RECOMMENDATION SYSTEM BASED ON JOINT APPLICATION OF TEXT BIDIRECTIONAL ENCODER AND KNOWLEDGE GRAPH
Программирование / Programming and Computer Software · 2025 · (6) · 17
DOI: 10.7868/s3034584725060021 ↗Kurz gefasst
Fachartikel in der Zeitschrift Программирование / Programming and Computer Software, 2025. Verfasst von T.A. Dmitrieva und 3 weiteren Autor:innen.
Abstract (Originalsprache)
In this paper we develop recommendation system based on reasoning neural network and knowledge graph. Some aspects of domain knowledge graph completion and entity and link embeddings creation are considered. Composed graph is used for personal recommendations reasoning. Neural network generates personal recommendations depending on historic user activity patterns and semantic connections between the simulated entities in the graph. Our work is based on some existing researches. The scientific novelty lies in implementation of a transformer based language model to extract additional entities from the text description of domain objects and embed them in the graph structure. Quality recommendation metrics are statistically boosted in comparison with the some existed approaches in the field of path-based reasoning on knowledge graphs. The research may be used to ameliorate recommendation services in online retail.
Zitieren
APA
Dmitrieva, T., Kashirin, I., Ovechkin, G., & Uspenskiy, D. (2025). RECOMMENDATION SYSTEM BASED ON JOINT APPLICATION OF TEXT BIDIRECTIONAL ENCODER AND KNOWLEDGE GRAPH. Программирование / Programming and Computer Software, 17. https://doi.org/10.7868/s3034584725060021
Vancouver
Dmitrieva T, Kashirin I, Ovechkin G, Uspenskiy D. RECOMMENDATION SYSTEM BASED ON JOINT APPLICATION OF TEXT BIDIRECTIONAL ENCODER AND KNOWLEDGE GRAPH. Программирование / Programming and Computer Software. 2025:17. doi:10.7868/s3034584725060021
BibTeX
@article{Dmitrieva2025,
author = {Dmitrieva, T.A. and Kashirin, I.Y. and Ovechkin, G.V. and Uspenskiy, D.I.},
title = {RECOMMENDATION SYSTEM BASED ON JOINT APPLICATION OF TEXT BIDIRECTIONAL ENCODER AND KNOWLEDGE GRAPH},
journal = {Программирование / Programming and Computer Software},
year = {2025},
number = {6},
pages = {17},
doi = {10.7868/s3034584725060021}
}
EUVIMED