融合深度情感分析和评分矩阵的推荐模型

李淑芝; 余乐陶; 邓小鸿

doi:10.11999/JEIT200779

融合深度情感分析和评分矩阵的推荐模型

doi: 10.11999/JEIT200779

1.
江西理工大学信息工程学院赣州 341000
2.
江西理工大学应用科学学院赣州 341000

基金项目: 国家自然科学基金(61762046)，江西省教育厅科学技术研究项目(GJJ181505)

详细信息

作者简介:
李淑芝：女，1964年生，教授，硕士生导师，研究方向为软件工程和信息隐藏

余乐陶：女，1996年生，硕士生，研究方向为数据挖掘

邓小鸿：男，1982年生，副教授，硕士生导师，研究方向为网络信息安全

通讯作者:
余乐陶　jxby568923279@qq.com

中图分类号: TP391
计量
- 文章访问数: 1208
- HTML全文浏览量: 602
- PDF下载量: 153
- 被引次数: 0
出版历程
- 收稿日期: 2020-09-02
- 修回日期: 2021-04-09
- 网络出版日期: 2021-07-13
- 刊出日期: 2022-01-10

Recommendation Model Combining Deep Sentiment Analysis and Scoring Matrix

1.
School of Information Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, China
2.
College of Applied Science, Jiangxi University of Science and Technology, Ganzhou 341000, China

Funds: The National Natural Science Foundation of China (61762046), The Science and Technology Research Project of Education Department of Jiangxi Province (GJJ181505)

摘要

摘要: 目前，大多数推荐系统都具有评分数据稀疏性的问题，它会限制模型的有效性。而用户对于某件商品撰写的评论中隐含了很多信息，对评论文本进行情感分析并提取关键的因素来用于模型的学习，可以有效地缓解数据稀疏问题，但仅使用评论数据而忽略了评分数据的主要因素会影响推荐精度。对此，为了进一步提高推荐精度，该文提出一个评论文本和评分矩阵交互(RTRM)的深度模型，该模型能够提取评论文本和评分矩阵的深层次特征，并结合它们进行评分预测；其次，通过使用预训练的Electra模型得到每条评论的隐表达，并结合深度情感分析及注意力机制实现从上下文语义层面对评论文本的分析，解决了短文本的语义难以分析的问题；同时，在融合层模块中，用户(物品)评论和评分矩阵进行交互，最终预测出用户对商品的评分；最后，在6组数据集上，采用均方误差(MSE)进行性能对比实验，实验结果表明该文模型性能优于其他系统，且平均预测误差最大降低了12.821%，该模型适用于向用户推荐精确的物品。
- 推荐系统 /
- 矩阵分解 /
- 评论文本 /
- 评分数据 /
- 深度学习
Abstract: Most recommendation systems have a data sparsity problem, which limits the validity of the model that they use. However, the user’s comments on a commodity contain a lot of information. Emotional analysis of the comment text and the extraction of key factors for model learning can effectively alleviate the data sparsity problem, but only the use of comment data and ignore the main factors of the scoring data will affect the recommendation accuracy. To improve further the precision of recommendations, a deep model for the processing of Review Texts and Rating Matrices (RTRM) is proposed. The model extracts deep-level features and combines them to make rating predictions. Then, by using the pre-trained Electra model, the implicit expression of each comment is get, and combining with the deep emotion analysis and attention mechanism, the analysis of the comment text is realized from the context semantic level. It solves the problem that it is difficult to analyze the semantics of short text; User (item) reviews interact with a rating matrix to predict the user’s rating of a product in the fusion layer module. Finally, the Mean Square Error (MSE) is used to perform performance comparison experiments on 6 sets of data sets. Experimental results show that the performance of the proposed model outperforms significantly other systems on a variety of datasets, and the average prediction error is reduced by a maximum of 12.821%, the model is suitable for recommending accurate items to users.
- Recommendation system /
- Matrix decomposition /
- Review text /
- Scoring data /
- Deep learning

HTML全文

图 1 RTRM模型结构

下载: 全尺寸图片幻灯片

图 2 评论情感分析模型

下载: 全尺寸图片幻灯片

图 3 不同隐因子个数对模型性能的影响

下载: 全尺寸图片幻灯片

图 4 Dropout比率对模型性能的影响

下载: 全尺寸图片幻灯片

图 5 不同近邻个数对模型性能的影响

下载: 全尺寸图片幻灯片

图 6 余弦相似度和云模型对模型性能的影响

下载: 全尺寸图片幻灯片

表 1 数据集基本信息

	Arts and Crafts	Digital Music	Software	Kindle Store	Luxury Beauty	Video Games	平均数
用户	33423	19819	3864	167569	5228	36789	44448
物品	29800	12876	1293	76438	1578	25212	24532
评论数目	494485	169781	12805	2222983	34278	497577	571984

下载: 导出CSV

表 2 RTRM模型与其他模型的性能比较

数据集	SVD ++	HFT	PMF	Deep CoNN	NA RRE	RTRM	提升(%)	提升(%)	提升(%)
数据集	a	b	c	d	e	f	f vs a	f vs c	f vs e
Arts and Crafts	0.862	0.809	0.794	0.776	0.738	0.699	18.910	11.965	5.285
Digital Music	0.835	0.816	0.803	0.787	0.763	0.742	11.138	7.597	2.752
Software	0.956	0.961	0.916	0.902	0.893	0.882	7.741	3.712	1.232
Kindle Store	1.371	1.313	1.105	1.084	0.993	0.972	29.103	12.036	2.115
Luxury Beauty	0.941	0.938	0.929	0.912	0.901	0.887	5.739	4.521	1.554
Video Games	0.675	0.698	0.681	0.662	0.655	0.646	4.296	5.147	1.374
AVG	0.940	0.923	0.871	0.854	0.824	0.805	12.821	7.496	2.385

下载: 导出CSV

参考文献(17)

[1]	BOBADILLA J, ORTEGA F, HERNANDO A, et al. Recommender systems survey[J]. Knowledge-Based Systems, 2013, 46: 109–132. doi: 10.1016/j.knosys.2013.03.012
[2]	黄立威, 江碧涛, 吕守业, 等. 基于深度学习的推荐系统研究综述[J]. 计算机学报, 2018, 41(7): 1619–1647. doi: 10.11897/SP.J.1016.2018.01619 HUANG Liwei, JIANG Bitao, LÜ Shouye, et al. Survey on deep based recommender systems[J]. Chinese Journal of Computers, 2018, 41(7): 1619–1647. doi: 10.11897/SP.J.1016.2018.01619
[3]	KOREN Y, BELL R, and VOLINSKY C. Matrix factorization techniques for recommender systems[J]. Computer, 2009, 42(8): 30–37. doi: 10.1109/MC.2009.263
[4]	张光卫, 李德毅, 李鹏, 等. 基于云模型的协同过滤推荐算法[J]. 软件学报, 2007, 18(10): 2403–2411. doi: 10.1360/jos182403 ZHANG Guangwei, LI Deyi, LI Peng, et al. A collaborative filtering recommendation algorithm based on cloud model[J]. Journal of Software, 2007, 18(10): 2403–2411. doi: 10.1360/jos182403
[5]	SUNDERMANN C V, DOMINGUES M A, SINOARA R A, et al. Using opinion mining in context-aware recommender systems: A systematic review[J]. Information, 2019, 10(2): 42. doi: 10.3390/info10020042
[6]	CHEN Li, CHEN Guanliang, and WANG Feng. Recommender systems based on user reviews: The state of the art[J]. User Modeling And User-adapted Interaction, 2015, 25(2): 99–154. doi: 10.1007/s11257-015-9155-5
[7]	MCAULEY J and LESKOVEC J. Hidden factors and hidden topics: Understanding rating dimensions with review text[C]. The 7th ACM Conference on Recommender Systems, Hong Kong, China, 2013: 165–172. doi: 10.1145/2507157.2507163.
[8]	REN Zhaochun, LIANG Shangsong, LI Piji, et al. Social collaborative viewpoint regression with explainable recommendations[C]. The Tenth ACM International Conference on Web Search and Data Mining, Cambridge, UK, 2017: 485–494. doi: 10.1145/3018661.3018686.
[9]	ZHENG Lei, NOROOZI V, and YU P S. Joint deep modeling of users and items using reviews for recommendation[C]. The Tenth ACM International Conference on Web Search and Data Mining, Cambridge, UK, 2017: 425–434. doi: 10.1145/3018661.3018665.
[10]	LI Piji, WANG Zihao, REN Zhaochun, et al. Neural rating regression with abstractive tips generation for recommendation[C]. The 40th International ACM SIGIR Conference on Research and Development in Information Retrieval, Tokyo, Janpan, 2017: 345–354. doi: 10.1145/3077136.3080822.
[11]	SEO S, HUANG Jing, YANG Hao, et al. Interpretable convolutional neural networks with dual local and global attention for review rating prediction[C]. The Eleventh ACM Conference on Recommender Systems, Como, Italy, 2017: 297–305. doi: 10.1145/3109859.3109890.
[12]	CHEN Chong, ZHANG Min, LIU Yiqun, et al. Neural attentional rating regression with review-level explanations[C]. The 2018 World Wide Web Conference, Lyon, France, 2018: 1583–1592. doi: 10.1145/3178876.3186070.
[13]	CHO K, VAN MERRIENBOER B, BAHDANAU D, et al. On the properties of neural machine translation: Encoder-decoder approaches[EB/OL]. https://arxiv.org/abs/1409.1259, 2014.
[14]	CLARK K, LUONG M T, LE Q V, et al. ELECTRA: Pre-training text encoders as discriminators rather than generators[C]. International Conference on Learning Representations 2020, Addis Ababa, Ethiopia, 2020: 1–13.
[15]	赵衎衎, 张良富, 张静, 等. 因子分解机模型研究综述[J]. 软件学报, 2019, 30(3): 799–821. doi: 10.13328/j.cnki.jos.005698 ZHAO Kankan, ZHANG Liangfu, ZHANG Jing, et al. Survey on factorization machines model[J]. Journal of Software, 2019, 30(3): 799–821. doi: 10.13328/j.cnki.jos.005698
[16]	李琳, 朱阁, 解庆, 等. 一种潜在特征同步学习和偏好引导的推荐方法[J]. 软件学报, 2019, 30(11): 3382–3396. doi: 10.13328/j.cnki.jos.005542 LI Lin, ZHU Ge, XIE Qing, et al. Recommendation approach by simultaneous learning latent features and preferences guidance[J]. Journal of Software, 2019, 30(11): 3382–3396. doi: 10.13328/j.cnki.jos.005542
[17]	KIM D, PARK C, OH J, et al. Convolutional matrix factorization for document context-aware recommendation[C]. The 10th ACM Conference on Recommender Systems, Boston, USA, 2016: 233–240. doi: 10.1145/2959100.2959165.