Deep learning methods for LSTM-based personalized search: a comparative analysis

doi:10.1007/s13042-024-02418-7

Return

Deep learning methods for LSTM-based personalized search: a comparative analysis

2024-10-25

PRE

Sara Abri *

Rayan Abri

DOI：10.1007/s13042-024-02418-7

Original

Original request for help

Save

Abstract

En 中文

Deep learning techniques have become effective tools for addressing the difficulties associated with personalized search. Long Short-Term Memory (LSTM) models, one of the significant extended models of Recurrent Neural Networks (RNNs), are now widely used to re-rank search results. Thanks to LSTM, the neural network can handle large time-series data and have a memory function. By thoroughly comparing enhanced deep learning-based and especially LSTM-based models, we analyze the literature in the field of personalized search to rank the results in this survey. By focusing on the existence of popular methods, we evaluate the performance of the methods on two datasets in various settings to obtain the best results. The goal is to evaluate the effectiveness of each method and advance the state-of-the-art for deep learning LSTM-based personalized search. We hope to help researchers in choosing the most efficient deep learning algorithm for LSTM-based re-ranking in personalized search using this comparison.

Keywords:

Deep learning methods

LSTM model

Personalized web search

User profiling