Multi-trait Essay Scoring (MES) aims to evaluate the quality of essays across multiple traits (e.g., Language, Content, and Organization). The task can be summarized into three crucial steps: essay content encoding, trait feature learning, and multi-trait scoring. However, previous methods fall short in these steps due to neglecting essential scoring-oriented knowledge, leading to suboptimal performance. To solve these issues, we propose a novel multi-trait scoring framework with multi-knowledge enhancement. Specifically, linguistic knowledge is used to model syntactic structural relations between words, highlighting structurally-informed essay encoding. We learn trait knowledge by capturing the knowledge dependencies between traits to enhance trait-specific features. Further, score-aware ordinal knowledge is integrated to promote ordinal alignment in trait-specific features associated with score rankings, improving scoring performance. Extensive experiments show that our proposed method achieves significant performance.

Few-shot Multi-label Aspect Category Detection (FMACD) is an essential task, which aims to identify multiple aspect categories in a given sentence with limited data. Recently, the prototypical network as a mainline has been used for the task due to its powerful capacity. However, existing methods mostly rely on intra-cluster samples to generate prototypes, and they struggle to extract robust prototype features in very few data cases (e.g., 1-shot). Therefore, these methods may fail to estimate label-query relevance during multi-label prediction. To solve the above issues, we propose a novel relation graph guided learning method for FMACD by considering all intra- and inter-cluster samples. Specifically, the proposed method explicitly models a relation graph to generate more robust prototypes by exploring sample relations among intra- and inter-cluster. Then, a multi-label inference strategy is proposed to enhance label-query relevance for multi-label prediction. Besides, graph contrastive learning enhances intra-cluster commonality and inter-cluster uniqueness to improve performance. Experiments show that the proposed method achieves significant performance, esp., it obtains an average of 1.55% AUC and 5.01% Macro-F1 improvement in 1-shot scenarios.

Few-shot Multi-label Intent Detection (MID) is crucial for dialogue systems, aiming to detect multiple intents of utterances in low-resource dialogue domains. Previous studies focus on a two-stage pipeline. They first learn representations of utterances with multiple labels and then use a threshold-based strategy to identify multi-label results. However, these methods rely on representation classification and ignore instance relations, leading to error propagation. To solve the above issues, we propose a multi-label joint learning method for few-shot MID in an end-to-end manner, which constructs an instance relation learning network with label knowledge propagation to eliminate error propagation. Concretely, we learn the interaction relations between instances with class information to propagate label knowledge between a few labeled (support set) and unlabeled (query set) instances. With label knowledge propagation, the relation strength between instances directly indicates whether two utterances belong to the same intent for multi-label prediction. Besides, a dual relation-enhanced loss is developed to optimize support- and query-level relation strength to improve performance. Experiments show that we outperform strong baselines by an average of 9.54% AUC and 11.19% Macro-F1 in 1-shot scenarios.

Abstract Few-shot Aspect Category Sentiment Analysis (ACSA) is a crucial task for aspect-based sentiment analysis, which aims to detect sentiment polarity for a given aspect category in a sentence with limited data. However, few-shot learning methods focus on distance metrics between the query and support sets to classify queries, heavily relying on aspect distributions in the embedding space. Thus, they suffer from overlapping distributions of aspect embeddings caused by irrelevant sentiment noise among sentences with multiple sentiment aspects, leading to misclassifications. To solve the above issues, we propose a metric-free method for few-shot ACSA, which models the associated relations among the aspects of support and query sentences by Dual Relations Propagation (DRP), addressing the passive effect of overlapping distributions. Specifically, DRP uses the dual relations (similarity and diversity) among the aspects of support and query sentences to explore intra-cluster commonality and inter-cluster uniqueness for alleviating sentiment noise and enhancing aspect features. Additionally, the dual relations are transformed from support-query to class-query to promote query inference by learning class knowledge. Experiments show that we achieve convincing performance on few-shot ACSA, especially an average improvement of 2.93% accuracy and 2.10% F1 score in the 3-way 1-shot setting.

Multi-label Aspect Category Detection (MACD) is essential for aspect-based sentiment analysis, which aims to identify multiple aspect categories in a given sentence. Few-shot MACD is critical due to the scarcity of labeled data. However, MACD is a high-noise task, and existing methods fail to address it with only two or three training samples per class, which limits the application in practice. To solve above issues, we propose a group of Few-shot Sample-set Operations (FSO) to solve noisy MACD in fewer sample scenarios by identifying the semantic contents of samples. Learning interactions among intersection, subtraction, and union networks, the FSO imitates arithmetic operations on samples to distinguish relevant and irrelevant aspect contents. Eliminating the negative effect caused by noises, the FSO extracts discriminative prototypes and customizes a dedicated query vector for each class. Besides, we design a multi-label architecture, which integrates with score-wise loss and multi-label loss to optimize the FSO for multi-label prediction, avoiding complex threshold training or selection. Experiments show that our method achieves considerable performance. Significantly, it improves by 11.01% at most and an average of 8.59% Macro-F in fewer sample scenarios.

Nowadays, encrypted traffic classification has become a challenge for network monitoring and cyberspace security. However, the existing methods cannot meet the requirements of encrypted traffic classification because of the encryption protocol in communication. Therefore, we design a novel neural network named Stereo Transform Neural Network (STNN) to classify encrypted network traffic. In STNN, we combine Long Short Term Memory (LSTM) and Convolution Neural Network (CNN) based on statistical features. STNN gains average precision about 95%, average recall about 95%, average F1-measure about 95% and average accuracy about 99.5% in multi-classification. Besides, the experiment shows that STNN obviously accelerates the convergence rate and improves the classification accuracy.