A Multi-Layer Network for Aspect-Based Cross-Lingual Sentiment Classification

Abstract

In the recent era, the advancement of communication technologies provides a valuable interaction source between people of different regions. Nowadays, many organizations adopt the latest approaches, i.e., sentiment analysis and aspect-oriented sentiment classification, to evaluate user reviews to improve the quality of their products. The processing of multi-lingual user reviews is a key challenge in Natural Language Processing (NLP). This paper proposes a multi-layer network with divided attention to perform aspectbased sentiment classification for cross-lingual data. It extracts the Part-of-Speech (POS) tagging information of the given reviews, preprocesses them, and converts them into tokens. Furthermore, bi-lingual dictionaries are leveraged to map the converted tokens from one language to another. Given the preprocessed and mapped reviews, vectors are generated by leveraging the multi-lingual BERT and passed to the proposed deep learning classifier. The 10351 restaurant reviews from SemEval-2016 Task 5 dataset are exploited for the prediction of aspect-based sentiment. The results of cross-lingual validation suggest that the proposed approach significantly outperforms the state-of-the-art approaches and improves the precision, recall, and F1 by more than 23%, 20%, and 22%, respectively.

Removing Non-Stationary Knowledge From Pre-Trained Language Models for Entity-Level Sentiment Classification in Finance

“…Further in this work, we use the pipeline in figure 1 over KorF inASC to offer a comprehensive investigation into whether datasets of different domains, languages, and tasks can be leveraged to improve classification performance. Also, referred to as intermediate transfer learning, multiple researches has already reported that cross-lingual transfer from English improves performance in general domain tasks (Pruksachatkun et al 2020;Sattar et al 2021). We find that the nature of financespecific ASC problems differs significantly from that of general domain tasks requiring additional measures to be properly transferred.…”

“…The majority of previous research on cross-lingual transfer for general-domain ASC involves parallel annotated data for both the source and target languages; ironically such data is equally difficult to acquire (Balahur and Turchi 2012;Lambert 2015). Only very recently unsupervised cross-lingual transfer was proposed, and surprisingly it has reported 20 to 23% improvement on state-of-the-art approaches (Sattar et al 2021). However, whether such behavior is equally applicable to the finance domain has not yet been explored.…”

Implicit aspect based sentiment analysis for restaurant review using LDA topic modeling and ensemble approach

“…These reviews serve as crucial sources of information for both customers and decision-makers in the food and beverage industry. Sattar et al [33] highlighted that advancements in communication technology have facilitated valuable connections among individuals in different regions. In response, many companies are adopting modern approaches such as sentiment analysis and profile-based sentiment classification to assess user reviews and enhance product quality.…”

Zero-Shot Emotion Detection for Semi-Supervised Sentiment Analysis Using Sentence Transformers and Ensemble Learning

“…According to Medhat et al [39], we describe the taxonomy of sentiment analysis techniques and divide it into two main paradigms: rule-/lexicon-based and machine learning. Lexicon-based methods [40] rely on the assumption that the overall sentiment depends on the words that explicitly express these sentiments. Words (adjectives, adverbs, sometimes verbs, and nouns) that define different sentiments are searched in the text and counted: the overall sentiment of the text depends on the majority.…”