WXWenwen Xia

Wenwen Xia

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Wenwen Xia has a total of 28 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Contrastive Flow Matching for Collaborative Filtering

    by Wangyu Jin, Jiansheng Qian, Wenwen Xia, Hongliang He, Guanfeng Liu, Pengpeng Zhao

  • Open Access

    SurvFormer: A Transformer Based Framework for Survival Analysis in Insurance Underwriting

    by Yingxue Li, Xiaolu Zhang, Jun Hu, Wenwen Xia, Ziyi Liu, Xiaobo Qin, Binjie Fei, Jun Zhou


  • Open Access

    Enhancing Sequential Recommendation via LLM-based Semantic Embedding Learning

    by Jun Hu, Wenwen Xia, Xiaolu Zhang, Chilin Fu, Weichang Wu, Zhaoxin Huan, Ang Li, Zuoli Tang, Jun Zhou

    Sequential recommendation systems (SRS) are crucial in various applications as they enable users to discover relevant items based on their past interactions. Recent advancements involving large language models (LLMs) have shown significant promise in addressing intricate recommendation challenges. However, these efforts exhibit certain limitations. Specifically, directly extracting representations from an LLM based on items' textual features and feeding them into a sequential model hold no guarantee that the semantic information of texts could be preserved in these representations. Additionally, concatenating textual descriptions of all items in an item sequence into a long text and feeding it into an LLM for recommendation results in lengthy token sequences, which largely diminishes the practical efficiency. In this paper, we introduce SAID, a framework that utilizes LLMs to explicitly learn Semantically Aligned item ID embeddings based on texts. For each item, SAID employs a projector module to transform an item ID into an embedding vector, which will be fed into an LLM to elicit the exact descriptive text tokens accompanied by the item. The item embeddings are forced to preserve fine-grained semantic information of textual descriptions. Further, the learned embeddings can be integrated with lightweight downstream sequential models for practical recommendations. In this way, SAID circumvents lengthy token sequences in previous works, reducing resources required in industrial scenarios and also achieving superior recommendation performance. Experiments on six public datasets demonstrate that SAID outperforms baselines by about 5% to 15% in terms of NDCG@10. Moreover, SAID has been deployed in Alipay's online advertising platform, achieving a 3.07% relative improvement of cost per mille (CPM) over baselines, with an online response time of under 20 milliseconds.


  • Open Access

    Predictive value of the C-reactive protein/albumin ratio in severity and prognosis of acute pancreatitis

    by Yi Zhao, Wenwen Xia, You Lu, Wei Chen, Yan Zhao, Yugang Zhuang

    Aim: To investigate the predictive value of C-reactive protein (CRP) to serum albumin (ALB) ratio in the severity and prognosis of acute pancreatitis (AP), and compare the predictive value of the CRP/ALB ratio with the Ranson score, modified computed tomography severity index (MCTSI) score, and Bedside Index of Severity in Acute Pancreatitis (BISAP) score.Methods: This cohort study retrospectively analyzed clinical data of AP patients from August 2018 to August 2020 in our hospital. Logistic regression analysis was utilized to determine the effects of CRP/ALB ratio, Ranson, MCTSI, and BISAP score on severe AP (SAP), pancreatic necrosis, organ failure, and death. The predictive values of CRP/ALB ratio, Ranson, MCTSI, and BISAP score were examined with the area under the curve (AUC) of the receiver operator characteristic (ROC) curve analysis. DeLong test was used to compare the AUCs between CRP/ALB ratio, Ranson, MCTSI, and BISAP score.Results: Totally, 284 patients were included in this study, of which 35 AP patients (12.32%) developed SAP, 29 (10.21%) organ failure, 30 (10.56%) pancreatic necrosis and 11 (3.87%) died. The result revealed that CRP/ALB ratio on day 2 was associated with SAP [odds ratio (OR): 1.74, 95% confidence interval (CI): 1.32 to 2.29], death (OR: 1.73, 95%CI: 1.24 to 2.41), pancreatic necrosis (OR: 1.28, 95%CI: 1.08 to 1.50), and organ failure (OR: 1.43, 95%CI: 1.18 to 1.73) in AP patients. Similarly, CRP/ALB on day 3 was related to a higher risk of SAP (OR: 1.50, 95%CI: 1.24 to 1.81), death (OR: 1.8, 95%CI: 1.34 to 2.65), pancreatic necrosis (OR: 1.22, 95%CI: 1.04 to 1.42), and organ failure (OR: 1.21, 95%CI: 1.04 to 1.41). The predictive value of CRP/ALB ratio for pancreatic necrosis was lower than that of MCTSI, for organ failure was lower than that of Ranson and BISAP, and for death was higher than that of MCTSI.Conclusion: The CRP/ALB ratio may be a novel but promising, easily measurable, reproducible, non-invasive prognostic score that can be used to predict SAP, death, pancreatic necrosis, and organ failure in AP patients, which can be a supplement of Ranson, MCTSI, and BISAP scores.


  • On the Substructure Countability of Graph Neural Networks

    by Wenwen Xia, Yuchen Li, Shenghong Li

    With the empirical success of Graph Neural Networks (GNNs) on graph-related tasks, it is intriguing to investigate their theoretical power on these tasks. In this paper, we focus on GNNs' theoretical power on substructure counting, a fundamental yet challenging task in many applications. Previous works have proven that the 2-dimensional Weisfeiler-Leman algorithm (2-WL) equivalent GNNs can only count limited substructures. However, the substructure counting ability of theoretically more powerful and computationally tractable GNNs remains unclear. In this paper, we study conditions for substructures to be theoretically countable by k-WL equivalent GNNs, and then focus on 3-WL equivalent ones, which are currently the most theoretically powerful instances with practical computational cost. Further, we propose an algorithm to determine the countability of substructures for 3-WL equivalent GNNs. Our results reveal that 3-WL equivalent GNNs can count considerably more substructures than 2-WL equivalent ones. However, the proportion of countable patterns and prediction performance decrease as the pattern size increases. Therefore, we propose a Layer Permutation Pooling (LPP) model for better substructure counting performance. LPP first decomposes the data graph into subgraphs. Then we propose a layer permutation scheme to represent each decomposed subgraph as a set of matrices. Finally, LPP utilizes a neural network to conduct predictions on matrices. We compare LPP with several state-of-the-art GNNs on various datasets. Experimental results show that LPP outperforms baselines by 84% on average with the RMSE metric.