Search

Aspect-based sentiment analysis, which aims to predict the sentiment polarities for the given aspects or targets, is a broad-spectrum and challenging research area. Recently, pre-trained models, such as BERT, have been used in aspect-based sentiment analysis. This fine-grained task needs auxiliary information to distinguish each aspect. But the input form of BERT is only a words sequence which can not provide extra contextual information. To address this problem, we introduce a new method named GBCN which uses a gating mechanism with context-aware aspect embeddings to enhance and control the BERT representation for aspect-based sentiment analysis. Firstly, the input texts are fed into BERT and context-aware embedding layer to generate BERT representation and refined context-aware embeddings separately. These refined embeddings contain the most correlated information selected in the context. Then, we employ a gating mechanism to control the propagation of sentiment features from BERT output with context-aware embeddings. The experiments of our model obtain new state-of-the-art results on the SentiHood and SemEval-2014 datasets, achieving a test F1 of 88.0 and 92.9 respectively.

Event detection is a particularly challenging problem in information extraction. The current neural network models have proved that dependency tree can better capture the correlation between candidate trigger words and related context in the sentence. However, syntactic information conveyed by the original dependency tree is insufficient for detecting trigger since the dependency tree obtained from natural language processing toolkits ignores semantic context information. Existing approaches employ a static graph structure based on original dependency tree which is incompetent in terms of distinguishing interrelations among trigger words and contextual words. So how to effectively make use of relevant information while ignoring irrelevant information from the dependency trees remains a challenging research question. To address this problem, we investigate a graph convolutional network over multiple latent context-aware graph structures to perform event detection. We exploit a multi-head attention mechanism on BERT representation and original adjacency matrix to generate multiple latent context-aware graph structures (a “dynamic cutting” strategy), which can automatically learn how to select the useful dependency information. Furthermore, we investigate graph convolutional networks with residual connections to combine the local and non-local contextual information. Experimental results on ACE2005 dataset show that our model achieves competitive performances compared with the methods based on dependency tree for event detection.

Shotcrete spray is an indispensable process in tunnel construction. At present, the construction of tunnels in China is mainly depend on labor or mobile concrete sprayer, which has lots problems like time-consuming, low precision, and labor intensive. An intelligent detection method for tunnel shotcrete spraying is proposed in this article. There are two main issues need to be solved, one is the modeling of tunnel in real-time to monitor the thickness of shotcrete and other is the detection of spraying area in the tunnel. The LiDAR can obtain a 3D model of tunnel after performing necessary preprocess on it in real-time. On the other hand, the spraying areas are usually divided by arches in the tunnel, so we can detect the position of arches to determine the spraying areas. Inspired by the YOLO algorithm, we proposed a novel neural network structure to detect the approximate bounding boxes of the arches and a line-detection algorithm is used to determine the final positions of the spraying area in the image. The size of the weight file of our neural network is only 2.57 MB after the use of some deep compression tricks, which means our model is device friendly. After that, the object detection results in the image will be projected to the point cloud data. The experimental results suggest that our method performed well in the detection for tunnel shotcrete spraying, and the mAP for spraying area detection was found to be 91.4%.

Knowledge Base Question Answering (KBQA) is a promising approach for users to access substantial knowledge and has become a research focus in recent years. Our paper focuses on relation detection, a subtask of KBQA and proposes an adversarial training improved multi-path multi-scale relation detector (AdvT-MMRD) to improve the performance of a common KBQA system. To solve the problem of matching the casual form of a question with the logical form of a predicate, we use question pattern-relation matching, in which an attention-based bidirectional recurrent neural network with gated recurrent units (Bi-GRUs) is used to match semantic similarity and a convolutional neural network (CNN) is used to learn literal similarity between question and relation. We also explore two ways to measure the relevance of entity type-relation pairs through several level representations. Additionally, an adversarial training strategy is conducted to enhance our model. The experimental results demonstrate that our approach not only achieves a state-of-the-art accuracy of 93.8% on relation detection task, but contributes our KBQA system to reaching an outstanding accuracy of 79.0% on the SimpleQuestions benchmark.

Owing to a dramatic increase of traffic, high demand for quality of service (QoS), and insufficient radio resources, real-time video streaming transmission with stringent delay constraints has been intensely concerned. By exploiting the device-to-device (D2D) multicast communications, this paper proposes a video streaming transmission scheme based on the frame priority (FP) to improve the QoS perceived by users and the users' satisfaction about the video quality. First, the FP strategy is proposed, which mainly considers the encoding characteristics of video streaming and users' feedback to ensure that the frames to be retransmitted are valuable for decoding. Then, in order to transmit a sufficient number of valuable frames within the delay constraints, the optimization of time consumption of the retransmitted frame is formulated. Based on this, the physical-layer resource allocation in the D2D multicast networks is discussed, where the relay selection, D2D subgroup forming, and channel allocation are jointly investigated with the aid of three-dimensional channel quality matrix. Furthermore, a heuristic algorithm is proposed to obtain a near-optimal performance with low complexity. The simulation results verify the advantages of our proposed transmission scheme in users' video reception quality and the satisfaction of all users.

List-based simulated annealing (LBSA) algorithm, which uses list-based cooling scheme to control the change of parameter temperature, was first proposed for traveling salesman problem. This paper extends the application of LBSA algorithm for 0-1 knapsack problem (0-1 KP). A hybrid greedy repair and optimization operator, which combines density-based and value-based greedy repair and optimization operators, is designed to get better balance between intensification and diversification. Extensive experiments were performed to show the effectiveness and parameter robustness of a list-based cooling scheme and to verify the advantage of using hybrid greedy repair and optimization operator. Comparing experiments, which were conducted on small-scale, medium-scale, and large-scale 0-1 KP instances, have shown that LBSA algorithm is better than or competitive with other state-of-the-art metaheuristics.