Article Details

Abstract

To analyze the impact of Internet trolls in the smartphone product field on the authenticity and credibility of reviews, this study proposes a hybrid model for identifying smartphone product online troll comments based on multi-dimensional features and Random Forest. The multi-dimensional feature correlation degree is introduced. This model uses the Random Forest algorithm to integrate multi-dimensional features and apply them to the detection of troll comments in smartphone product online reviews. Comparative experiments on the dataset show that the accuracy, F1-score, and precision of this model are 80.6%, 80.9%, and 78.7%, respectively. All performance indicators show competitive results. Moreover, the ablation experiment results further indicate that different features have varying impacts on the model. This study provides an effective technical means for the management of online reviews of smartphone products, aiming to present more authentic and reliable review information to consumers and to provide a reference for future related research.

Keywords

Internet troll; Internet troll identification; Machine Learning; multi-dimensional feature; Random Forest

References

1. [1] Wu Jing, Guo Caiping The impact and strategies of short video live streaming marketing on consumer purchasing behavior [J]. Journal of Shanxi Open University, 2025, 30 (2): 95-98
2. [2] Lei Xia Research on the Development, Evolution and Characteristics of Network Water Army [J]. People's Forum, 2024 (21): 102-105
3. [3] Zhou Mei, Chang Jianhua, Chen Sicheng, etc A model for aerosol type recognition based on naive Bayes classifier [J]. Acta Optica Sinica, 2022, 42 (18): 49-57
4. [4] Lv Ning, Luo Qian A Telecom Customer churn Prediction Model Integrating XGBoost and Logistic Regression Algorithm [J]. Modern Electronic Technology, 2025, 48 (11): 136-143
5. [5] Li Ying, Hou Xuerui, Liu Chenhui Analysis of Electric Vehicle Charging Behavior Based on Ordered Logistic Regression Model [J]. Science, Technology and Engineering, 2025, 25 (8): 3473-3479
6. [6] Zou Jie, Zhu Guosheng, Qi Xiaoyun, etc HTTPS encrypted traffic classification method based on C4.5 decision tree [J]. Computer Science, 2020, 47 (S1): 381-385
7. [7] Yang Chao, Qin Tingdong, Fan Bo, etc Research on Weibo Navy User Detection Based on Artificial Immune Hazard Theory [J]. Computer Science, 2018, 45 (11): 138-142159
8. [8] Wen Xiaohui Design and Implementation of Weibo Navy Recognition System Based on Graph Neural Network [D]. Jining: Qufu Normal University, 2021
9. [9] Wang Siyuan Weibo Navy Identification Based on CBOW Model [D]. Suzhou: Soochow University, 2022
10. [10] Ning Xinli, Sun Yuan Network Water Army Identification Based on Douban Short Reviews [J]. Statistics and Consulting, 2022 (3): 6-9
11. [11] Zhang Jiyang, Zhang Peng, Gong Siyu, etc Research on deep learning based network navy recognition technology [J]. Journal of Agricultural Library and Information Science, 2023, 35 (12): 49-59

Funding

Discipline Co construction Project of Philosophy and Social Sciences Planning in Guangdong Province in 2023 (GD23XXW13); 2025 Guangzhou University Teaching Reform Project Teaching Master Cultivation Project (2025JXMS009); 2024 School level Quality Engineering Teacher Teaching Innovation Team Project (XTD202402)

About the authors

Cao Huiru (1981-), female, Han ethnicity, from Weinan, Shaanxi, associate professor, master's student, research direction: big data, social networks.