Neural Rating Regression with Abstractive Tips Generation for Recommendation

By Piji Li, Zihao Wang, Zhaochun Ren, Lidong Bing, Wai Lam. 2017

338 citations recorded by the source corpus.

Recently, some E-commerce sites launch a new interaction box called Tips on their mobile apps. Users can express their experience and feelings or provide suggestions using short texts typically several words or one sentence. In essence, writing some tips and giving a numerical rating are two facets of a user's product assessment action, expressing the user experience and feelings. Jointly modeling these two facets is helpful for designing a better recommendation system. While some existing models integrate text information such as item specifications or user reviews into user and item latent factors for improving the rating prediction, no existing works consider tips for improving recommendation quality. We propose a deep learning based framework named NRT which can simultaneously predict precise ratings and generate abstractive tips with good linguistic quality simulating user experience and feelings. For abstractive tips generation, gated recurrent neural networks are employed to "translate'' user and item latent representations into a concise sentence. Extensive experiments on benchmark datasets from different domains show that NRT achieves significant improvements over the state-of-the-art methods. Moreover, the generated tips can vividly predict the user experience and feelings.

Subfields: Artificial Intelligence, Information Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

Citations in the corpus

Explainable Recommendation: A Survey and New Perspectives (2018, background): “Tips are more concise than reviews and can reveal user experience, feelings, and suggestions with only a few words (Li et al. [2017]).”
Justifying Recommendations using Distantly-Labeled Reviews and Fine-Grained Aspects (2019, background): “More recently, there has been work studying the task of tip generation where tips are concise summaries of reviews (Li et al., 2017).”
Generate Natural Language Explanations for Recommendation (2021, background): “As a result, researchers have looked into explainable recommendation systems in the recent years [2, 3, 8, 16, 27, 38, 39, 44, 45], which can not only provide users with the recommendation lists, but also intutive explanations about why these items are recommended.”
Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5) (2022, background): “We compare with the following explanation generation baselines:Attn2Seq [8],NRT [27], PETER [26] and its feature-enhanced version PETER+ [26].”
Visually Explainable Recommendation (2018, background): “• NRT: The neural rating regression model [16] is a state-of-the-art neural recommender which can generate user textual reviews.”
Explainable Outfit Recommendation with Joint Outfit Matching and Comment Generation (2018, methodology): “For comparison, we adapt NRT to make it compatible with outfit recommendation.”
AliCoCo: Alibaba E-commerce Cognitive Concept Net (2020, neutral): “, “how many people also viewed” [9, 17]).”
Recommendation system based on deep learning methods: a systematic review and new directions (2019, methodology): “One of the advantages of utilizing multi-task learning in a deep neural network includes its ability to reduce the data sparsity problem through implicit data augmentation (Li et al. 2017).”
Adversarial Distillation for Efficient Recommendation with External Knowledge (2018, background): “generation [23] and machine translation [28] – our review prediction network is more likely to focus on the sentiment and opinion-rich part of the text, and the result does not heavily rely on the specific position of the sentiment words.”
Neural Attentional Rating Regression with Review-level Explanations (2018, methodology): “NRT[25] combines gated recurrent neural networks and collaborative ltering to simultaneously predict ratings and generate abstractive tips simulating user experience and feelings.”
When large language models meet personalization: perspectives of challenges and opportunities (2023, neutral): “…explanation methods are tightly coupled with speciﬁc recommendation models (e.g., NETE [141]), or directly design a new recommendation model (e.g., NRT [139], 123 PETER [144]), and they are often powerless when faced with existing advanced recommendation models, which limits their generalizability.”
A Survey on Accuracy-Oriented Neural Recommendation: From Collaborative Filtering to Information-Rich Recommendation (2021, methodology): “Compared to review generation models [112], the visual embedding is injected into both”
Personalized Reason Generation for Explainable Song Recommendation (2019, background): “Li et al. [50] propose a deep-learning-based framework called NRT that can simultaneously predict precise ratings and summarize the massive reviews to generate abstractive tips.”
Co-Attentive Multi-Task Learning for Explainable Recommendation (2019, methodology): “To measure explainability, we leverage BLEU [Papineni et al., 2002] and ROUGE [Lin, 2004], which are widely used in 4https://www.yelp.com/dataset/challenge machine translation and natural language generation to evaluate the similarity between ground truth texts and generated texts [Li et al., 2016;…”
Neural Attentive Session-based Recommendation (2017, background): “Deep models can be trained to learn discriminative representations from unstructured data [9, 11, 19].”
Personalized Fashion Recommendation with Visual Explanations based on Multimodal Attention Network: Towards Visually Explainable Recommendation (2019, background): “The overall comparison between our VECF model and the baselines are presented in Table 3, we can see: • By integrating user reviews or product images, VBPR and NRT obtained better performance than BPR in most cases, which verifies the effectiveness of these information for the task of Top-N recommen”
Deep Learning Based Recommender System (2017, methodology): “Neural Attention [14, 44, 70, 90, 100, 103, 128, 146, 170, 190, 195, 206], [62, 147, 194] Adversary Network [9, 52, 163, 165] DRL [16, 21, 107, 169, 199–201] Hybrid Models [17, 38, 41, 82, 84, 87, 119, 136, 161, 193, 194]”
A Reinforcement Learning Framework for Explainable Recommendation (2018, methodology): “Here we use GRU in our implementation as it is computationally efficient and has comparable performance with LSTM [20].”
Multi-Pointer Co-Attention Networks for Recommendation (2018, positive): “[15] proposed a encoder-decoder based framework for both rating prediction and generating tips.”
Is ChatGPT a Good Recommender? A Preliminary Study (2023, methodology): “For explanation generation, we utilize Attn2Seq [15], NRT 1.0751 0.6977 [30] and PETER [29] as baselines.”