Yue Shang

Semantic retrieval, which retrieves semantically matched items given a textual query, has been an essential component to enhance system effectiveness in e-commerce search. In this paper, we study the multimodal retrieval problem, where the visual information (e.g, image) of item is leveraged as supplementary of textual information to enrich item representation and further improve retrieval performance. Though learning from cross-modality data has been studied extensively in tasks such as visual question answering or media summarization, multimodal retrieval remains a non-trivial and unsolved problem especially in the asymmetric scenario where the query is unimodal while the item is multimodal. In this paper, we propose a novel model named SMAR, which stands for Semantic-enhanced Modality-Asymmetric Retrieval, to tackle the problem of modality fusion and alignment in this kind of asymmetric scenario. Extensive experimental results on an industrial dataset show that the proposed model outperforms baseline models significantly in retrieval accuracy. We have open sourced our industrial dataset for the sake of reproducibility and future research works.

We propose a novel domain-specific generative pre-training (DSGPT) method for text generation and apply it to the product title and review summarization problems on E-commerce mobile display. First, we adopt a decoder-only transformer architecture, which fits well for fine-tuning tasks by combining input and output all together. Second, we demonstrate utilizing only small amount of pre-training data in related domains is powerful. Pre-training a language model from a general corpus such as Wikipedia or the Common Crawl requires tremendous time and resource commitment, and can be wasteful if the downstream tasks are limited in variety. Our DSGPT is pre-trained on a limited dataset, the Chinese short text summarization dataset (LCSTS). Third, our model does not require product-related human-labeled data. For title summarization task, the state of art explicitly uses additional background knowledge in training and predicting stages. In contrast, our model implicitly captures this knowledge and achieves significant improvement over other methods, after fine-tuning on the public Taobao.com dataset. For review summarization task, we utilize JD.com in-house dataset, and observe similar improvement over standard machine translation methods which lack the flexibility of fine-tuning. Our proposed work can be simply extended to other domains for a wide range of text generation tasks.

Due to the rapid growth of network data, the authenticity and reliability of network information have become increasingly important and have presented challenges. Most of the methods for fake review detection start with textual features and behavioral features. However, they are time-consuming and easily detected by fraudulent users. Although most of the existing neural network-based methods address the problems presented by the complex semantics of reviews, they do not account for the implicit patterns among users, reviews, and products; additionally, they do not consider the usefulness of information regarding fine-grained aspects in identifying fake reviews. In this paper, we propose an attention-based multilevel interactive neural network model with aspect constraints that mines the multilevel implicit expression mode of reviews and integrates four dimensions, namely, users, review texts, products and fine-grained aspects, into review representations. We model the relationships between users and products and use these relationships as a regularization term to redefine the model’s objective function. The experimental results from three public datasets show that the model that we propose is superior to the state-of-the-art methods; thus showing the effectiveness and portability of our model.