Centre de documentation
scientifique

Graph neural network based model for multi-behavior session-based recommendation / Bo Yu in Geoinformatica, vol 26 n° 2 (April 2022)  

Public [article]  inGeoinformatica > vol 26 n° 2 (April 2022) . - pp 429 - 447 Titre : Graph neural network based model for multi-behavior session-based recommendation Type de document : Article/Communication Auteurs : Bo Yu, Auteur ; Ruoqian Zhang, Auteur ; Wei Chen, Auteur ; Junhua Fang, Auteur Année de publication : 2022 Article en page(s) : pp 429 - 447 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Intelligence artificielle [Termes IGN] attention (apprentissage automatique) [Termes IGN] comportement [Termes IGN] consommation [Termes IGN] modèle de simulation [Termes IGN] réseau neuronal de graphes [Termes IGN] réseau sémantique [Termes IGN] service fondé sur la position Mots-clés libres : session Résumé : (auteur) Multi-behavior session-based recommendation aims to predict the next item, such as a location-based service (LBS) or a product, to be interacted by a specific behavior type (e.g., buy or click) in a session involving multiple types of behaviors. State-of-the-art methods generally model multi-behavior dependencies in item-level, but ignore the potential of discovering useful patterns of multi-behavior transition through feature-level representation learning. Besides, sequential and non-sequential patterns should be properly fused in session modeling to capture dynamic interests within the session. To this end, this paper proposes a Graph Neural Network based Hybrid Model GNNH, which enables feature-level deeper representations of multi-behavior interaction sequences for session-based recommendation. Specifically, we first construct multi-relational item graph (MRIG) and feature graph (MRFG) based on session sequences. On top of the MRIG and MRFG, our model takes advantage of GNN to capture item and feature representations, such that global item-to-item and feature-to-feature relations are fully preserved. Afterwards, each multi-behavior session is modeled by a seamless fusion of interacted item and feature representations, where self-attention and mean-pooling are used to obtain sequential and non-sequential patterns simultaneously. Experiments on two real datasets show that the GNNH model significantly outperforms the state-of-the-art methods. Numéro de notice : A2022-326 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE/INFORMATIQUE Nature : Article DOI : 10.1007/s10707-021-00439-w Date de publication en ligne : 29/05/2021 En ligne : https://doi.org/10.1007/s10707-021-00439-w Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100489 [article]

MTLM: a multi-task learning model for travel time estimation / Saijun Xu in Geoinformatica, vol 26 n° 2 (April 2022)  

Public [article]  inGeoinformatica > vol 26 n° 2 (April 2022) . - pp 379 - 395 Titre : MTLM: a multi-task learning model for travel time estimation Type de document : Article/Communication Auteurs : Saijun Xu, Auteur ; Ruoqian Zhang, Auteur ; Wanjun Cheng, Auteur ; Jiajie Xu, Auteur Année de publication : 2022 Article en page(s) : pp 379 - 395 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Géomatique [Termes IGN] analyse coût-avantage [Termes IGN] apprentissage automatique [Termes IGN] données d'entrainement (apprentissage automatique) [Termes IGN] durée de trajet [Termes IGN] modèle de simulation [Termes IGN] transport collectif [Termes IGN] transport intermodal Résumé : (auteur) Travel time estimation (TTE) is an important research topic in many geographic applications for smart city research. However, existing approaches either ignore the impact of transportation modes, or assume the mode information is known for each training trajectory and the query input. In this paper, we propose a multi-task learning model for travel time estimation called MTLM, which recommends the appropriate transportation mode for users, and then estimates the related travel time of the path. It integrates transportation-mode recommendation task and travel time estimation task to capture the mutual influence between them for more accurate TTE results. Furthermore, it captures spatio-temporal dependencies and transportation mode effect by learning effective representations for TTE. It combines the transportation-mode recommendation loss and TTE loss for training. Extensive experiments on real datasets demonstrate the effectiveness of our proposed methods. Numéro de notice : A2022-325 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE Nature : Article DOI : https://doi.org/10.1007/s10707-020-00422-x Date de publication en ligne : 15/08/2020 En ligne : https://doi.org/10.1007/s10707-020-00422-x Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100488 [article]