Recommendation Algorithm of Personalized Environmental Art Design Based on Deep Learning

With the deep integration of digital technology and environmental art design, the personalized needs and dynamic preferences contained in user behavior data put forward higher requirements for recommendation systems. In view of the shortcomings of traditional recommendation methods in temporal modeling and design element association analysis, this paper proposes a personalized environment art design recommendation algorithm based on Bidirectional Long-Short-Term Memory (BiLSTM). By constructing a bidirectional temporal modeling framework, integrating the positive cumulative effect and reverse explanatory association of user behavior sequence, and combining with in-depth learning feature extraction and dynamic decision-making mechanism, the precise capture of user preference transfer law and nonlinear interaction relationship of design elements is realized. Experiments show that the model is significantly better than comparison model in accuracy (85.2 %), recall (78.1 %) and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$F 1$</tex> value (81.5 %), and has a real-time response capability of 0.5 seconds, which provides an effective technical path for the intelligent transformation of environmental art design.