Search

Graph representation learning aims to embed each vertex in a graph into a low-dimensional vector space. Existing graph representation learning methods can be classified into two categories: generative models that learn the underlying connectivity distribution in a graph, and discriminative models that predict the probability of edge between a pair of vertices. In this paper, we propose GraphGAN , an innovative graph representation learning framework unifying the above two classes of methods, in which the generative and the discriminative model play a game-theoretical minimax game. Specifically, for a given vertex, the generative model tries to fit its underlying true connectivity distribution over all other vertices and produces “fake” samples to fool the discriminative model, while the discriminative model tries to detect whether the sampled vertex is from ground truth or generated by the generative model. With the competition between these two models, both of them can alternately and iteratively boost their performance. Moreover, we propose a novel graph softmax as the implementation of the generative model to overcome the limitations of traditional softmax function, which can be proven satisfying desirable properties of normalization , graph structure awareness , and computational efficiency . Through extensive experiments on real-world datasets, we demonstrate that GraphGAN achieves substantial gains in a variety of applications, including graph reconstruction, link prediction, node classification, recommendation, and visualization, over state-of-the-art baselines.

Reduced graphene oxide (rGO) fiber as a carbon-based fiber sensor has aroused widespread interest in the field of gas sensing. However, the low response value and poor flexibility of the rGO fiber sensor severely limit its application in the field of flexible wearable electronics. In this paper, a flexible and wearable nylon fiber sensor modified by rGO and ZnO quantum dots (QDs) is proposed for wide-range NO2 gas detection at room temperature. The response value of the nylon fiber sensor to 100 ppm NO2 gas is as high as 0.4958, and the response time and recovery time are 216.2 s and 667.9 s, respectively. The relationship between the sensor’s response value and the NO2 concentration value is linear in the range of 20–100 ppm, and the fitting coefficient is 0.998. In addition, the test results show that the sensor also has good repeatability, flexibility, and selectivity. Moreover, an early warning module was also designed and is proposed in this paper to realize the over-limit monitoring of NO2 gas, and the flexible sensor was embedded in a mask, demonstrating its great application potential and value in the field of wearable electronics.

Radiotherapy (RT) represents one of the major treatment methods for cancers. However, many studies have observed that in descendant surviving tumor cells, sublethal irradiation can promote metastatic ability, which is closely related to the tumor microenvironment. We therefore investigated the functions and mechanisms of sublethal irradiated liver nonparenchymal cells (NPCs) in hepatocellular carcinoma (HCC). In this study, primary rat NPCs and McA‐RH7777 hepatoma cells were irradiated with 6 Gy X‐ray. Conditioned media (CM) from nonirradiated (SnonR), irradiated (SR), or irradiated plus radiosensitizer celecoxib‐treated (S[R + D]) NPCs were collected and added to sublethal irradiated McA‐RH7777 cells. We showed that CM from sublethal irradiated NPCs significantly promoted the migration and invasion ability of sublethal irradiated McA‐RH7777 cells, which was reversed by celecoxib. The differentially expressed genes in differently treated McA‐RH7777 cells were enriched mostly in the AMP‐activated protein kinase/mammalian target of rapamycin (AMPK/mTOR) signaling pathway. SR increased the migration and invasion ability of HCC cells by inhibiting AMPK/mTOR signaling, which was enhanced by the AMPK inhibitor compound C and blocked by the AMPK activator GSK‐621. Analyses of HCC tissues after neoadjuvant radiotherapy confirmed the effects of radiation on the AMPK/mTOR pathway. Cytokine antibody arrays and further functional investigations showed that matrix metalloproteinase‐8 (MMP‐8) partly mediates the promotion effects of SR on the migration and invasion ability of HCC cells by regulating AMPK/mTOR signaling. In summary, our data indicate that MMP‐8 secreted by irradiated NPCs enhanced the migration and invasion of HCC by regulating AMPK/mTOR signaling, revealing a novel mechanism mediating sublethal irradiation–induced HCC metastasis at the level of the tumor microenvironment., In this work, we observed that conditioned media from sublethally irradiated liver nonparenchymal cells (NPCs) promoted the migration and invasion ability of sublethally irradiated McA‐RH7777 hepatoma cells. Further mechanistic studies revealed that matrix metalloproteinase‐8 (MMP‐8) secreted by NPCs enhanced the migration and invasion of hepatocellular carcinoma (HCC) by regulating the AMPK/mTOR signaling pathway. These findings might offer a new strategy for improving the therapeutic effect of radiotherapy.

With the advancement and development of science and technology, the detection role of small-scale robot has become increasingly prominent, especially for environmental detection in hazardous situations such as explosion and fire. In this study, to realise the detection of carbon monoxide gas in hazardous locations, a flexible carbon monoxide sensor consisting of polyimide, interdigital electrode, and composite film is proposed and designed, which has the advantages of simple structure, high sensitivity, good flexibility, and low cost. For the interdigital electrode, the distance between the electrodes and the width of the electrode is both 50 μm, and the sizes of the polyimide substrate and the electrode area are 1.5 × 4 cm2 and 1 × 1 cm2, respectively. The composite film is made by mixing single-walled carbon nanotubes and platinum powders. The test results show that the sensitivity of the sensor is 0.0201 Ω/ppm in the range of 0–100 ppm at room temperature with a linear fit of 94.63%.

Background Nasopharyngeal carcinoma (NPC), arising from nasopharynx epithelium, is a rare type of malignant carcinoma that has a specific geographical distribution and a high risk of distant metastases. For most of the diagnosed NPC patients, the total survival rate decreased significantly due to the high local recurrence rate and metastasis rate. Concurrent chemoradiotherapy (CCRT), as routine therapy strategy of NPC, usually accompanies with high-dosage cytotoxic agents and serious toxic side reaction. Therefore, there is an urgent need for update the existing therapy strategies. In this study, we sought to investigate the effects of a combined therapy strategy, erlotinib combined with cisplatin and radiotherapy, on biological characteristics of NPC CNE2 cells and the potential reasons. Methods CNE2 cells in logarithmic phase seeding in 96-well plates received concentration gradients of erlotinib (at 0, 10, 20, 40, 80, 160, 320 mmol/L) or cisplatin (at 0, 0.25, 0.5, 1, 2, 4, 8 mg/L), in order to obtain the optimal working concentration of erlotinib and cisplatin via 3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. Then, cells were divided into control group and four treatment groups (Group I-IV). All treatment groups received irradiation of 4 Gy, moreover, Group II-IV respectively received erlotinib, cisplatin and erlotinib plus cisplatin at optimal working concentration. After 24 and 48 h of irradiation, growth inhibition rate was determined; invasion ability and migration ability was respectively detected by Boyden's chamber assay and cell scratch test; flow cytometry was performed for determining apoptosis rate and cell cycle distribution; the expression levels of epidermal growth factor receptor (EGFR) signal pathway proteins were semi-quantitatively analyzed by Western-blot. Results Compared with Group I, all the other treatment groups showed better inhibition effect on cell viability, invasion and migration ability and higher apoptosis rate, while Group IV showed the strongest growth inhibition effect and highest apoptosis rate. In addition, EGFR signal pathway proteins of Group IV showed the lowest expression level. Conclusions In combined therapy with radiotherapy/chemotherapy, erlotinib could enhance radiotherapy/chemotherapy sensitivity, probably because it could suppress DNA damage repair after radiotherapy/chemotherapy, thus weakening radiotherapy/chemotherapy resistance of tumor cells.

Background This study aimed to compare the efficacy and toxicity of raltitrexed (Saiweijian®) plus cisplatin (SP regimen) and 5‐fluorouracil plus cisplatin (FP regimen) as concurrent chemoradiotherapy (CCRT) in patients with locally advanced nasopharyngeal carcinoma (LA‐NPC). Methods Eligible patients (N = 135) were allocated randomly in a ratio of 1:1 to receive CCRT with either SP or FP. At least 2 cycles of chemotherapy was administrated during radiotherapy. Progression free survival (PFS) was primary endpoint. Secondary endpoints included overall survival (OS), loco‐regional relapse free survival (LRRFS), distant metastasis free survival (DMFS) and toxicity. Results In this study, 68 patients received SP as CCRT, and 67 received FP. Objective responses were noted in 97.1% of the patients in the SP group and in 97.0% of the patients in the FP group (P = 1.00). At the end of a median 36 months follow‐up period, the estimated 3‐year PFS rates were 70.1% for SP and 66.6% for FP, respectively. The 3‐year LRRFS, DMFS and OS rates were 88.9%, 74.7% and 84.0%, respectively, for the SP group, and 92.3%, 71.0% and 73.7%, respectively, for the FP group. Overall, there was no difference between treatment groups with regard to response or survival. The most frequent acute toxicities monitored in both groups were bone marrow suppression, gastrointestinal side effects and oral mucositis (OM). The overall incidence of grade 3‐4 OM in the FP group (47.8%) was higher than in the SP group (11.8%). However, the incidence of other adverse effects observed in both groups was similar (P > .05). Conclusions These data indicate that SP and FP therapies have similar efficacy in treating LA‐NPC. The SP regimen showed a tolerable safety profile along with a lower frequency of severe OM and therefore, an improved life quality. In conclusion, SP was a well tolerated, effective, regimen for LA‐NPC treatment., The incidence of other adverse effects seen in both groups was similar (P > .05).The efficacy of both regimens i.e. SP and FP are found similar. The SP regimen has tolerable safety profile, with a lower incidence of severe OM and consequently, an improved quality of life.

Real-time assessment of therapeutic response in patients with advanced lung cancer presents a major challenge throughout the treatment process. Currently, computed tomography imaging is often used; however, it is radiation-based and hysteretic and is not suitable for repeated use as a real-time assessment. Blood biomarkers represent a novel solution for assessing therapeutic response in patients with advanced lung cancer. In the present study, the efficacy of a methylation marker [methylated prostaglandin E receptor 4 (mPTGER4)] and four protein markers [carcinoma antigen 125 (CA125), carcinoembryonic antigen (CEA), cytokeratin 19-fragments (cyfra21-1) and neuron-specific enolase (NSE)] were simultaneously evaluated to determine their potential in facilitating therapeutic response monitoring as well as their prognostic values in patients with stage IV lung cancer. The results indicated that, following treatment, the blood levels of methylated PTGER4 and NSE had significantly decreased, and mPRGER4, CA125, CEA and NSE exhibited a significant decrease in percentage level. Since mPTGER4 exhibited a higher rate of positive detection prior to therapy, and a greater response of sensitivity to therapy compared to the protein markers, it may represent an improved marker for the monitoring of therapeutic response. The efficacy of the markers in predicting the overall survival (OS) rate of patients with stage IV lung cancer was also assessed. Results from the follow-up of patients (up to 891 days) revealed that the blood levels of mPTGER4, CA125 and NSE before treatment were able to predict overall survival (OS) rate. Additionally, the percentage change in expression levels of CA125, CEA and NSE was also able to predict the OS rate. In conclusion, the present results indicate that mPTGER4 represents an improved biomarker for monitoring therapeutic efficacy compared with CA125, CEA, Cyfra21-1 and NSE. In predicting the long-term survival of patients with stage IV lung cancer; however, the pre-treatment levels of mPTGER4, CA125 and NSE and the percentage changes of CA125, CEA and NSE may be used as the markers.

Cerebral revascularization strategies through extracranial to intracranial bypass have been adopted in the management of complex intracranial aneurysms. The internal maxillary artery used as a donor in a bypass is an effective method. At present, there are few quantitative analyses of cerebral blood flow perfusion. The main focus of this study was to evaluate the effectiveness of blood perfusion after bypass grafting.From April 2015 to December 2017, 19 patients who underwent internal maxillary artery radial artery middle cerebral artery bypass surgery with unobstructed bypass vessels were selected. Cerebral blood flow perfusion before and after bypass surgery was quantitatively evaluated by computed tomography perfusion imaging. The cerebral blood perfusion in the region of interest was measured by computed tomography perfusion.The aneurysms were excised after trapping in 2 cases with mass effects and neural compression. Proximal occlusion of the parent artery was performed in 9 cases of fusiform or giant dissecting aneurysms. Trapping was performed after bypass surgery in 8 cases. Within 3 months after surgery, 17 patients had good outcomes. After the hypothesis test, there was a significant difference between the preoperative △cerebral blood volume and postoperative △cerebral blood volume in the anterior area of the semioval center cross section (P = 0.001 0.05).The internal maxillary artery as a bypass donor is an effective method that can provide sufficient intracranial blood perfusion, and there is usually no cerebral ischemia in the surrounding area.

Flexible sensors used to detect NO

Knowledge graph completion (KGC) has become a focus of attention across deep learning community owing to its excellent contribution to numerous downstream tasks. Although recently have witnessed a surge of work on KGC, they are still insufficient to accurately capture complex relations, since they adopt the single and static representations. In this work, we propose a novel Disentangled Knowledge Graph Attention Network (DisenKGAT) for KGC, which leverages both micro-disentanglement and macro-disentanglement to exploit representations behind Knowledge graphs (KGs). To achieve micro-disentanglement, we put forward a novel relation-aware aggregation to learn diverse component representation. For macro-disentanglement, we leverage mutual information as a regularization to enhance independence. With the assistance of disentanglement, our model is able to generate adaptive representations in terms of the given scenario. Besides, our work has strong robustness and flexibility to adapt to various score functions. Extensive experiments on public benchmark datasets have been conducted to validate the superiority of DisenKGAT over existing methods in terms of both accuracy and explainability., CIKM2021

The influence of surface roughness on adhesion hysteresis of Cu thin films with different thickness was studied. The Cu thin films were fabricated using dc magnetron sputtering technology and their surface adhesion forces were measured using force-displacement curves of atomic force microscopy. The experiment results show that the unload curve exhibits obvious adhesion hysteresis and that the roughness and adhesion force have opposite change tendency. The theoretical analysis on the influence of thin film roughness to the adhesion hysteresis was carried out using the Johnson-Kendall-Roberts model. The results show that the load curve and unload curve constitute a closed curve and that the smoother the surface, the more obvious adhesion hysteresis, i.e. the rougher the surface, the less obvious adhesion hysteresis. The theoretical analysis results are in good agreement with the experimental results.

Objective To evaluate the impact of systematic nutrition management (SNM) on nutritional status, treatment-related toxicity, quality of life (QoL), response rates, and survival in patients with locally advanced nasopharyngeal carcinoma (LA-NPC) treated by radiotherapy (RT). Methods In this retrospective study, 56 patients with LA-NPC were selected as nutrition management group (NG) for SNM during RT till 1 month later. Another 56 patients with LA-NPC receiving RT without SNM as control group (CG) were identified from the hospital database and matched pairs with NG patients according to age, gender, stage, and body mass index (BMI) prior to RT. Results At 1 month after RT, the percentage of malnourished patients with BMI 0.05) for the 2-year progression-free survival and overall survival (84.2% versus 79.5% and 94.7% versus 92.3%, respectively.). Conclusion SNM for LA-NPC patients treated by RT resulted in better nutritional status, reduced treatment-related toxicity and improved QoL.

To address the sparsity and cold-start problem of collaborative filtering, researchers usually make use of side information, such as social networks or item attributes, to improve the performance of recommendation. In this article, we consider the knowledge graph (KG) as the source of side information. To address the limitations of existing embedding-based and path-based methods for KG-aware recommendation, we propose RippleNet , an end-to-end framework that naturally incorporates the KG into recommender systems. RippleNet has two versions: (1) The outward propagation version, which is analogous to the actual ripples on water, stimulates the propagation of user preferences over the set of knowledge entities by automatically and iteratively extending a user’s potential interests along links in the KG. The multiple “ripples” activated by a user’s historically clicked items are thus superposed to form the preference distribution of the user with respect to a candidate item. (2) The inward aggregation version aggregates and incorporates the neighborhood information biasedly when computing the representation of a given entity. The neighborhood can be extended to multiple hops away to model high-order proximity and capture users’ long-distance interests. In addition, we intuitively demonstrate how a KG assists with recommender systems in RippleNet, and we also find that RippleNet provides a new perspective of explainability for the recommended results in terms of the KG. Through extensive experiments on real-world datasets, we demonstrate that both versions of RippleNet achieve substantial gains in a variety of scenarios, including movie, book, and news recommendations, over several state-of-the-art baselines.

To solve the problem of near-field measurement of aeroengines a novel large-range high-precision Fabry–Perot interferometer (FPI) is developed, which is verified by a high-temperature experiment. Based on the principle of FPI wavelength drift and frequency spectrum drift, a double-beam-interference FPI is designed. Through an analysis of the optical path difference between the two beams, the conclusion that the spectrum drifts to the long-wave direction with the increase of temperature is obtained. Moreover, through frequency spectrum analysis, the measurement error caused by the distortion of the spectrum is avoided, and it is found that the increase in temperature will cause the change in frequency spectrum. The temperature sensitivity of the glass-type FPI is only 0.0011 nm °C−1. A ceramic material with a higher thermal expansion coefficient is selected as the collimating tube to make the sensitivity of the temperature sensor as high as 0.691 nm °C−1 from normal temperature to 100 °C. To meet the needs of a wide range of measurements from room temperature to 1000 °C the frequency drift method is utilized. A field experiment is carried out on the ceramic FPI at the tail spray of the aeroengine simulation platform. The temperature response test from normal temperature to 1000 °C is completed, and the accuracy of the sensor reached 0.043%. In this study, the principle, design, production, and testing of optical fiber sensors are carried out. The developed optical fiber sensor has significance for high temperature monitoring.

Recently, the retrieval models based on dense representations have been gradually applied in the first stage of the document retrieval tasks, showing better performance than traditional sparse vector space models. To obtain high efficiency, the basic structure of these models is Bi-encoder in most cases. However, this simple structure may cause serious information loss during the encoding of documents since the queries are agnostic. To address this problem, we design a method to mimic the queries on each of the documents by an iterative clustering process and represent the documents by multiple pseudo queries (i.e., the cluster centroids). To boost the retrieval process using approximate nearest neighbor search library, we also optimize the matching function with a two-step score calculation procedure. Experimental results on several popular ranking and QA datasets show that our model can achieve state-of-the-art results., 11 pages, 2 figures, Accepted by ACL 2021

Sentiment analysis has attracted increasing attention in e-commerce. The sentiment polarities underlying user reviews are of great value for business intelligence. Aspect category sentiment analysis (ACSA) and review rating prediction (RP) are two essential tasks to detect the fine-to-coarse sentiment polarities. %Considering the sentiment of the aspects(ACSA) and the overall review rating(RP) simultaneously has the potential to improve the overall performance. ACSA and RP are highly correlated and usually employed jointly in real-world e-commerce scenarios. While most public datasets are constructed for ACSA and RP separately, which may limit the further exploitation of both tasks. To address the problem and advance related researches, we present a large-scale Chinese restaurant review dataset \textbf{ASAP} including $46,730$ genuine reviews from a leading online-to-offline (O2O) e-commerce platform in China. Besides a $5$-star scale rating, each review is manually annotated according to its sentiment polarities towards $18$ pre-defined aspect categories. We hope the release of the dataset could shed some light on the fields of sentiment analysis. Moreover, we propose an intuitive yet effective joint model for ACSA and RP. Experimental results demonstrate that the joint model outperforms state-of-the-art baselines on both tasks., 11 Pages, 5 Figures, Accepted at NAACL 2021

The optical fiber temperature and refractive index sensor combined with the hollow needle structure for medical treatment can promote the standardization of traditional acupuncture techniques and improve the accuracy of body fluid analysis. A double-parameter sensor based on fiber Bragg grating (FBG) is developed in this paper. The sensor materials are selected through X-ray diffraction (XRD) analysis, and the sensor sensing principle is theoretically analyzed and simulated. Through femtosecond laser writing pure silica fiber, a high temperature resistant wavelength type FBG temperature sensor is obtained, and the FBG is corroded by hydrofluoric acid (HF) to realize a high-sensitivity intensity-type refractive index sensor. Because the light has dual characteristics of energy and wavelength, the sensor can realize simultaneous dual-parameter sensing. The light from the lead-in optical fiber is transmitted to the sensor and affected by temperature and refractive-index, then, the reflection peak is reflected back to the lead-out fiber by the FBG. The high temperature response and the refractive index response of the sensor were measured in the laboratory, and the high temperature characteristics of the sensor were verified in the accredited institute. It is demonstrated that the proposed sensor can achieve temperature sensing up to 1150 °C with the sensitivity of 0.0134 nm/°C, and refractive sensing over a refractive range of 1.333 to 1.4027 with the sensitivity of −49.044 dBm/RIU. The sensor features the advantages of two-parameter measurement, compact structure, and wide temperature range, and it exhibits great potential in acupuncture treatment.

Background Despite the capability of emergency surgery to reduce the mortality of severe spontaneous intracranial hemorrhage (SSICH) patients, the effect and safety of surgical treatment for severe spontaneous intracranial hemorrhage (SSICH) patients receiving long-term oral antiplatelet treatment (LOAPT) remains unclear. In consideration of this, the cohort study is aimed at figuring out the effect and safety of emergency surgery for SSICH patients on LOAPT. Methods As a multicenter and prospective cohort study, it will be conducted across 7 representative clinical centers. Starting in September 2019, the observation is scheduled to be completed by December 2022, with a total of 450 SSICH patients recruited. The information on clinical, radiological, and laboratory practices will be recorded objectively. All of the patients will be monitored until death or 6 months after the occurrence of primary hemorrhage. Results In this study, two comparative cohorts and an observational cohort will be set up. The primary outcome is the effect of emergency surgery, which is subject to assessment using the total mortality and comparison in the survival rate of SSICH patients on LOAPT between surgical treatment and conservative treatment. The second outcome is the safety of surgery, with the postoperative hemorrhagic complication which is compared between the operated SSICH patients on and not on LOAPT. Based on the observation of the characteristics and outcome of SSICH patients on LOAPT, the ischemic events after discontinuing LOAPT will be further addressed, and the coagulation function assessment system for operated SSICH patients on LOAPT will be established. Conclusions In this study, we will investigate the effect and safety of emergency surgery for SSICH patients on LOAPT, which will provide an evidence for management in the future. Ethics and dissemination The research protocol and informed consent in this study were approved by the Institutional Review Board of Beijing Tiantan Hospital (KY2019-096-02). The results of this study are expected to be disseminated in peer-reviewed journals in 2023. Trial registration Name: Effect and safety of surgical intervention for severe spontaneous intracerebral hemorrhage patients on long-term oral antiplatelet treatment. ChiCTR1900024406. Date of registration is July 10, 2019.

In this paper, we focus on the problem of keyword and document matching by considering different relevance levels. In our recommendation system, different people follow different hot keywords with interest. We need to attach documents to each keyword and then distribute the documents to people who follow these keywords. The ideal documents should have the same topic with the keyword, which we call topic-aware relevance. In other words, topic-aware relevance documents are better than partially-relevance ones in this application. However, previous tasks never define topic-aware relevance clearly. To tackle this problem, we define a three-level relevance in keyword-document matching task: topic-aware relevance, partially-relevance and irrelevance. To capture the relevance between the short keyword and the document at above-mentioned three levels, we should not only combine the latent topic of the document with its deep neural representation, but also model complex interactions between the keyword and the document. To this end, we propose a Two-stage Interaction and Topic-Aware text matching model (TITA). In terms of “topic-aware”, we introduce neural topic model to analyze the topic of the document and then use it to further encode the document. In terms of “two-stage interaction”, we propose two successive stages to model complex interactions between the keyword and the document. Extensive experiments reveal that TITA outperforms other well-designed baselines and shows excellent performance in our recommendation system.

Recommender system usually suffers from severe popularity bias -- the collected interaction data usually exhibits quite imbalanced or even long-tailed distribution over items. Such skewed distribution may result from the users' conformity to the group, which deviates from reflecting users' true preference. Existing efforts for tackling this issue mainly focus on completely eliminating popularity bias. However, we argue that not all popularity bias is evil. Popularity bias not only results from conformity but also item quality, which is usually ignored by existing methods. Some items exhibit higher popularity as they have intrinsic better property. Blindly removing the popularity bias would lose such important signal, and further deteriorate model performance. To sufficiently exploit such important information for recommendation, it is essential to disentangle the benign popularity bias caused by item quality from the harmful popularity bias caused by conformity. Although important, it is quite challenging as we lack an explicit signal to differentiate the two factors of popularity bias. In this paper, we propose to leverage temporal information as the two factors exhibit quite different patterns along the time: item quality revealing item inherent property is stable and static while conformity that depends on items' recent clicks is highly time-sensitive. Correspondingly, we further propose a novel Time-aware DisEntangled framework (TIDE), where a click is generated from three components namely the static item quality, the dynamic conformity effect, as well as the user-item matching score returned by any recommendation model. Lastly, we conduct interventional inference such that the recommendation can benefit from the benign popularity bias while circumvent the harmful one. Extensive experiments on three real-world datasets demonstrated the effectiveness of TIDE.

Point-of-Interest (POI) recommendation helps users find their interested places to visit based on the time and user location. Unlike traditional recommendation tasks, POI recommendation is personalized, spatial-aware, and temporally dependent. Although many previous works have tried modeling spatial and temporal characteristics, most of them suffer from the following two limitations: For the spatial aspect, existing works only consider the user-POI distance or POI-POI distance. However, we find that a user prefers different regions at different times, which is known as user-region periodic pattern. For the temporal aspect, most works treat user and time as two independent factors. However, different users may prefer the same POI in different time periods, which is known as user-POI periodic pattern. To address the limitation of existing works, we propose a novel Spatial-Temporal aware Graph Convolutional Neural Network (STGCN) for POI recommendation. Specifically, we first design a user record multigraph to fuse all the context information into a unified graph. Then, we propose a time-based neighborhood sampling algorithm and take advantage of the flexible propagation mechanism of GCNs to learn the representations of each node at a specific time. Furthermore, multiple scoring functions are proposed to exploit user-region periodic pattern and user-POI periodic pattern, respectively. We also develop a time smoothing strategy to alleviate the data sparsity problem. Extensive experiments are conducted on two real-world datasets, and the experimental results demonstrate the effectiveness of our method.

Recently, conversational recommender system (CRS) has become an emerging and practical research topic. Most of the existing CRS methods focus on learning effective preference representations for users from conversation data alone. While, we take a new perspective to leverage historical interaction data for improving CRS. For this purpose, we propose a novel pre-training approach to integrating both item-based preference sequence (from historical interaction data) and attribute-based preference sequence (from conversation data) via pre-training methods. We carefully design two pre-training tasks to enhance information fusion between item- and attribute-based preference. To improve the learning performance, we further develop an effective negative sample generator which can produce high-quality negative samples. Experiment results on two real-world datasets have demonstrated the effectiveness of our approach for improving CRS., Comment: Accepted as CIKM short paper

As a concise form of user reviews, tips have unique advantages to explain the search results, assist users' decision making, and further improve user experience in vertical search scenarios. Existing work on tip generation does not take query into consideration, which limits the impact of tips in search scenarios. To address this issue, this paper proposes a query-aware tip generation framework, integrating query information into encoding and subsequent decoding processes. Two specific adaptations of Transformer and Recurrent Neural Network (RNN) are proposed. For Transformer, the query impact is incorporated into the self-attention computation of both the encoder and the decoder. As for RNN, the query-aware encoder adopts a selective network to distill query-relevant information from the review, while the query-aware decoder integrates the query information into the attention computation during decoding. The framework consistently outperforms the competing methods on both public and real-world industrial datasets. Last but not least, online deployment experiments on Dianping demonstrate the advantage of the proposed framework for tip generation as well as its online business values., Accepted By CIKM 2020 Applied Research Track

To provide more accurate personalized product search (PPS) results, it is compulsory to go beyond modeling user-query-item interaction. Graph embedding techniques open the potential to integrate node information and topological structure information. Existing graph embedding enhanced PPS methods are mostly based on entity-relation-entity graph learning. In this work, we propose to consider structural relationship in users' product search scenario with graph embedding by latent representation learning. We argue that explicitly modeling the structural relationship in graph embedding is essential for more accurate PPS results. We propose a novel method, Graph embedding based Structural Relationship Representation Learning (GraphSRRL), which explicitly models the structural relationship in users-queries-products interaction. It combines three key conjunctive graph patterns to learn graph embedding for better PPS. In addition, GraphSRRL facilitates the learning of affinities between users (resp. queries or products) in the designed geometric operation in low-dimensional latent space. We conduct extensive experiments on four datasets to evaluate GraphSRRL for PPS. Experimental results show that GraphSRRL outperforms the state-of-the-art algorithm on real-world search datasets by at least 50.7% in term of Hit@10 and 48.7% in terms of NDCG@10.

Document retrieval (DR) is a crucial task in NLP. Recently, the pre-trained BERT-like language models have achieved remarkable success, obtaining a state-of-the-art result in DR. In this paper, we come up with a new BERT-based ranking model for DR task, named TABLE. In the pre-training stage of TABLE, we present a domain-adaptive strategy. More essentially, in the fine-tuning stage, we develop a two-phase task-adaptive process, i.e., type-adaptive pointwise fine-tuning and listwise fine-tuning. In the type-adaptive pointwise fine-tuning phase, the model can learn different matching patterns regarding different query types. In the listwise fine-tuning phase, the model matches documents with regard to a given query in a listwise fashion. This task-adaptive process makes the model more robust. In addition, a simple but effective exact matching feature is introduced in fine-tuning, which can effectively compute matching of out-of-vocabulary (OOV) words between a query and a document. As far as we know, we are the first who propose a listwise ranking model with BERT. This work can explore rich matching features between queries and documents. Therefore it substantially improves model performance in DR. Notably, our TABLE model shows excellent performance on the MS MARCO leaderboard.

Recommender systems have shown great potential to solve the information explosion problem and enhance user experience in various online applications. To tackle data sparsity and cold start problems in recommender systems, researchers propose knowledge graphs (KGs) based recommendations by leveraging valuable external knowledge as auxiliary information. However, most of these works ignore the variety of data types (e.g., texts and images) in multi-modal knowledge graphs (MMKGs). In this paper, we propose Multi-modal Knowledge Graph Attention Network (MKGAT) to better enhance recommender systems by leveraging multi-modal knowledge. Specifically, we propose a multi-modal graph attention technique to conduct information propagation over MMKGs, and then use the resulting aggregated embedding representation for recommendation. To the best of our knowledge, this is the first work that incorporates multi-modal knowledge graph into recommender systems. We conduct extensive experiments on two real datasets from different domains, results of which demonstrate that our model MKGAT can successfully employ MMKGs to improve the quality of recommendation system.

As the perfect combination of fiber optics and nanotechnology, micro-nano fiber is one of the frontier research directions in fiber optics and micro-nano photonics developed in recent years. In this paper, a high-sensitivity temperature sensor is proposed and realized based on the micro-nano fiber structure. The degree of linearity and sensitivity of the sensor are 98.6% and -0.0121 dBm / °C, respectively. Besides, the preparation process and influencing factors of micro-nano fiber are also introduced and discussed in detail. The configuration features the advantages of high sensitivity, simple structure, low transmission loss, low cost, and easy fabrication, making it a good candidate for the field of temperature measurement.

PurposeGastric carcinoma (GC) is one of the most common cause of tumor-related death. Chemotherapy resistance usually occurs, leading to cancer relapse and poor survival of GC patients. To investigate the role of miRNAs in chemotherapy resistance for GC patients, we conducted an integrated analysis of miRNA expression and survival information using data obtained from the TCGA project. Materials and methodsGenome-wide screening of chemotherapy response-specific miRNAs was performed using Cox proportional hazards regression analyses for patients received chemotherapy or those had never received chemotherapy, respectively. CCK-8 assay was used for assessing cell viability. GSEA and single sample GSVA were performed for functional enrichment analysis. CIBERSORT cellular composition analysis was used to infer immune cell populations in GC tumors.ResultsA four-miRNA expression signature involving miR-130, -152, -199 and -200b was predicted as a specific indicator for GC chemoresistance (P = 0.00053; hazard ratio = 8.63), outperforming those clinicopathological factors. Functional experiments confirmed the roles of these signature miRNAs in regulation of chemotherapy response. Functional enrichment of these signature miRNAs and risk score revealed positive association with epithelial-mesenchymal transition (EMT), and negative association with cell cycle checkpoint and DNA damage response. Furthermore, the immune infiltration-miRNA functional network analysis revealed transformation from activated effector cells to resting immunosuppressive cells are preferred in GCs with adverse chemotherapy response.ConclusionsOur work identifies a four‑miRNA expression signature as a promising chemoresistance biomarker in gastric carcinoma, which provides novel insights into developing new strategies to overcome GC chemoresistance.

Tumor-associated macrophages (TAMs) are important components of the tumor microenvironment, which are characterized by pro-tumor M2 phenotype and correlate with poor survival of nasopharyngeal carcinoma (NPC). Heme oxygenase-1 (HO-1) plays a crucial role in macrophage polarization toward M2 phenotype, but its prognosis significance in NPC has been rarely determined. To gain insights into the HO-1 expression profile and to determine the clinical significance of HO-1 in NPC, we performed immunohistochemistry analyses in 126 NPC specimens. CD163, a highly specific marker of M2 macrophages, was used as a surrogate for the polarization state of TAMs. Our results showed that high expression of HO-1 and CD163 were detected in TAMs for 57.9% (73/126) and 61.9% (78/126) of the studied patients, and both of them were significantly associated with worse survival. Additionally, a significant correlation between the intensities of HO-1 and CD163 was identified, and HO-1 exhibited a superior ability in predicting survival compared with CD163. Our study revealed for the first time that overexpression of HO-1 characterized a poor-prognosis subtype in NPC. Individualized therapy targeting HO-1 might serve as a promising treatment modality for NPC.

Background: Persistent primitive hypoglossal artery (PPHA) is a rare remnant of the congenital carotid-basilar artery anastomosis, sometimes associated with subarachnoid hemorrhage (SAH), and aneurysms are the more common cause. However, the association of PPHA with perimesencephalic nonaneurysmal subarachnoid hemorrhage (PNSAH) is extremely rare. This report is the first to describe the occurrence of PNSAH concerning PPHA. Case description: A 53-year-old man suffered from subarachnoid hemorrhage with a sudden severe headache. An emergency brain computed tomography (CT) disclosed a perimesencephalic pattern of SAH. CT angiography (CTA) revealed a right PPHA entering the posterior fossa through the right enlarged hypoglossal canal. No source of the hemorrhage was identified in the imaging examinations, even with the performance of a 4-vessel angiography. Conclusion: We present a case of PPHA associated with PNSAH and review the literature. The main cause for the PPHA related to SAH is an aneurysm. The four vessels angiography helped rule out an aneurysm that is associated with a poorer prognosis and requires immediate active management. It is, therefore, essential for physicians to be aware of the increased incidence of aneurysms in PPHA to make the correct diagnosis. Keywords: Persistent primitive hypoglossal artery, Subarachnoid hemorrhage, Perimesencephalic nonaneurysmal subarachnoid hemorrhage, Carotid-basilar anastomoses, Intracranial aneurysm

Recently, significant progress has been made in sequential recommendation with deep learning. Existing neural sequential recommendation models usually rely on the item prediction loss to learn model parameters or data representations. However, the model trained with this loss is prone to suffer from data sparsity problem. Since it overemphasizes the final performance, the association or fusion between context data and sequence data has not been well captured and utilized for sequential recommendation. To tackle this problem, we propose the model S^3-Rec, which stands for Self-Supervised learning for Sequential Recommendation, based on the self-attentive neural architecture. The main idea of our approach is to utilize the intrinsic data correlation to derive self-supervision signals and enhance the data representations via pre-training methods for improving sequential recommendation. For our task, we devise four auxiliary self-supervised objectives to learn the correlations among attribute, item, subsequence, and sequence by utilizing the mutual information maximization (MIM) principle. MIM provides a unified way to characterize the correlation between different types of data, which is particularly suitable in our scenario. Extensive experiments conducted on six real-world datasets demonstrate the superiority of our proposed method over existing state-of-the-art methods, especially when only limited training data is available. Besides, we extend our self-supervised learning method to other recommendation models, which also improve their performance., Comment: Accepted as CIKM2020 long paper

Verifying fact on semi-structured evidence like tables requires the ability to encode structural information and perform symbolic reasoning. Pre-trained language models trained on natural language could not be directly applied to encode tables, because simply linearizing tables into sequences will lose the cell alignment information. To better utilize pre-trained transformers for table representation, we propose a Structure-Aware Transformer (SAT), which injects the table structural information into the mask of the self-attention layer. A method to combine symbolic and linguistic reasoning is also explored for this task. Our method outperforms baseline with 4.93% on TabFact, a large scale table verification dataset.

Location recommendation plays an essential role in helping people find attractive places. Though recent research has studied how to recommend locations with social and geographical information, few of them addressed the cold-start problem of new users. Because mobility records are often shared on social networks, semantic information can be leveraged to tackle this challenge. A typical method is to feed them into explicit-feedback-based content-aware collaborative filtering, but they require drawing negative samples for better learning performance, as users’ negative preference is not observable in human mobility. However, prior studies have empirically shown sampling-based methods do not perform well. To this end, we propose a scalable Implicit-feedback-based Content-aware Collaborative Filtering (ICCF) framework to incorporate semantic content and to steer clear of negative sampling. We then develop an efficient optimization algorithm, scaling linearly with data size and feature size, and quadratically with the dimension of latent space. We further establish its relationship with graph Laplacian regularized matrix factorization. Finally, we evaluate ICCF with a large-scale LBSN dataset in which users have profiles and textual content. The results show that ICCF outperforms several competing baselines, and that user information is not only effective for improving recommendations but also coping with cold-start scenarios.

Illuminating and viewing geometries may strongly affect the color appearance of metallic coatings, which can be characterized accurately by bidirectional reflectance distribution function (BRDF) measurement devices. However, such devices with hundreds or even thousands of different geometries are usually expensive and complex. Accordingly, two modified models were developed in this study, based on the distribution of aluminum-flake pigments in the coatings, to, respectively, estimate the spectral radiance factors and the CIE tristimulus values of metallic coatings utilizing the measurements at 6 common geometries of portable multiangle spectrophotometers. Their performance was examined with 65 achromatic and 20 chromatic metallic coating samples under D65, A, and F11 illuminants. The average CIEDE2000 color differences over all 19 geometries were found to be less than 1.8 for both models, while the average CAM02-SCD and CIELAB color differences can, respectively, reach 1.7 and 2.0, indicating the effectiveness of our methods.

Jianfeng Huang,1,* Qinzhou Zou,1,* Danqi Qian,1 Leyuan Zhou,1 Bo Yang,1 Jianjun Chu,1 Qingfeng Pang,2 Kewei Wang,2 Fuzheng Zhang1 1Department of Radiation Oncology, Affiliated Hospital of Jiangnan University, 2Department of Epidemiology, Wuxi Medical School of Jiangnan University, Wuxi, Jiangsu, People’s Republic of China *These authors contributed equally to this work Objective: This study aimed to evaluate the safety and efficacy of intensity-modulated radiotherapy (IMRT) plus nimotuzumab with or without concurrent chemotherapy (CCT) for patients with locally advanced nasopharyngeal carcinoma (LA-NPC). Patients and methods: A total of 50 newly diagnosed patients with LA-NPC treated at the Affiliated Hospital of Jiangnan University between November 2011 and January 2017 were retrospectively analyzed. All patients received the combined treatment modality of nimotuzumab plus IMRT. Nimotuzumab was administered concurrently with IMRT at a weekly dose of 200 mg. Neoadjuvant, concurrent or adjuvant chemotherapy with the doublet regimen of taxanes (docetaxel or paclitaxel) plus platinum (cisplatin or nedaplatin) were administered. Among the 50 patients, 43 (86.0%) received ≥6 cycles of nimotuzumab (median 7 cycles, range 2–14 cycles) and 29 (58.0%) received two cycles of CCT with docetaxel plus nedaplatin. Results: With a median follow-up of 28.0 months, the 2-year progression-free survival (PFS) and overall survival were 83.29% (95% confidence interval [CI]: 67.93%–91.72%) and 97.67% (95% CI: 84.62%–99.67%), respectively. Both univariate and multivariate analyses revealed that cycles of nimotuzumab were significantly associated with PFS. Patients who received≥6cycles of nimotuzumab showed a better PFS than those receiving