Aspect-based sentiment analysis via multitask learning for online reviews

Article Computer Science, Artificial Intelligence

TADC: A Topic-Aware Dynamic Convolutional Neural Network for Aspect Extraction

Zusheng Zhang et al.

Summary: Aspect extraction is a key task in fine-grained sentiment analysis. This study proposes a CNN-based model that uses dynamic filters generated from aspect information to effectively identify and extract aspects.

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2023)

Add to Collection

Article Computer Science, Artificial Intelligence

Meta-Based Self-Training and Re-Weighting for Aspect-Based Sentiment Analysis

Kai He et al.

Summary: The study proposes a meta-based self-training method for aspect-based sentiment analysis (ABSA). By generating pseudo-labels and controlling convergence rates, the method improves model performance and accuracy in fine-grained sentiment analysis.

IEEE TRANSACTIONS ON AFFECTIVE COMPUTING (2023)

Add to Collection

Article Computer Science, Artificial Intelligence

The Biases of Pre-Trained Language Models: An Empirical Study on Prompt-Based Sentiment Analysis and Emotion Detection

Rui Mao et al.

Summary: With the breakthrough of large-scale pre-trained language model (PLM) technology, prompt-based classification tasks, such as sentiment analysis and emotion detection, have gained increasing attention. This study conducts a systematic empirical study on prompt-based sentiment analysis and emotion detection to investigate the biases of PLMs in affective computing.

IEEE TRANSACTIONS ON AFFECTIVE COMPUTING (2023)

Add to Collection

Article Computer Science, Artificial Intelligence

Aspect-based sentiment analysis via affective knowledge enhanced graph convolutional networks

Bin Liang et al.

Summary: This paper proposes a graph convolutional network model Sentic GCN based on SenticNet to enhance the affective dependencies of sentences for aspect-based sentiment analysis. By integrating emotional knowledge from SenticNet, the model effectively handles contextual affective information in sentences, improving the effectiveness of sentiment polarity detection towards specific aspects.

KNOWLEDGE-BASED SYSTEMS (2022)

Add to Collection

Article Computer Science, Information Systems

Relation construction for aspect-level sentiment classification

Jiandian Zeng et al.

Summary: Aspect-level sentiment classification aims to classify fine-grained sentiment polarities of different aspects in a sentence. Existing methods overlook the aspect relations, while this paper proposes a new multi-task learning network that extracts aspect relations to improve the classification results, and conducts experiments on multiple datasets to demonstrate its effectiveness.

INFORMATION SCIENCES (2022)

Add to Collection

Article Computer Science, Information Systems

Convolutional attention neural network over graph structures for improving the performance of aspect-level sentiment analysis

Huyen Trang Phan et al.

Summary: This study proposes a new method called the CANN-SSCG model to address the limitations of graph convolutional network (GCN)-based aspect-level sentiment analysis methods. The method constructs three different heterogeneous graphs and combines them into a general heterogeneous graph. Finally, a convolutional neural network algorithm is used for aspect-level sentiment analysis, and promising results are achieved.

INFORMATION SCIENCES (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

An enhanced guided LDA model augmented with BERT based semantic strength for aspect term extraction in sentiment analysis

Manju Venugopalan et al.

Summary: Aspect level sentiment analysis is a fine-grained task that extracts aspects and their sentiment polarity from opinionated text. This research proposes an unsupervised model that uses minimal aspect seed words to guide the extraction process and enhance the performance. The model incorporates guided inputs, multiple pruning strategies, and semantic filters to improve performance. Evaluation results show competitive and appreciable performance on restaurant domain datasets.

KNOWLEDGE-BASED SYSTEMS (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

Aspect-based sentiment analysis with attention-assisted graph and variational sentence representation

Shi Feng et al.

Summary: Aspect-based sentiment analysis is a fine-grained task that detects the sentiment polarities of aspect words in a sentence. Current ABSA models primarily use graph-based methods but may rely excessively on the quality of dependency trees and overlook global sentence information. To address these issues, we propose a new ABSA model AG-VSR, which utilizes A2GR and VSR for final classification.

KNOWLEDGE-BASED SYSTEMS (2022)

Add to Collection

Article Computer Science, Information Systems

Aggregated graph convolutional networks for aspect-based sentiment classification

Meng Zhao et al.

Summary: This paper proposes an aggregated graph convolutional network (AGCN) to enhance the representation ability of target nodes in aspect-based sentiment analysis. The AGCN updates the node representation iteratively using aggregator functions, and uses subdependency and attention mechanism to extract and capture sentiment dependencies between node feature information. Experimental results show that AGCN is effective compared to other GCN-based methods.

INFORMATION SCIENCES (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

Attention-based aspect sentiment classification using enhanced learning through cnn-Bilstm networks

Eniafe Festus Ayetiran

KNOWLEDGE-BASED SYSTEMS (2022)

Add to Collection

Proceedings Paper Computer Science, Information Systems

PERD: Personalized Emoji Recommendation with Dynamic User Preference

Xuanzhi Zheng et al.

Summary: This paper proposes a personalized emoji recommendation method, which helps users find appropriate emojis from a large number of candidates based on user historical information and semantic similarity. Experimental results validate the superiority of this method.

PROCEEDINGS OF THE 45TH INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL (SIGIR '22) (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

A dependency syntactic knowledge augmented interactive architecture for end-to-end aspect-based sentiment analysis

Yunlong Liang et al.

Summary: This paper proposes an end-to-end aspect-based sentiment analysis model that utilizes dependency syntactic knowledge and multi-task learning, outperforming existing methods significantly by leveraging Dependency Relation Embedded Graph Convolutional Network and message-passing mechanism.

NEUROCOMPUTING (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Enhanced aspect-based sentiment analysis models with progressive self-supervised attention learning

Jinsong Su et al.

Summary: In this study, a progressive self-supervised attention learning approach is proposed to enhance the performance of aspect-based sentiment analysis (ABSA) models by continuously learning useful attention supervision information through iterative processes. Experimental results demonstrate that this method can effectively improve the attention mechanism of ABSA models and enhance their performance.

ARTIFICIAL INTELLIGENCE (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

CAPER: Context-Aware Personalized Emoji Recommendation

Guoshuai Zhao et al.

Summary: In this paper, a Context-Aware Personalized Emoji Recommendation (CAPER) model is proposed, which fuses contextual and personal information to improve recommendation accuracy. Experimental results show better performance of the CAPER model compared to existing methods, demonstrating the effectiveness of considering contextual and personal factors.

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Positionless aspect based sentiment analysis using attention mechanism

Rohan Kumar Yadav et al.

Summary: This paper simplifies preprocessing by incorporating polarity lexicon replacement and masking techniques to eliminate positional embedding. The experimental results demonstrate a significant improvement in accuracy for publicly available ABSA datasets using this approach.

KNOWLEDGE-BASED SYSTEMS (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Knowledge-enabled BERT for aspect-based sentiment analysis

Anping Zhao et al.

Summary: The knowledge-enabled language representation model BERT proposed in this work enhances aspect-based sentiment analysis by injecting domain knowledge and leveraging an external sentiment knowledge graph, resulting in more accurate and explainable results.

KNOWLEDGE-BASED SYSTEMS (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Dynamic evolution of multi-graph based collaborative filtering for recommendation systems

Hao Tang et al.

Summary: The recommendation system is an important technology in the era of Big Data. Current methods have integrated side information to alleviate the sparsity problem, but not all side information can be obtained with high quality. By proposing the DMGCF model and dynamically evolving multi-graph collaborative filtering, the approach successfully mines and reuses side information, as demonstrated by experimental results.

KNOWLEDGE-BASED SYSTEMS (2021)

Add to Collection

Article Acoustics

Investigating Typed Syntactic Dependencies for Targeted Sentiment Classification Using Graph Attention Neural Network

Xuefeng Bai et al.

Summary: Researchers propose a novel approach, a relational graph attention network that integrates typed syntactic dependency information, to improve targeted sentiment classification performance, which achieves good results. This method effectively utilizes label information and outperforms existing state-of-the-art syntax-based approaches.

IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING (2021)

Add to Collection

Article Computer Science, Information Systems