The paper introduces a novel neural network framework called Neural Attentive Recommendation Machine (NARM) to address session-based recommendation problems, particularly in e-commerce scenarios where user profiles are not available. NARM aims to capture both the user's sequential behavior and the main purpose of the current session, which is often overlooked in previous models. The framework uses a hybrid encoder with an attention mechanism to model these aspects, combining them into a unified session representation. This representation is then used to compute recommendation scores for each candidate item through a bi-linear matching scheme. Extensive experiments on two benchmark datasets, YOOCHOOSE and DIGINETICA, demonstrate that NARM outperforms state-of-the-art baselines, especially on long sessions, highlighting its effectiveness in modeling both sequential behavior and the user's main purpose. The paper also discusses the impact of different session features and lengths, and visualizes the attention weights to illustrate the attention mechanism's role in capturing important items.The paper introduces a novel neural network framework called Neural Attentive Recommendation Machine (NARM) to address session-based recommendation problems, particularly in e-commerce scenarios where user profiles are not available. NARM aims to capture both the user's sequential behavior and the main purpose of the current session, which is often overlooked in previous models. The framework uses a hybrid encoder with an attention mechanism to model these aspects, combining them into a unified session representation. This representation is then used to compute recommendation scores for each candidate item through a bi-linear matching scheme. Extensive experiments on two benchmark datasets, YOOCHOOSE and DIGINETICA, demonstrate that NARM outperforms state-of-the-art baselines, especially on long sessions, highlighting its effectiveness in modeling both sequential behavior and the user's main purpose. The paper also discusses the impact of different session features and lengths, and visualizes the attention weights to illustrate the attention mechanism's role in capturing important items.