Search

Abstract In recent years, recommendation explanation methods have received widespread attention due to their potentials to enhance user experience and streamline transactions. In scenarios where auxiliary information such as text and attributes are lacking, counterfactual explanation has emerged as a crucial technique for explaining recommendations. However, existing counterfactual explanation methods encounter two primary challenges. First, a substantial bias indeed exists in the calculation of the group impact function, leading to the inaccurate predictions as the counterfactual explanation group expands. In addition, the importance of collaborative filtering as a counterfactual explanation is overlooked, which results in lengthy, narrow, and inaccurate explanations. To address such issues, we propose a counterfactual explanation method based on Modified Group Influence Function for recommendation. In particular, via a rigorous formula derivation, we demonstrate that a simple summation of individual influence functions cannot reflect the group impact in recommendations. After that, building upon the improved influence function, we construct the counterfactual groups by iteratively incorporating the individuals from the training samples, which possess the greatest influence on the recommended results, and continuously adjusting the parameters to ensure accuracy. Finally, we expand the scope of searching for counterfactual groups by incorporating the collaborative filtering information from different users. To evaluate the effectiveness of our method, we employ it to explain the recommendations generated by two common recommendation models, i.e., Matrix Factorization and Neural Collaborative Filtering, on two publicly available datasets. The evaluation of the proposed counterfactual explanation method showcases its superior performance in providing counterfactual explanations. In the most significant case, our proposed method achieves a 17% lead in terms of Counterfactual precision compared to the best baseline explanation method.

Abstract Recommender system always suffers from various recommendation biases, seriously hindering its development. In this light, a series of debias methods have been proposed in the recommender system, especially for two most common biases, i.e., popularity bias and amplified subjective bias. However, existing debias methods usually concentrate on correcting a single bias. Such single-functionality debiases neglect the bias-coupling issue in which the recommended items are collectively attributed to multiple biases. Besides, previous work cannot tackle the lacking supervised signals brought by sparse data, yet which has become a commonplace in the recommender system. In this work, we introduce a disentangled debias variational auto-encoder framework (DB-VAE) to address the single-functionality issue as well as a counterfactual data enhancement method to mitigate the adverse effect due to the data sparsity. In specific, DB-VAE first extracts two types of extreme items only affected by a single bias based on the collier theory, which are, respectively, employed to learn the latent representation of corresponding biases, thereby realizing the bias decoupling. In this way, the exact unbiased user representation can be learned by these decoupled bias representations. Furthermore, the data generation module employs Pearl’s framework to produce massive counterfactual data to help fully train the model, making up the lacking supervised signals due to the sparse data. Extensive experiments on three real-world data sets demonstrate the effectiveness of our proposed model. Specifically, our model outperforms the best baseline by 19.5% in terms of Recall@20 and 9.5% in terms of NDCG@100 in the best scenario. Besides, the counterfactual data can further improve DB-VAE, especially on the data set with low sparsity.

Document-level Event Argument Extraction (DEAE) aims to extract structural event knowledge composed of arguments and roles beyond the sentence level. Existing methods mainly focus on designing prompts and using Abstract Meaning Representation (AMR) graph structure as additional features to enrich event argument representation. However, two challenges still remain: (1) the long-range dependency between event trigger and event arguments and (2) the distracting context in the document towards an event that can mislead the argument classification. To address these issues, we propose a novel document-level event argument extraction model named AMR Parser and Sparse Representation (APSR). Specifically, APSR sets inter- and intra-sentential encoders to capture the contextual information in different scopes. Especially, in the intra-sentential encoder, APSR designs three types of sparse event argument attention mechanisms to extract the long-range dependency. Then, APSR constructs AMR semantic graphs, which capture the interactions among concepts well. Finally, APSR fuses the inter- and intra-sentential representations and predicts what role a candidate span plays. Experimental results on the RAMS and WikiEvents datasets demonstrate that APSR achieves a superior performance compared with competitive baselines in terms of F1 by 1.27% and 3.12%, respectively.

Abstract Recently, segmentation-based approaches have been proposed to tackle arbitrary-shaped text detection. The trade-off between speed and accuracy is still a challenge that hinders its deployment in practical applications. Previous methods adopt complex pipelines to improve accuracy while ignoring inference speed. Moreover, the performance of most efficient scene text detectors often suffers from weak feature extraction when equipping lightweight networks. In this paper, we propose a novel distillation method for efficient and accurate arbitrary-shaped text detection, termed kernel-mask knowledge distillation. Our approach equips a low computational-cost visual transformer module (VTM) and a feature adaptation layer to make full use of feature-based and response-based knowledge in distillation. More specifically, first, the text features are obtained by aggregating the multi-level information extracted in the respective backbones of the teacher and student networks. Second, the text features are respectively sent to the VTM to enhance the feature representation ability. Then, we distill the feature-based and response-based kernel knowledge of the teacher network to obtain an efficient and accurate arbitrary-shaped text detection model. Extensive experiments on publicly available datasets demonstrate the state-of-the-art performance of our method. It is worth noting that our method can achieve a competitive F-measure of 86.92% at 34.5 FPS on Total-text. Code is available at https://github.com/giganticpower/KKDnet .

The Yangtze River Valley (YRV) experienced record-breaking extreme high temperature and weakened rainfall events in August 2022, which resulted in severe disasters and large socioeconomic losses. The importance of the intensification and westward movement of the western Pacific subtropical high (WPSH) with abnormal subsidence has been emphasized in previous studies. However, the source of the abnormal subsidence remains unknown. This study investigates the source of the abnormal subsidence over the YRV and discusses its possible causes by adopting the three-pattern decomposition of the atmospheric circulation (3P-DGAC). Meridional circulation (MC) was the main contributor to the abnormal vertical velocity (114%), while the contribution of zonal circulation (ZC) was negative (–14%). Additionally, the negative rainfall anomaly over the YRV can be explained mainly by the MC. The anomalous MC was characterized as a ‘negative-positive-negative-positive-negative’ quintuple distribution with sinking motion over the YRV. Anomalous MC is closely related to the sea surface temperature anomaly (SSTA) over the three oceans. The negative phase of the Indian Ocean dipole (IOD) and La Niña SSTA leads to an anomalous rising motion of the ZC over the Maritime Continent, favoring the existence of the rising motion of the MC by the coupling effect. The positive phase of the North Atlantic triple (NAT) SSTA results in an anomalous Rossby wave train, which further leads to a sinking motion over the YRV.

Text classification is a machine learning technique employed to assign a given text to predefined categories, facilitating the automatic analysis and processing of textual data. However, an important problem is that the number of new text categories is growing faster than that of human annotation data, which makes many new categories of text data lack a lot of annotation data. As a result, the conventional deep neural network is forced to over-fit, which damages the application in the real world. As a solution to this problem, academics recommend addressing data scarcity through few-shot learning. One of the efficient methods is prompt-tuning, which transforms the input text into a mask prediction problem featuring [MASK]. By utilizing descriptors, the model maps output words to labels, enabling accurate prediction. Nevertheless, the previous prompt-based adaption approaches often relied on manually produced verbalizers or a single label to represent the entire label vocabulary, which makes the mapping granularity low, resulting in words not being accurately mapped to their label. To address these issues, we propose to enhance the verbalizer and construct the refined external knowledge into a prompt-tuning (REKP) model. We employ the external knowledge bases to increase the mapping space of tagged terms and design three refinement methods to remove noise data. We conduct comprehensive experiments on four benchmark datasets, namely AG’s News, Yahoo, IMDB, and Amazon. The results demonstrate that REKP can outperform the state-of-the-art baselines in terms of Micro-F1 on knowledge-enhanced text classification. In addition, we conduct an ablation study to ascertain the functionality of each module in our model, revealing that the refinement module significantly contributes to enhancing classification accuracy.

Abstract For a selected train journey with multiple trains in the same traction power system, the given driving strategies of other trains can be abstracted as available external power flow. Aims at reducing the energy consumption of the multi‐train system, this paper presents an analytical solution for optimising the selected train trajectory to maximise the utilisation of the external power flow. Specifically, a spatial‐temporal area which combine the track grade along space and the external power over time is constructed and the problem is converted into the energy‐efficient control of a train travels in the spatial‐temporal area. Classical methods of constrained optimisation is used to show that the set of the optimal control modes and the optimal driving speed for the proposed problem are in a state of change. In order to obtain the optimal driving strategy, the spatial‐temporal area is divided into grids, and the evolution of a key adjoint variable when the external power changes is discussed. The authors illustrate our results by finding optimal control sequences and associated speed profile for the freight train in some elementary but realistic examples.

Aims. To evaluate the value of endoscopic screening during endoscopic submucosal dissection (ESD) in the detection of synchronous multiple early gastric cancer (SMEGC) and the risk factors for missed diagnosis of SMEGC. Methods. We conducted gastric endoscopic screening during ESD operation in 271 patients with early gastric cancer (EGC) referred for ESD, and endoscopic follow-up within 1 year after the operation. The detection and characteristics of SMEGC were analyzed in three stages: before ESD, during ESD operation, and within 1 year after ESD. Results. SMEGC was detected in 37 of 271 patients (13.6%). Among them, 21 patients with SMEGC (56.8%) were diagnosed before ESD, 9 (24.3%) were diagnosed with SMEGC by endoscopic screening during ESD operation, and 7 (18.9%) were found to have EGC lesions in the stomach during postoperative endoscopic follow-up within 1 year. The preoperative missed detection rate of SMEGC was 43.2%, and the rate of missed detection could be reduced by 24.3% (9/37) with endoscopic screening during ESD operation. Missed SMEGC lesions were more common in flat or depressed type and smaller in size than the lesions found before ESD. The presence of severe atrophic gastritis and age ≥60 years were significantly correlated with SMEGC (P

Graph Collaborative Filtering (GCF) methods have emerged as an effective recommendation approach, capturing users’ preferences over items by modeling user–item interaction graphs. However, these methods suffer from data sparsity in real scenarios, and their performance can be improved using contrastive learning. In this paper, we propose an optimized method, named LoRA-NCL, for GCF based on Neighborhood-enriched Contrastive Learning (NCL) and low-rank dimensionality reduction. We incorporate low-rank features obtained through matrix factorization into the NCL framework and employ LightGCN to extract high-dimensional representations. Extensive experiments on five public datasets demonstrate that the proposed method outperforms a competitive graph collaborative filtering base model, achieving 4.6% performance gains on the MovieLens dataset, respectively.

Taking COVID-19 as an example, we know that a pandemic can have a huge impact on normal human life and the economy. Meanwhile, the population flow between countries and regions is the main factor affecting the changes in a pandemic, which is determined by the airline network. Therefore, realizing the overall control of airports is an effective way to control a pandemic. However, this is restricted by the differences in prevention and control policies in different areas and privacy issues, such as how a patient’s personal data from a medical center cannot be effectively combined with their passenger personal data. This prevents more precise airport control decisions from being made. To address this, this paper designed a novel data-sharing framework (i.e., PPChain) based on blockchain and federated learning. The experiment uses a CPU i7-12800HX and uses Docker to simulate multiple virtual nodes. The model is deployed to run on an NVIDIA GeForce GTX 3090Ti GPU. The experiment shows that the relationship between a pandemic and aircraft transport can be effectively explored by PPChain without sharing raw data. This approach does not require centralized trust and improves the security of the sharing process. The scheme can help formulate more scientific and rational prevention and control policies for the control of airports. Additionally, it can use aerial data to predict pandemics more accurately.

Purpose: This is the first study to compare the pharmacokinetics, safety and, immunogenicity of QL1209, a biosimilar of Perjeta®.Methods: This study was a randomized, double-blind, parallel-controlled clinical trial evaluating the biosimilarity between QL1209 (specification: 420 mg:14 ml, single use via, manufacturer: Qilu Pharmaceutical Co., Ltd., batch number: 201808001KJL) and Perjeta® (specification: 420 mg: 14 ml, single use via, manufacturer: Roche Pharma AG, batch number: H0309H02). The trial period was 99 days (blood samples for PK were collected 99 days after infusion). Serum concentrations were determined using a validated assay. PK parameters were calculated using a non-compartmental model and analyzed statistically. Anti-drug antibody (ADA)-positive samples were further tested for the presence of neutralization antibody detection (NAb).Results: A total of 137 healthy subjects were administrated. The subjects were randomized 1:1 to receive QL1209 or Perjeta® 420 mg intravenously. The geometric mean ratio (GMRs) for QL1209 versus Perjeta® are 104.14%, 104.09%, and 110.59% for Cmax, AUC0-t, and AUC0-∞, respectively, and their 90% confidence interval (CIs) all fell within the predefined bioequivalence margin 80.00–125%. The incidence of drug-related adverse events was 95.6% and 95.5% in the QL1209 and Perjeta® groups, respectively, also comparable between the two groups.Conclusion: The results of this comparative clinical pharmacology study demonstrated the PK similarity of QL1209 (420 mg: 14 ml) and Perjeta® (420 mg: 14 ml) and there was no significant difference in safety and immunogenicity between QL1209 and Perjeta® manufactured by Roche Pharma AG.

The long non-coding RNA (lncRNA) ASAP1-IT1 has been recently shown to aberrantly increase in ovarian and bladder cancer, while its role in other malignancies remains unexplored. This study was to characterize the expression and assess the potential role of ASAP1-IT1 in hepatocellular carcinoma (HCC). Fifty-four paired HCC and histologically normal tissues were obtained from HCC patients. Human HCC cell lines (HepG2, Huh7, SMMC-7721, and BEL-7402) and a normal liver cell line (LO2) were used for in vitro studies. ASAP1-IT1-specific siRNAs were used to silence ASAP1-IT1 expression, while the pcDNA-ASAP1-IT1 vector was constructed to up-regulate its expression. In situ hybridization and qRT-PCR were performed to characterize subcellular localization and expression of ASAP1-IT1. Cell proliferation and migration assays were conducted to examine the role of ASAP1-IT1 in the progression of HCC. In silico analysis was conducted to predict putative miRNA binding sites, which were validated by luciferase reporter assays. ASAP1-IT1 levels were significantly increased in HCC tissues and cells compared with controls. Notably, higher ASAP1-IT1 levels were significantly associated with poorer prognosis of HCC patients. In situ hybridization analysis revealed that ASAP1-IT1 was mainly localized in the nucleus of hepatoma cells and differentially expressed in trabecular, compact, and pseudoglandular forms of liver cancer. Furthermore, knockdown of ASAP1-IT1 significantly suppressed cell proliferation and migration, while its overexpression significantly promoted cell proliferation and migration of HCC cells. Mechanistically, ASAP1-IT1 might exert its role in HCC progression, at least in part, by directly interacting with miR-221-3p. In conclusion, ASAP1-IT1 is abnormally elevated in HCC, and higher levels are correlated with poorer prognosis. An underlying mechanism has been proposed for ASAP1-IT1-associated promotion of proliferation and migration in HCC cells. These findings have provided evidence supporting the oncogenic role of ASAP1-IT1 in HCC.

Relation extraction aims to predict the relation triple between the tail entity and head entity in a given text. A large body of works adopt meta-learning to address the few-shot issue faced by relation extraction, where each relation category only contains few labeled data for demonstration. Despite promising results achieved by existing meta-learning methods, these methods still struggle to distinguish the subtle differences between different relations with similar expressions. We argue this is largely owing to that these methods cannot capture unbiased and discriminative features in the very few-shot scenario. For alleviating the above problems, we propose a taxonomy-aware prototype network, which consists of a category-aware calibration module and a task-aware training strategy module. The former implicitly and explicitly calibrates the representation of prototype to become sufficiently unbiased and discriminative. The latter balances the weight between easy and hard instances, which enables our proposal to focus on data with more information during the training stage. Finally, comprehensive experiments are conducted on four typical meta tasks. Furthermore, our proposal presents superiority over the competitive baselines with an improvement of 3.30% in terms of average accuracy.

Text classification aims to assign predefined labels to unlabeled sentences, which tend to struggle in real-world applications when only a few annotated samples are available. Previous works generally focus on using the paradigm of meta-learning to overcome the classification difficulties brought by insufficient data, where a set of auxiliary tasks is given. Accordingly, prompt-based approaches are proposed to deal with the low-resource issue. However, existing prompt-based methods mainly focus on English tasks, which generally apply English pretrained language models that can not directly adapt to Chinese tasks due to structural and grammatical differences. Thus, we propose a prompt-based Chinese text classification framework that uses generated natural language sequences as hints, which can alleviate the classification bottleneck well in low-resource scenarios. In detail, we first design a prompt-based fine-tuning together with a novel pipeline for automating prompt generation in Chinese. Then, we propose a refined strategy for dynamically and selectively incorporating demonstrations into each context. We present a systematic evaluation for analyzing few-shot performance on a wide range of Chinese text classification tasks. Our approach makes few assumptions about task resources and expertise and therefore constitutes a powerful, task-independent approach for few-shot learning.

Intent recognition aims to identify users’ potential intents from their utterances, which is a key component in task-oriented dialog systems. A real challenge, however, is that the number of intent categories has grown faster than human-annotated data, resulting in only a small amount of data being available for many new intent categories. This lack of data leads to the overfitting of traditional deep neural networks on a small amount of training data, which seriously affects practical applications. Hence, some researchers have proposed few-shot learning should address the data-scarcity issue. One of the efficient methods is text augmentation, which always generates noisy or meaningless data. To address these issues, we propose leveraging the knowledge in pre-trained language models and constructed the cloze-style data augmentation (CDA) model. We employ unsupervised learning to force the augmented data to be semantically similar to the initial input sentences and contrastive learning to enhance the uniqueness of each category. Experimental results on CLINC-150 and BANKING-77 datasets show the effectiveness of our proposal by its beating of the competitive baselines. In addition, we conducted an ablation study to verify the function of each module in our models, and the results illustrate that the contrastive learning module plays the most important role in improving the recognition accuracy.

Unmanned Aerial Vehicles, commonly known as drones, have been widely used in transmission line inspection and traffic patrolling due to their flexibility and environmental adaptability. To take advantage of drones and overcome their limited endurance, the patrolling tasks are parallelized by concurrently dispatching the drones from a truck which travels on the road network to the nearby task arc. The road network considered in previous research is undirected; however, in reality, the road network usually contains unidirectional arcs, i.e., the road network is asymmetric. Hence, we propose an asymmetric coordinated vehicle-drones arc routing mode for traffic patrolling. In this mode, a truck travelling on an asymmetric road network with multiple drones needs to patrol multiple task arcs, and the drones can be launched and recovered at certain nodes on the truck route, making it possible for drones and the truck to patrol the task in parallel. The total patrol time is the objective function that needs to be minimized given the time limit constraints of drones. The whole problem can be considered as an asymmetric arc routing problem of coordinating a truck and multiple drones. To solve this problem, a large-scale neighborhood search with simulated annealing algorithm (LNS-SA) is proposed. Finally, extensive computation experiments and a real case are carried out. The experimental results show the efficiency of the proposed algorithm. Moreover, a detailed sensitivity analysis is performed on several drone-parameters of interest.

Traditional construction machinery’s full hydraulic steering system has high energy consumption. An electro-hydraulic flow matching steering system for electric wheel loaders based on closed-circuit pump control is proposed to solve the problem. The transfer function of the electro-hydraulic system is established, and the system is stable according to analysis via Routh matrix. A test platform is built to verify the effectiveness of the system and the control strategy. Taking a 1.6T wheel loader as an example, the energy consumption of the traditional steering system and the new steering system under zero-load, positive-load (shovel loaded with 600 kg gravel), and offset-load (the center of gravity of the gravel is off the center of the bucket) conditions is compared. The results show that the energy consumption of the proposed steering system is greatly reduced compared to the traditional system. Under the condition of zero-load with medium steering speed, compared to the traditional system, consumption is reduced by 22.8%.

Chinese Medical Named Entity Recognition (Chinese-MNER) aims to identify potential entities and their categories from the unstructured Chinese medical text. Existing methods for this task mainly incorporate the dictionary knowledge on the basis of traditional BiLSTM-CRF or BERT architecture. However, the construction of high-quality dictionaries is typically time consuming and labor-intensive, which may also damage the robustness of NER models. What is more, the limited amount of annotated Chinese-MNER data can easily lead to the over-fitting problem while training. With the aim of dealing with the above problems, we put forward a BERT-BiLSTM-CRF model by integrating the part-of-speech (POS) tagging features and a Regularization method (BBCPR) for Chinese-MNER. In BBCPR, we first leverage a POS fusion layer to incorporate external syntax knowledge. Next, we design a novel REgularization mothod with Adversarial training and Dropout (READ) to improve the model robustness. Specifically, READ focuses on reducing the difference between the predictions of two sub-models through minimizing the bidirectional KL divergence between the adversarial output and original output distributions for the same sample. Comprehensive evaluations on two public data sets, namely, cMedQANER and cEHRNER from the Chinese Biomedical Language Understanding Evaluation benchmark (ChineseBLUE), demonstrate the superiority of our proposal in Chinese-MNER. In addition, ablation study shows that READ can effectively improve the model performance. Our proposal does well in exploring the technical terms and identifying the word boundary.