Cross-Domain Sentiment Analysis using Transfer Learning and Domain Tokens

DOI

10.2991/978-94-6239-674-6_26How to use a DOI?

Keywords

Transfer Learning; Domain Tokens; Few-Shot Learning; Amazon Review Dataset; Contextual Embeddings

Abstract

Most existing models of sentiment analysis degrade in performance significantly when used out of domains they are originally trained on, which is essentially because vocabularies, semantics, and contextual patterns all vary across domains. In this paper, we will introduce a simple, lightweight, and flexible transfer learning approach to improve assumption classification tasks on hidden spaces with only a few available labeled data. Our model proposes that domain tokens are introduced at the input level. That is, special tokens like [DVD], [BOOKS] will indicate different domains. These domain tokens explicitly cue the model to learn domain-aware representations without modification of its architecture. Second, we integrate a few-shot learning process where the model is fine-tuned with only a few labeled cases of the target domain. Our approach has been evaluated on the widely used dataset of Amazon Review, which includes four domains: Electronics, DVD, Books, and Kitchen. The results reveal that our model with these domain tokens consistently preserves the top rank position and reaches F1 values of up to 0.66 in both zero-shot and few-shot learning. The above points and discoveries reveal the capability and potential of space tokens in performing opinion analysis in a domain.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Dhirendra Yadav
AU  - Mohit Singh
AU  - Ramesh Chundi
AU  - S. Senthil
PY  - 2026
DA  - 2026/05/28
TI  - Cross-Domain Sentiment Analysis using Transfer Learning and Domain Tokens
BT  - Proceedings of the International Conference on Sustainable Computing and Artificial Intelligence (ICSCAI 2025)
PB  - Atlantis Press
SP  - 299
EP  - 312
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6239-674-6_26
DO  - 10.2991/978-94-6239-674-6_26
ID  - Yadav2026
ER  -