Sarcasm is an expression of contempt expressed through verbal irony. It is a nuanced form of language that individuals use to imply the opposite of what they are actually saying, and thus it can be difficult to detect at times. The lack of large, annotated datasets is one of the major challenges and limitations of building systems to detect sarcasm automatically. To address this issue, we propose a hybrid graph-based framework, namely, SINCERE, to build compact sarcasm detection models with sentiment and emotion analysis by leveraging only a small amount of prior data. To automatically extract patterns from a small dataset collected by distant supervision, a graph is first constructed. This approach is used to discover latent representations of vertices in a network, as the basis for a language model. We demonstrate that simple classifiers built from the model can detect sarcasm and generalize better than the state-of-the-art approach. According to the experimental results, the proposed SINCERE framework is able to outperform the SOTA baselines on accuracy by 5%.

Generative Adversarial Networks (GAN) is a popular machine learning method that possesses powerful image generation ability, which is useful for different multimedia applications (e.g., photographic filters, image editing). However, typical GAN models have a large memory footprint that limits their practical applications for resource-constrained devices (e.g., smartphones). To deploy GAN models on devices with various hardware constraints, we propose our method, AdjustableGAN, which can compress a pretrained GAN model to different compression ratios. Our method compresses GAN by performing filter-wise pruning that follows these objectives: (1) deactivate convolutional filters for minimal performance decrease, (2) reactivate convolutional filters for maximal performance increase. We implement multiple Genetic Algorithms (GA) to perform each of these objectives— Downsize GA for best filter deactivations, while Upsize GA searches for best filter reactivations. By selective utilization of Upsize/Downsize GA, we could explicitly control the compression ratio of the model. For finalization, we fine-tune the compressed output model using the training dataset of the original input model. Our experimental results show that our method can reliably compress generative networks with minimal accuracy drop compared to other state-of-the-art compression algorithms.

Nowadays, there are numerous social platforms for people to discuss and share information. However, certain users may post contents that discriminate against specific groups or spread inappropriate behaviors. Therefore, it is desired to have a detection mechanism that is able to accurately identify these offensive comments and inappropriate messages before appearing on social platforms. To address this need, we propose a novel model called Attention-based Multi-filter Convolutional Neural Network (AMCNN) for inappropriate speech detection. Our model combines CNN and BiLSTM to acquire the advantages of the two. AM CNN leverages CNN filters of various kernel sizes for convolution operations to capture the relation between words of different lengths. The convolutional features are then used as the inputs of BiLSTM, which is able to obtain the bidirectional semantics via the forward and backward directions. AMCNN further incorporates the attention mechanism to highlight the important parts of the hidden layer output for better classification. Our model can be applied to various types and lengths of text, and it is also able to support different languages. To evaluate the model performance, we conduct extensive experiments on various real-world datasets. According to the experimental results, our model achieves the best performance, which is on average 12.63% higher than the state-of-the-art approach on the F1-score indicator. We also conduct ablation analysis, and the results show that the attention mechanism and the usage of BiLSTM can considerably improve the classification performance and effectively detect inappropriate speech in sentences.

Graph sparsification aims to reduce the number of edges of a network while maintaining its accuracy for given tasks. In this study, we propose a novel method called GSGAN, which is able to sparsify networks for community detection tasks. GSGAN is able to capture those relationships that are not shown in the original graph but are relatively important, and creating artificial edges to reflect these relationships and further increase the effectiveness of the community detection task. We adopt GAN as the learning model and guide the generator to produce random walks that are able to capture the structure of a network. Specifically, during the training phase, in addition to judging the authenticity of the random walk, discriminator also considers the relationship between nodes at the same time. We design a reward function to guide the generator creating random walks that contain useful hidden relation information. These random walks are then combined to form a new social network that is efficient and effective for community detection. Experiments with real-world networks demonstrate that the proposed GSGAN is much more effective than the baselines, and GSGAN can be applied and helpful to various clustering algorithms of community detection.

Electricity load forecasting is a prevalent research topic in recent years. In this study, we predict the electricity consumption using only previous power data (i.e., without using weather information or other features). We survey existing univariate methods such as MLP-based, CNN-based, XGBoost-based, RF-based, and EN3-bestK. However, these existing methods do not perform well due to that the range of power values varies a lot. Therefore, we present an electricity consumption forecast system called Dynamic Weight Ensemble Model (DWEM). There are three stages in the proposed DWEM. First of all, we provide three types of data serialization in data preprocessing. Second, we train four types of models (i.e., MLP-based, CNN-based, XGBoost-based, and RF-based) for building the ensemble model later. Finally, we combine the four types of models into an ensemble model, using the proposed Two-Phase Ensemble. In the two-phase ensemble, the first phase is to ensemble the models trained using the same algorithm but different serializations, and the second phase is to ensemble the models from different algorithms. The two-phase ensemble method is designed to dynamically adjust weights based on the previous performance of the corresponding models. Moreover, we notice that properly handling missing values is an important factor in system performance. Therefore, we present a statistical method to estimate the missing values. We compare DWEM with various state-of-the-art methods. Comparison of DWEM and the state-of-the-art ensemble method, the results show that DWEM is on average about 46.95% and 44.47% better than EN3-bestK on the MAPE and MAE indicators, respectively.

In recent years, due to the vigorous development of social network media, a large amount of social data can be obtained through social media. This trend has made it easier for researchers in natural language processing to obtain textual research materials, and has further stimulated the development of natural language processing. Emotion recognition is an important task in the field of natural language processing, and emotion recognition helps improve interaction in social. In this study, the emotion recognition of the content of twitter users' posts and comments is based on the sentence level. We use large datasets to collect users' posts and comments on Twitter for comparative analysis, and we attempt to understand the emotions expressed by users. Most of the recent studies focus on the analysis of vocabulary and syntactic features in sentences to learn the emotion of the sentence. However, the property of multi-feature may be conducive for promoting the performance of the model and may also be a disturbance of the training of the model. Our study is based on the latest of attention mechanism and XLNet to improve the recognition of emotions. We propose a novel emotion recognition model called Dynamic Weighted Attention with Multichannel Convolutional Neural Network (DACNN), which combines the multi-channel convolutional neural network and the attention mechanism of automatic weight adjustment to effectively improve the outcome of emotion recognition. In addition, we use the latest word embedding technology XLNet to obtain high-quality feature vectors of sentences. In the experiments, we compare the DACNN with the baselines on various datasets, and the results show that DACNN is better than the state-of-the-art methods in accuracy, precision, recall, and fl-score. The experimental results also prove that our attention mechanism and dynamic weighting can well promote the accuracy of emotion recognition.

Three essential criteria are important for social activity planning: (1) finding attendees familiar with the initiator, (2) ensuring most attendees have tight social relations with each other, and (3) selecting an activity period available to all. In this paper, we propose the Social-Temporal Group Query (STGQ) to find suitable time and attendees with minimum total social distance. We first prove that the problem is NP-hard and inapproximable within any ratio. Next, we design two algorithms, SGSelect and STGSelect, which include effective pruning techniques to substantially reduce running time. Moreover, as users may iteratively adjust query parameters to fine tune the results, we study the problem of Subsequent Social Group Query (SSGQ). We propose the Accumulative Search Tree and Social Boundary, to cache and index intermediate results of previous queries in order to accelerate subsequent query processing. Experimental results indicate that SGSelect and STGSelect are significantly more efficient than baseline approaches. With the caching mechanisms, processing time of subsequent queries can be further reduced by 50-75 percent. We conduct a user study to compare the proposed approach with manual activity coordination. The results show that our approach obtains higher quality solutions with lower coordination effort, thereby increasing the users' willingness to organize activities.

Social media websites have become an important channel for content sharing and communication between users on social networks. The shared images on the websites, even the ones from the same user, tend to receive a quite diverse distribution of views. This raises the problem of image popularity prediction on social media. To address this important research topic, we explore three essential components that have considerable impact of the image popularity, which are user profile, post metadata, and photo aesthetics. Moreover, we make use of state-of-the-art predictive modeling approaches to demonstrate the effectiveness of our proposed features in predicting image popularity. We then evaluate the proposed method through a large number of real image posts from Flickr. The experimental results show significant statistical evidence that incorporating the proposed features with ensemble learning method that combines predictions from support vector regression (SVR) and classification and regression tree (CART) models offers a satisfactory popularity prediction. By understanding the social behavior and the underlying structure of content popularity, our research results can also contribute to designing better algorithms for important applications like content recommendation and advertisement placement.

Three essential criteria are important for activity planning, including: (1) finding a group of attendees familiar with the initiator, (2) ensuring each attendee in the group to have tight social relations with most of the members in the group, and (3) selecting an activity period available for all attendees. Therefore, this paper proposes Social-Temporal Group Query to find the activity time and attendees with the minimum total social distance to the initiator. Moreover, this query incorporates an acquaintance constraint to avoid finding a group with mutually unfamiliar attendees. Efficient processing of the social-temporal group query is very challenging. We show that the problem is NP-hard via a proof and formulate the problem with Integer Programming. We then propose two efficient algorithms, SGSelect and STGSelect , which include effective pruning techniques and employ the idea of pivot time slots to substantially reduce the running time, for finding the optimal solutions. Experimental results indicate that the proposed algorithms are much more efficient and scalable. In the comparison of solution quality, we show that STGSelect outperforms the algorithm that represents manual coordination by the initiator.