Session-Based Recommendation by combining Probabilistic Models and LSTM
Introduction
Figure 1: Example Session and Purchase Data of ACM RecSys Challenge 2022.
Figure 2: The schema of the proposed hybrid system architecture.
We present the approach, we used as team "DataLab HMU.GR", for the ACM RecSys Challenge 2022 [1]. The challenge aims to predict the item that was purchased for a given sequence of item views (session). The full dataset, provided by Dressipi, consists of 1.1 million online retail sessions.
Methodology
Our proposed method, that solves the Session-Based Recommendation problem, relies on an efficient deterministic system based on a weighted combination of Probabilistic models and an LSTM neural network.
Probabilistic models learn the transition probabilities between item-item interactions of each session, that are used to predict the purchase probability of an item in a new session.
The LSTM neural network takes as input the context representation of the items in a session and a candidate item and predicts the purchase probability of the candidate item.
Experiments - Downloads
The code of the proposed method is available on github: https://github.com/cpanag79/recsys-Challenge-2022
Related Publications
[1] C. Panagiotakis and H. Papadakis, Session-Based Recommendation by combining Probabilistic Models and LSTM, ACM RecSys Challenge 2022.
[2] C. Panagiotakis, H. Papadakis, A. Papagrigoriou and P. Fragopoulou, Improving Recommender Systems via a Dual Training Error based Correction Approach, Expert Systems with Applications, 2021.
[3] H. Papadakis, C. Panagiotakis and P. Fragopoulou, SCoR: A Synthetic Coordinate based Recommender System, Expert Systems with Applications, vol. 79, pp.8-19, 2017.
[4] C. Panagiotakis, H. Papadakis, A. Papagrigoriou, and P. Fragopoulou, DTEC: Dual Training Error based Correction approach for Recommender Systems, Software Impacts, 2021.