The Exploration and Application of Sentiment Analysis Technology in Meituan

In May 2021, Meituan NLP Center open sourced the largest Chinese attribute-level sentiment analysis data set ASAP based on real scenes to date. The relevant papers of this data set were accepted by NAACL2021, and the data set was added to Chinese open source data. A thousand words are planned to promote the progress of Chinese information processing technology together with other open source data sets. This article reviews the evolution of Meituan sentiment analysis technology and its application in typical business scenarios, including text/sentence-level sentiment analysis, attribute-level sentiment analysis, and opinion triple analysis. In business applications, relying on the ability of sentiment analysis technology to build online real-time prediction services and offline batch prediction services. So far, sentiment analysis services have provided services for more than ten business scenarios within Meituan.

references

• [1] https://github.com/Meituan-Dianping/asap.
• [2] Bu J, Ren L, Zheng S, et al. ASAP: A Chinese Review Dataset Towards Aspect Category Sentiment Analysis and Rating Prediction. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.
• [3] https://www.luge.ai/
• [4] Zhang, L. , S. Wang , and B. Liu . "Deep Learning for Sentiment Analysis : A Survey." Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery (2018):e1253.
• [5] Liu, Bing. "Sentiment analysis and opinion mining." Synthesis lectures on human language technologies 5.1 (2012): 1-167.
• [6] Peng, Haiyun, et al. "Knowing what, how and why: A near complete solution for aspect-based sentiment analysis." In Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 34. No. 05. 2020.
• [7] Zhang, Chen, et al. "A Multi-task Learning Framework for Opinion Triplet Extraction." In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: Findings. 2020.
• [8] Yoon Kim. 2014. Convolutional neural networks for sentence classification. arXiv preprint arXiv:1408.5882.
• [9] Peng Zhou, Wei Shi, Jun Tian, Zhenyu Qi, Bingchen Li,Hongwei Hao, and Bo Xu. 2016. Attention-based bidirectional long short-term memory networks for relation classification. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 207–212.
• [10] Devlin, Jacob, et al. “Bert: Pre-training of deep bidirectional transformers for language understanding.” arXiv preprint arXiv:1810.04805 (2018).
• [11] Yang Yang, Jia Hao, etc. Exploration and practice of BERT in Meituan.
• [12] Pontiki, Maria, et al. "Semeval-2016 task 5: Aspect based sentiment analysis." International workshop on semantic evaluation. 2016.
• [13] Pontiki, M. , et al. "SemEval-2014 Task 4: Aspect Based Sentiment Analysis." In Proceedings of International Workshop on Semantic Evaluation at (2014).
• [14] Yequan Wang, Minlie Huang, and Li Zhao. 2016. Attention-based lstm for aspect-level sentiment classification. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 606–615.
• [15] Sara Sabour, Nicholas Frosst, and Geoffrey E Hinton. 2017. Dynamic routing between capsules. In Advances in neural information processing systems, pages 3856–3866.
• [16] Chi Sun, Luyao Huang, and Xipeng Qiu. 2019. Utilizing bert for aspect-based sentiment analysis via constructing auxiliary sentence. arXiv preprint arXiv:1903.09588.
• [17] Qingnan Jiang, Lei Chen, Ruifeng Xu, Xiang Ao, and Min Yang. 2019. A challenge dataset and effective models for aspect-based sentiment analysis. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 6281–6286.
• [18] Wu, Zhen, et al. "Grid Tagging Scheme for End-to-End Fine-grained Opinion Extraction." In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: Findings. 2020.
• [19] Liu, Yinhan, et al. "Roberta: A robustly optimized bert pretraining approach." arXiv preprint arXiv:1907.11692 (2019).
• [20] Clark, Kevin, et al. "Electra: Pre-training text encoders as discriminators rather than generators." arXiv preprint arXiv:2003.10555 (2020).
  0- [21] Timothy Dozat and Christopher D. Manning. 2017.Deep biaffine attention for neural dependency parsing. In 5th International Conference on Learning Representations, ICLR 2017.

about the author

Ren Lei, Jiahao, Zhang Chen, Yang Yang, Mengxue, Ma Fang, King Kong, Wuwei, etc., all come from the NLP Center of the Meituan Platform Search and NLP Department.

Job Offers

Meituan Search and NLP Department/NLP Center is the core team responsible for the research and development of Meituan artificial intelligence technology. The mission is to build world-class core technology and service capabilities for natural language processing.

The NLP Center has been recruiting natural language processing algorithm experts/machine learning algorithm experts for a long time. Interested students can send their resumes to [e-mail]. Specific requirements are as follows.

Job responsibilities

1. Prospective exploration of pre-training language models, including but not limited to knowledge-driven pre-training, task-based pre-training, multi-modal model pre-training, and cross-language pre-training;
2. Responsible for the training and performance optimization of super-large models with more than tens of billions of parameters;
3. Model fine-tuning and forward-looking technology exploration, including but not limited to Prompt Tuning, Adapter Tuning, and various Parameter-efficient transfer learning directions;
4. Prospective exploration of model inference/training compression technology, including but not limited to quantization, pruning, tensor analysis, KD and NAS, etc.;
5. Complete the application of pre-training models in search, recommendation, advertising and other business scenarios and achieve business goals;
6. Participate in the construction and promotion of the internal NLP platform of Meituan

job requirements

1. More than 2 years of relevant work experience, participated in algorithm development in at least one of search, recommendation, and advertising, and paid attention to the progress of the industry and academia;
2. A solid algorithm foundation, familiar with natural language processing, knowledge graph and machine learning technology, and passion for technology development and application;
3. Familiar with programming languages such as Python/Java, and have certain engineering skills;
4. Familiar with Tensorflow, PyTorch and other deep learning frameworks and have actual project experience;
5. Familiar with NLP models such as RNN/CNN/Transformer/BERT/GPT and have actual project experience;
6. Strong sense of purpose, good at analyzing and finding problems, dismantling and simplifying, and being able to discover new spaces in daily work;
7. Organized and motivated, able to sort out complicated tasks and establish effective mechanisms to promote upstream and downstream cooperation to complete goals.

bonus item

1. Familiar with the basic principles of each Optimizer for model training, and understand the basic methods and framework of distributed training;
2. Have an understanding of the latest training acceleration methods, such as mixed precision training, low-bit training, distributed gradient compression, etc.

Read more technical articles from the

the front | algorithm | backend | data | security | operation and maintenance | iOS | Android | test

| in the public account menu bar dialog box, and you can view the collection of technical articles from the Meituan technical team over the years.

| This article is produced by the Meituan technical team, and the copyright belongs to Meituan. Welcome to reprint or use the content of this article for non-commercial purposes such as sharing and communication, please indicate "the content is reproduced from the Meituan technical team". This article may not be reproduced or used commercially without permission. For any commercial activity, please send an email to tech@ meituan.com apply for authorization.