MF-CNN-BILSTM: A Deep Learning-Based Sentiment Analysis Approach and Topic Modeling of Tweets Related to the Ukraine-Russia Conflict

Aslan, Serpil

doi:10.2139/ssrn.4218398

Cited by 4 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, the final 2-additive fuzzy measure on each feature and each pair of features can be obtained by solving the optimization model (19). Then, the fuzzy weights of all subsets of the feature set C are obtained using Eq.…”

Section: Phase 3: Fuzzy Weights Determinationmentioning

confidence: 99%

“…Then, partial realistic preferences for the feature interaction directions are considered, and these constraints are added to the optimization model. And the maximum separation model is constructed as (19). LINGO 18.0 software is used to solve the model to obtain optimal feature capacities as ψ(C j ) = {s −2 (0.004), s −1 (0.009), s 0 (0.019), s 1 (0.182), s 2 (0.786)} {s −2 (0.048), s −1 (0.044), s 0 (0.018), s 1 (0.246), s 2 (0.643)} A 10 {s −2 (0.011), s −1 (0.015), s 0 (0.013), s 1 (0.146), s 2 (0.815)} {s −2 (0.039), s −1 (0.042), s 0 (0.015), s 1 (0.166), s 2 (0.738)} A 11 {s −2 (0.007), s −1 (0.024), s 0 (0.05), s 1 (0.264), s 2 (0.655)} {s −2 (0.059), s −1 (0.086), s 0 (0.035), s 1 (0.217), s 2 (0.604)} A 12 {s −2 (0.013), s −1 (0.011), s 0 (0.018), s 1 (0.17), s 2 (0.787)} {s −2 (0.081), s −1 (0.071), s 0 (0.024), s 1 (0.197), s 2 (0.629)} A 13 {s −2 (0.008), s −1 (0.012), s 0 (0.019), s 1 (0.112), s 2 (0.849)} {s −2 (0.04), s −1 (0.032), s 0 (0.018), s 1 (0.13), s 2 (0.78)} A 14 {s −2 (0.006), s −1 (0.017), s 0 (0.029), s 1 (0.154), s 2 (0.794)} {s −2 (0.098), s −1 (0.275), s 0 (0), s 1 (0.137), s 2 (0.49)} A 15 {s −2 (0.014), s −1 (0.025), s 0 (0.021), s 1 (0.…”

Section: Phase 3: Fuzzy Weights Determinationmentioning

confidence: 99%

“…Liu et al [17] manually extended the medical review corpus using a lexicon-based SA approach in the problem of doctor selection. By contrast, deep learning models, such as BERT [18], LSTM [6], RoBERTa [9], and Bidirectional Long and Short-Term Memory (BiL-STM) [19], show strong language understanding and transfer learning, including high accuracy in SA. BERT-BiLSTM [20] is an improved algorithm that fuses Transformer Bidirectional Encoded Representation (BERT) and BiLSTM.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

RETRACTED: Online review–driven Product Recommendation Ranking Considering Interactions Among Features and Consumers’ Psychology

Liu,

Li,

Chin

2024

Preprint

View full text Add to dashboard Cite

In the digital economy, online reviews serve as valuable and real-time feedback for consumers, making them a crucial source of information for product selection. We propose a product recommendation ranking method based on sentiment analysis of online reviews to address information overload and support informed purchase decisions. The proposed model considers the interdependencies among product features and fully simulates the psychological state of consumers before and after purchasing products. First, online reviews of products are mined and preprocessed. Natural language processing techniques are employed to extract product features and perform aspect-based sentiment analysis. The results are transformed into an evaluation matrix of the probabilistic linguistic term sets (PLTSs). Then, we propose a 2-additive fuzzy measures recognition model based on multi-dimensional feature attention information to effectively identify redundant, independent, and complementary interactions among product features. Subsequently, the fuzzy weight and the proposed possibility comprehensive discrepancy measures are applied to the interactive multi-criteria decision-making based on CPT-TODIM (Cumulative Prospect Theory TODIM) method to rank the recommended priorities of alternative products. The feasibility of the proposed method is validated through a case study of smart sweeping robot products from Jingdong Mall, followed by the investigation of the impact of various psychological behaviours that influence purchase decisions on the recommendation ranking based on stochastic multi-criteria acceptability analysis (SMAA). Finally, a comparative analysis is conducted to demonstrate the advantages and effectiveness of the proposed method.

show abstract

Section: Phase 3: Fuzzy Weights Determinationmentioning

confidence: 99%

Section: Phase 3: Fuzzy Weights Determinationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

RETRACTED: Online review–driven Product Recommendation Ranking Considering Interactions Among Features and Consumers’ Psychology

Liu,

Li,

Chin

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…As for our aim, in the last year, other works proposed sentiment, emotion, and/or intention analysis over user-generated content concerning the Russia-Ukraine conflict in order to deepen the war perception [39], by also using ML-based strategies [40] or defining proper models, as for the MF-CNN-BiLSTM model defined by Aslan [41]. Most of them exploit Twitter API and/or previously published tweets' datasets.…”

Section: Related Workmentioning

confidence: 99%

Analyzing the worldwide perception of the Russia-Ukraine conflict through Twitter

Breve,

Caruccio,

Cirillo

et al. 2024

J Big Data

View full text Add to dashboard Cite

In this paper, we analyze the worldwide perception of the Russia-Ukraine conflict (RU conflict for short) on the Twitter platform. The study involved collecting over 17 million tweets written in 63 different languages and conducting a multi-language sentiment analysis, as well as an analysis of their geographical distribution and verification of their temporal relationship to daily events. Additionally, the study focused on analyzing the accounts producing pro-conflict tweets to evaluate the possible presence of bots. The results of the analysis showed that the war had a significant global impact on Twitter, with the volume of tweets increasing as the war’s threats materialized. There was a strong correlation between the succession of events, the volume of tweets, and the prevalence of a specific sentiment. Most tweets had a negative sentiment, while tweets with positive sentiment mainly contained support and hope for people directly involved in the conflict. Moreover, a bot detection analysis performed on the collected tweets revealed the presence of many accounts spreading tweets including pro-conflict hashtags that cannot be identified as real users. Overall, this study sheds light on the importance of social media in shaping public opinion during conflicts and highlights the need for reliable methods to detect bots.

show abstract

Analysis of Semantic Relationships in Ukrainian Text Content Based on Word2Vec and Machine Learning

Kryndach,

Vysotska,

Chyrun

et al. 2023

2023 IEEE 18th International Conference on Computer Science and Information Technologies (CSIT)

View full text Add to dashboard Cite

MF-CNN-BILSTM: A Deep Learning-Based Sentiment Analysis Approach and Topic Modeling of Tweets Related to the Ukraine-Russia Conflict

Cited by 4 publications

References 0 publications

RETRACTED: Online review–driven Product Recommendation Ranking Considering Interactions Among Features and Consumers’ Psychology

RETRACTED: Online review–driven Product Recommendation Ranking Considering Interactions Among Features and Consumers’ Psychology

Analyzing the worldwide perception of the Russia-Ukraine conflict through Twitter

Analysis of Semantic Relationships in Ukrainian Text Content Based on Word2Vec and Machine Learning

Contact Info

Product

Resources

About