interaction in recruitment data is sparse in nature, as people do not frequently change jobs. To tackle this problem, we built a graph neural network methods that infer recom- mendations from JTH, compensating for the lack of job-candidates interaction with semantic relationships. We named this first method RecruiterGCN. Lastly, based on our first result, we found that including temporality is crucial to our approach, as candidate applications and job positions have limited avail- ability. To overcome this challenge, we propose TIMBRE (Temporal Integrated Model for Better REcommendations). This approach introduces a new representa- tion of time in a heterogeneous graph through the addition of new node types and a novel graph sampling method. We demonstrate that traditional metrics employed in state-of-the-art graph-based temporal recommender systems are uncorrelated with metrics that assess ranking capabilities. On these metrics, TIMBRE vastly outperforms even large language models at job-candidate recommendation.<\/p>\n","protected":false},"excerpt":{"rendered":" L’Ecole doctorale : Ecole Doctorale de l’Institut Polytechnique de Paris et le Laboratoire de recherche SAMOVAR – Services r\u00e9partis, Architectures, Mod\u00e9lisation, Validation, Administration des R\u00e9seaux pr\u00e9sentent l\u2019AVIS DE SOUTENANCE de Monsieur Eric BEHAR Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en : […]<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ocean_post_layout":"","ocean_both_sidebars_style":"","ocean_both_sidebars_content_width":0,"ocean_both_sidebars_sidebars_width":0,"ocean_sidebar":"","ocean_second_sidebar":"","ocean_disable_margins":"enable","ocean_add_body_class":"","ocean_shortcode_before_top_bar":"","ocean_shortcode_after_top_bar":"","ocean_shortcode_before_header":"","ocean_shortcode_after_header":"","ocean_has_shortcode":"","ocean_shortcode_after_title":"","ocean_shortcode_before_footer_widgets":"","ocean_shortcode_after_footer_widgets":"","ocean_shortcode_before_footer_bottom":"","ocean_shortcode_after_footer_bottom":"","ocean_display_top_bar":"default","ocean_display_header":"default","ocean_header_style":"","ocean_center_header_left_menu":"","ocean_custom_header_template":"","ocean_custom_logo":0,"ocean_custom_retina_logo":0,"ocean_custom_logo_max_width":0,"ocean_custom_logo_tablet_max_width":0,"ocean_custom_logo_mobile_max_width":0,"ocean_custom_logo_max_height":0,"ocean_custom_logo_tablet_max_height":0,"ocean_custom_logo_mobile_max_height":0,"ocean_header_custom_menu":"","ocean_menu_typo_font_family":"","ocean_menu_typo_font_subset":"","ocean_menu_typo_font_size":0,"ocean_menu_typo_font_size_tablet":0,"ocean_menu_typo_font_size_mobile":0,"ocean_menu_typo_font_size_unit":"px","ocean_menu_typo_font_weight":"","ocean_menu_typo_font_weight_tablet":"","ocean_menu_typo_font_weight_mobile":"","ocean_menu_typo_transform":"","ocean_menu_typo_transform_tablet":"","ocean_menu_typo_transform_mobile":"","ocean_menu_typo_line_height":0,"ocean_menu_typo_line_height_tablet":0,"ocean_menu_typo_line_height_mobile":0,"ocean_menu_typo_line_height_unit":"","ocean_menu_typo_spacing":0,"ocean_menu_typo_spacing_tablet":0,"ocean_menu_typo_spacing_mobile":0,"ocean_menu_typo_spacing_unit":"","ocean_menu_link_color":"","ocean_menu_link_color_hover":"","ocean_menu_link_color_active":"","ocean_menu_link_background":"","ocean_menu_link_hover_background":"","ocean_menu_link_active_background":"","ocean_menu_social_links_bg":"","ocean_menu_social_hover_links_bg":"","ocean_menu_social_links_color":"","ocean_menu_social_hover_links_color":"","ocean_disable_title":"default","ocean_disable_heading":"default","ocean_post_title":"","ocean_post_subheading":"","ocean_post_title_style":"","ocean_post_title_background_color":"","ocean_post_title_background":0,"ocean_post_title_bg_image_position":"","ocean_post_title_bg_image_attachment":"","ocean_post_title_bg_image_repeat":"","ocean_post_title_bg_image_size":"","ocean_post_title_height":0,"ocean_post_title_bg_overlay":0.5,"ocean_post_title_bg_overlay_color":"","ocean_disable_breadcrumbs":"default","ocean_breadcrumbs_color":"","ocean_breadcrumbs_separator_color":"","ocean_breadcrumbs_links_color":"","ocean_breadcrumbs_links_hover_color":"","ocean_display_footer_widgets":"default","ocean_display_footer_bottom":"default","ocean_custom_footer_template":"","ocean_post_oembed":"","ocean_post_self_hosted_media":"","ocean_post_video_embed":"","ocean_link_format":"","ocean_link_format_target":"self","ocean_quote_format":"","ocean_quote_format_link":"post","ocean_gallery_link_images":"on","ocean_gallery_id":[],"footnotes":""},"categories":[286,549],"tags":[],"class_list":["post-7011","post","type-post","status-publish","format-standard","hentry","category-fractualites-ennews-fr","category-seminaire-acmes","entry"],"_links":{"self":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/7011","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/comments?post=7011"}],"version-history":[{"count":2,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/7011\/revisions"}],"predecessor-version":[{"id":7013,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/7011\/revisions\/7013"}],"wp:attachment":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media?parent=7011"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/categories?post=7011"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/tags?post=7011"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
M. Julien ROMERO<\/strong>, Ma\u00eetre de conf\u00e9rences, T\u00e9l\u00e9com SudParis, FRANCE – Co-encadrant de these<\/p>\n\n\n\n
<\/p>\n\n\n\n\u00ab Syst\u00e8mes de Recommandation Bas\u00e9s sur des Graphes Temporels pour le Recrutement \u00bb<\/h2>\n\n\n\n
pr\u00e9sent\u00e9 par Monsieur Eric BEHAR<\/h2>\n\n\n\n
Abstract :<\/strong><\/p>\n\n\n\n