Explore the vast range of titles available.

The rapid expansion of online education platforms has led to an influx of false reviews, complicating users’ ability to identify suitable courses promptly. Addressing these challenges, this paper introduces ICRA (Intelligent Course Review Analysis), a novel model that identifies and filters false reviews using a custom sentiment lexicon and a pre-trained ERNIE 3.0 model. ICRA enhances data quality by analyzing user reviews and course profiles comprehensively for recommendation purposes. The model utilizes the BERT lexicon and ERNIE 3.0 to obtain deep semantic representations. It integrates BiLSTM with a multi-head attention mechanism to capture essential review details, aiming to minimize overfitting and enhance generalization. By predicting user review scores and verifying review authenticity, ICRA boosts recommendation accuracy and robustness, addressing the cold-start issue. Experimental findings highlight ICRA’s excellence in predicting user ratings and delivering precise course recommendations efficiently. This capability streamlines course selection on online education platforms, improving learning experiences and efficiency.

Coxiella burnetii is the etiological agent of Q fever. The clinical diagnosis of Q fever is mainly based on several serological tests. These tests all need Coxiella organisms which are difficult and hazardous to culture and purify. An immunoproteomic study of C. burnetii Xinqiao strain isolated in China was conducted with the sera from experimentally infected BALB/c mice and Q fever patients. Twenty of whole proteins of Xinqiao recognized by the infection sera were identified by mass spectrometry. Nineteen of the 20 proteins were successfully expressed in Escherichia coli and used to fabricate a microarray which was probed with Q fever patient sera. As a result, GroEL, YbgF, RplL, Mip, OmpH, Com1, and Dnak were recognized as major seroreactive antigens. The major seroreactive proteins were fabricated in a small microarray and further analyzed with the sera of patients with rickettsial spotted fever, Legionella pneumonia or streptococcal pneumonia. In this analysis, these proteins showed fewer cross-reactions with the tested sera. Our results demonstrate that these 7 Coxiella proteins gave a modest sensitivity and specificity for recognizing of Q fever patient sera, suggesting that they are potential serodiagnostic markers for Q fever.

This paper investigates the balancing problem between popularity bias and the long-tail distribution in bundle recommendation systems. Existing approaches often focus exclusively on recommending popular items or solely on recommending long-tail items. Such approaches fail to effectively balance their respective impacts, resulting in insufficient diversity in system recommendations. To address this issue, we propose the GAMNBRec model. The model introduces the graph-adaptive mixed normalization (AdaMixNorm) method, which dynamically adjusts normalization strategies based on interactive graph structures. This strategy balances recommendation performance between popular and long-tail items. It enhances the recommendation quality of long-tail items through adaptive normalization adjustments, while simultaneously preserving the quality of popular item recommendations. In addition, GAMNBRec incorporates a residual-enhanced dynamic feature fusion mechanism to prevent feature loss for long-tail items in deep networks. It also integrates a Softmax-weighted BPR contrastive loss, which dynamically adjusts the importance of negative samples and thereby improves model training effectiveness. Experiments on the NetEase, Youshu, and iFashion datasets demonstrate that GAMNBRec significantly outperforms existing state-of-the-art methods in both Recall and NDCG metrics. These results validate the effectiveness and innovation of GAMNBRec in balancing recommendations for popular items and long-tail items.

Rickettsia heilongjiangensis, the agent of Far-Eastern spotted fever (FESF), is an obligate intracellular bacterium. The surface-exposed proteins (SEPs) of rickettsiae are involved in rickettsial adherence to and invasion of host cells, intracellular bacterial growth, and/or interaction with immune cells. They are also potential molecular candidates for the development of diagnostic reagents and vaccines against rickettsiosis. R. heilongjiangensis SEPs were identified by biotin-streptavidin affinity purification and 2D electrophoreses coupled with ESI-MS/MS. Recombinant SEPs were probed with various sera to analyze their serological characteristics using a protein microarray and an enzyme-linked immune sorbent assay (ELISA). Twenty-five SEPs were identified, most of which were predicted to reside on the surface of R. heilongjiangensis cells. Bioinformatics analysis suggests that these proteins could be involved in bacterial pathogenesis. Eleven of the 25 SEPs were recognized as major seroreactive antigens by sera from R. heilongjiangensis-infected mice and FESF patients. Among the major seroreactive SEPs, microarray assays and/or ELISAs revealed that GroEL, OmpA-2, OmpB-3, PrsA, RplY, RpsB, SurA and YbgF had modest sensitivity and specificity for recognizing R. heilongjiangensis infection and/or spotted fever. Many of the SEPs identified herein have potentially important roles in R. heilongjiangensis pathogenicity. Some of them have potential as serodiagnostic antigens or as subunit vaccine antigens against the disease.

Coxiella burnetii is the etiological agent of Q fever. To identify its major seroreactive proteins, a subgenomic protein array was developed. A total of 101 assumed virulence-associated recombinant proteins of C. burnetii were probed with sera from mice experimentally infected with C. burnetii and sera from Q fever patients. Sixteen proteins were recognized as major seroreactive antigens by the mouse sera. Seven of these 16 proteins reacted positively with at least 45% of Q fever patient sera. Notably, HspB had the highest fluorescence intensity value and positive frequency of all the proteins on the array when probed with both Q fever patient sera and mouse sera. These results suggest that these seven major seroreactive proteins, particularly HspB, are potential serodiagnostic and subunit vaccine antigens of Q fever.

Background Coxiella burnetii is an obligate intracellular bacterium and the etiologic agent of Q fever; both coxiella outer membrane protein 1 (Com1) and heat shock protein B (HspB) are its major immunodominant antigens. It is not clear whether Com1 and HspB have the ability to mount immune responses against C. burnetii infection. Results The recombinant proteins Com1 and HspB were applied to pulse human monocyte-derived dendritic cells (HMDCs), and the pulsed HMDCs were used to stimulate isogenic T cells. Com1-pulsed HMDCs expressed substantially higher levels of surface molecules (CD83, CD40, CD80, CD86, CD54, and CD58) and a higher level of interleukin-12 than HspB-pulsed HMDCs. Moreover, Com1-pulsed HMDCs induced high-level proliferation and activation of CD4 + and CD8 + cells, which expressed high levels of T-cell activation marker CD69 and inflammatory cytokines IFN-γ and TNF-α. In contrast, HspB-pulsed HMDCs were unable to induce efficient T-cell proliferation and activation. Conclusions Our results demonstrate that Com1-pulsed HMDCs are able to induce efficient T-cell proliferation and drive T cells toward Th1 and Tc1 polarization; however, HspB-pulsed HMDCs are unable to do so. Unlike HspB, Com1 is a protective antigen, which was demonstrated by the adoptive transfer of Com1-pulsed bone marrow dendritic cells into naive BALB/c mice.

Visible-thermal person re-identification (VT-ReID) is a challenging pedestrian retrieval problem in the field of security. Due to the intra-modality variations and cross-modality discrepancy caused by different spectrums, it is difficult to extract discriminative features. Existing works are devoted to projecting different-modality features into a shared space, which has weak discriminability and ignores the contextual relationship. In this paper, a novel dual-level information transfer framework is proposed to reduce the modality discrepancy in image level and feature level for VT-ReID. An auxiliary mix-modality is proposed and a mix-visible-thermal (MVT) learning strategy is built to reduce the discrepancy in image level. Firstly, the mix-modality is generated by a mixup scheme which alleviates the direct transfer. Secondly, under the MVT framework, we use ID loss and hetero center triplet loss to guide feature extraction for visible, thermal, and mixed modalities on a one-stream Network. To enhance the robustness of feature extraction, we introduce a graph information transfer module to transfer information across intra-modality and inter-modality in feature level. We build the agent node for modality by using the modality center, where the agent node aggregates the information of all samples in one modality, and then the information from one modality is transmitted to other modalities through the agent nodes. Extensive experimental results on SYSU-MM01 and RegDB datasets show that our method achieves excellent performance.

Target detection in aerial images by high altitude unmanned aerial vehicles is one of the hot research topics. A detection that is efficient and accurate is of a high value in the military and civilian fields. However, due to the irregular distribution of targets and their various scales shown in aerial images, it is difficult for existing backbone networks to effectively extract target features based on deep learning. To address these problems, in the paper, an efficient backbone network called AerNet is proposed, to which a local feature enhancement module (LFEM) is added to fully extract discriminative features of aerial targets via multi-scale convolutional layers. The network AerNet consists of 55 layers, which can keep high spatial resolution in deeper layers and maintain a large receptive field. Experimental results show that our AerNet has achieved a satisfying detection performance on the DOTA benchmark.

The domain adaptive person re-identification (Re-ID) has be more popular among researchers. Because it can save a lot of resources as it only exploits the source domain knowledge and does not need the complex annotation efforts in target domain. It aims to extend a model trained on a labeled dataset to another dataset which is unlabeled. Many works reduce feature distribution gap between two different datasets to solve the problem. However, these works ignore the problem which is the variations within an unlabeled dataset. In the paper, we propose a domain adaptive person Re-ID framework based on multiple granularity network and dynamic label (MGDL). Specifically, we send the images of two different datasets into multiple granularity network at the same time for joint training to reduce feature distribution gap which is between the two different datasets. The network is trained by two different kinds of pseudo labels, namely, conservative label and radical label. The two kinds of pseudo labels are used to alternating pull and push the feature distribution in the target domain to reduce the variations within an unlabeled dataset. Experiments have shown that the MGDL achieves competitive performance in person Re-ID which is under the cross-domain setting.

We present a method to recognize gestures made by Chinese traffic police based on the static and dynamic descriptor fusion for driver assistance systems and intelligent vehicles. Gesture recognition is made possible by combining the extracted static and dynamic features. First, the point cloud data of human upper body in each frame of input video is obtained to estimate the static descriptor with 2.5D gesture model. Then, the dynamic descriptor is estimated by computing the motion history image of the input RGB video sequence. Finally, the above two descriptors are fused and the mean structural similarity index is used to recognize the gestures made by Chinese traffic police. A comparative study and qualitative evaluation are proposed with other gesture recognition methods, which demonstrate that better recognition results can be obtained using the proposed method on a number of video sequences.