Recommender systems are critical for mitigating information overload, assisting users in uncovering their latent interests, and enhancing their overall experience. Sequential recommendation leverages users' historical interaction sequences to predict dynamic interests more effectively than traditional rec-ommendation approaches. However, existing models-including RNN-based and Transformer-based methods-face significant limitations. RNNs struggle with vanishing gradients and long-term dependency capture, while Transformers, though effective for long-range relationships, suffer from computational inefficiency due to their quadratic attention complexity. Recent advancements have employed contrastive learning for sequential recommendation, aiming to enhance the consistency between augmented views and improve self-supervised learning signals. Despite their promise, these methods often lack diversity in data augmentation strategies, which restricts their capacity for bias mitigation, resulting in augmented data that still retains inherent biases. To address these challenges, we propose MDEC, a novel sequential modeling framework that leverages State Space Models (SSM) combined with unbiased contrastive learning. MDEC utilizes Mamba to efficiently model user preferences as an alternative to Transformer-based models. Additionally, it integrates graph-based information, including item transition and co-interaction data, to improve data augmentation comprehensively. Finally, we introduce adaptive anchor-enhanced contrastive learning, which adaptively utilizes augmented samples to improve representation quality and bias mitigation. Extensive experiments on multiple datasets demonstrate that MDEC significantly out-performs existing models, showcasing improved efficiency, better mitigation of biases, and enhanced recommendation quality. Code is available at https://github.com/Echohuangyan/CSLP.

In personalized recommendation systems, Graph Neural Networks (GNNs) have gained attention for their ability to effectively model complex user-item interactions. The core concept of GNNs is to leverage ample and high-quality training data, recursively performing message passing along user-item interaction edges to progressively refine embedding representations. However, the highly imbalanced nature of user interaction data often leads traditional GNN-based recommendation systems to experience information redundancy, concentrating recommendations on a few popular items. Additionally, the long-tail problem further limits the visibility of less popular items, reducing recommendation diversity and impairing the user experience. In this paper, we propose a novel Diversity Augmented Graph Contrastive Learning method (DAGCL), which aims to increase recommendation diversity by improving the embedding generation process, while maintaining a balance between accuracy and diversity. Specifically, we design two graph augmentation methods, including a diversified subset selection module based on Deterministic Point Process (DPP) and a long-tail item interaction augmentation module, to generate diversified embeddings. Moreover, DAGCL incorporates contrastive learning, leveraging augmented views to optimize the recommendation process, alleviate data sparsity issue and capture potential user-item relationships. Experiments on two real-world datasets demonstrate that DAGCL significantly improves diversity while preserving recommendation accuracy, achieving an effective balance between these two aspects and validating its effectiveness. The code is available on https://github.com/XiaHaoZhi/DAGCL.

The goal of Emotion Recognition in Conversations (ERC) is to accurately identify the emotions expressed in each utterance within a dialogue. Despite advancements made by current ERC methods, particularly those using RNN-based and GCN-based models to capture emotional dynamics and model speaker relationships, there remain two primary limitations: first, an insufficient integration of multiple feature representations and commonsense knowledge, which hampers the model's ability for deep emotional understanding; and second, the reliance on a single cross-entropy loss for classification optimization, which restricts the accuracy and robustness of emotion recognition. To address these issues, we propose a method for ERC using Multilayer Feature Fusion and Joint Loss Optimization (MFFJL). This approach combines contextual information, speaker dependency, and commonsense knowledge features by extracting feature vectors through RoBERTa and COMET, utilizing bidirectional LSTM and attention mechanisms to capture conversational context, and applying a cross-fusion module to deeply integrate various features, thus enhancing comprehension of complex emotional expressions. Additionally, the feature classification module incorporates joint cross-entropy and KL divergence optimization, further improving classification accuracy and consistency. Experimental results demonstrate the effectiveness of our method, as evidenced by superior performance on the IEMOCAP and MELD datasets. Our code is available at https://anonymous.4open.science/r/MFFJL-D9D8.

Performing better text translation by integrating auxiliary inputs from visual information has gained widespread attention in recent years. While existing methods outperform the text-only translation models, the step-by-step generative style reduces the inference speed, which limits their applicability in real-world scenarios. In this paper, we propose the non-autoregressive language model (NA-LM) for multimodal machine translation. With NA-LM, we develop a Non-Autoregressive Multimodal Transformer (NA-MMT), which accelerates the generative translation via a parallel multimodal decoder. To retain the translation performance, we improve the NA-MMT in twofold: 1) We preprocess the image into a refined sequence of visual entities with length encoding to reduce irrelevant information; 2) We design cross fertility and cross-modal gate attention for multimodal decoder to enhance the generative quality. Experiments on Multi30k datasets show the NA-MMT can generate high-quality translation with over 11× speedup than the baselines, which is strongly competitive.

Multimodal dialogue emotion recognition integrates data from multiple modalities to accurately identify emotional states in conversations. However, differences in expression and information density across modalities complicate the fusion of features. Traditional methods may introduce redundant information from other utterances, reducing the accuracy of emotion recognition. Existing one-hot labels often fail to capture the full range of emotional expressions, leading to biased results. To address these issues, we propose a model that fuses different modalities within the same utterance to avoid redundancy. It employs a progressive classification process, refining emotion recognition from coarse to fine granularity. Additionally, we use emotion polarity probabilities as weights for fine-grained classification and introduce a multimodal information-rich label that considers both the data and their interactions. Experiments on IEMOCAP and MELD datasets demonstrate the model’s effectiveness, significantly improving dialog emotion recognition accuracy. Our code is available at https://anonymous.4open.science/r/LOCG-188E.

Emotion recognition in conversation (ERC) aims at accurately identifying emotional states expressed in conversational content. Existing ERC methods, although relying on semantic understanding, often encounter challenges when confronted with incomplete or misleading semantic information. In addition, when dealing with the interaction between emotional and semantic information, existing methods are often difficult to effectively distinguish the complex relationship between the two, which affects the accuracy of emotion recognition. To address the problems of semantic misdirection and emotional cross-talk encountered by traditional models when confronted with complex conversational data, we propose a semantic and emotional dual channel (SEDC) strategy for emotion recognition in conversations to process emotional and semantic information independently. Under this strategy, emotion information provides an auxiliary recognition function when the semantics are unclear or lacking, enhancing the accuracy of the model. Our model consists of two modules: the emotion processing module accurately captures the emotional features of each utterance through contrastive learning, and then constructs a dialogue emotion propagation map to simulate the emotional information conveyed in the dialogue; the semantic processing module combines an external knowledge base to enhance the semantic expression of the dialogue through knowledge enhancement strategies. This divide-and-conquer approach allows us to more deeply analyze the emotional and semantic dimensions of complex dialogues. Experimental results on the IEMOCAP, EmoryNLP, MELD, and DailyDialog datasets show that our approach significantly outperforms existing techniques and effectively improves the accuracy of dialogue emotion recognition.

A significant problem in sequential recommendation (SR) is the over-recommendation of popular items, leading to popularity bias, as users often follow these items due to conformity. Existing methods measure users’ conformity factors to reduce the impact of popularity bias. However, these methods do not consider the differences in users’ conformity behavior in the long-term and short-term. To address this, we propose LSDRec, a novel debiased SR method structured around three key tasks: degree-centrality conformity awareness, dual-scale interest encoding, and adaptive conformity information fusing. The degree-centrality conformity awareness task constructs a multiuser interaction graph, employs a graph convolutional network (GCN) to obtain global user conformity representations, and uses the degree centrality algorithm to compute users’ long-term and short-term conformity factors. The dual-scale interest encoding task models users’ long-term and short-term interests separately, obtaining corresponding interest representations and further enhancing them through the adaptive conformity information fusing task. The adaptive conformity information fusing task contrasts global conformity representations with long-term and short-term interest representations, adaptively integrating conformity factors and dynamically adjusting the degree of conformity information transfer. Together, these three tasks effectively mitigate popularity bias and improve the accuracy of user interest modeling. Our extensive evaluations of four diverse datasets demonstrate LSDRec's superior performance over current state-of-the-art methods.

Personalized news recommendation helps users find interesting content among much news information. Most existing news recommendation techniques densely interact a user’s historical clicked news with candidate news to capture their reading interest. However, during this process, even if certain historical clicked news is weakly correlated with the candidate news, the model still interacts with it, generating more noise information when constructing user interests. Simultaneously, the news representation learning process using neural networks is a fine-grained to coarse-grained learning process; i.e., as the network deepens, the fine-grained original news information is easily lost. Therefore, to solve the above problems, this paper proposes a news recommendation model based on hierarchical granular semantic embeddings and threshold attention (HGTA). In HGTA, we propose a threshold attention module to filter unimportant historical clicked news by setting an attention score threshold, i.e., interacting only with historical clicked news that is strongly correlated with the candidate news when capturing user interests. Thus, the model attention is focused, and the noisy interaction information is reduced. Furthermore, when training the neural network, to prevent the loss of fine-grained information, we design a hierarchical granular information extraction module to capture both coarse- and fine-grained news features to better enrich the news semantics. Experiments conducted on the large real-world Microsoft News Dataset (MIND) show that our model achieves highly competitive results in terms of popular metrics. Our code is published at the following URL: https://github.com/liyiweneven/HGTA.

In recent years, aspect sentiment quadruple prediction (ASQP) has become a popular task in the field of aspect-based sentiment analysis. When modeling comment statements, it is often necessary to consider structural information between sentiment labels. Existing approaches categorize these labels by encoding text and dependency tree structures separately and then incorporating their representations, but this approach fails to effectively capture the structural information between sentiment labels. To address this problem, we propose a method for ASQP called Opinion-Tree-guided Contrastive Learning (OTCL), which embeds structural information directly into the text instead of modeling it separately. Specifically, OTCL utilizes the opinion tree structure to guide the construction of positive samples of the input text during training. By associating the input text with its positive samples, the text encoder can learn to generate textual representations associated with the opinion tree structure. We validate the effectiveness of our method on two commonly used datasets.

The most critical task in personalized news recommendation is to perform exact matching between candidate news and users’ interests. Existing news recommendation methods usually model users’ interests from historical clicked news items without considering candidate news, which makes it difficult to precisely match candidate news with users’ interests. Moreover, it is challenging to distinguish similar news when forming a vector representation. Simply adding candidate news to the user interest modeling process without considering word-level information does not provide sufficient discrimination for recommending appropriate news. In this paper, we propose a news recommendation model with candidate-aware fine-grained interaction information (CAFI). In our approach, we propose a fine-grained interaction module that matches candidate news items and text fragments of each historical news item at each semantic granularity to achieve interaction between the two at the word level and help similar historical news form more specific and accurate representations. In addition, our proposed gated self-attention mechanism utilizes the candidate news features as channel moderation gates to implement the process of filtering out information that is irrelevant to the candidate news for learning user interest representations, thereby better matching the candidate news to specific user interests. Experiments conducted on the large real-world Microsoft News Dataset (MIND) show that our model significantly outperforms the previously developed models in terms of all metrics. Our code is published at the following URL: https://github.com/liyiweneven/CAFI.

Emotion recognition in conversation (ERC) is a prominent research topic in natural language processing, widely applicable in various scenarios. However, accessing external commonsense knowledge and related dialogue topics, along with the complexity of fusing utterances, presents significant challenges. In this paper, we introduce a Topic-based Multilayer Knowledge Filtering (TMKF) model to enhance dialog emotion recognition accuracy. TMKF acquires commonsense knowledge for each utterance and employs two Variational Autoencoder (VAE) modules to extract global and speaker-level dialogue topics. We utilize Global Knowledge Filtering (GKF) and Local Knowledge Filtering (LKF) to obtain topic-specific knowledge representations after filtering commonsense information through global and local topic hierarchies. Subsequently, we leverage relational graphs to convolve commonsense knowledge with utterances, resulting in knowledge-constrained utterance representations for classification. Our proposed model is evaluated through extensive experiments on four widely used benchmark datasets for conversational emotion recognition. The results underscore the effectiveness of TMKF, which significantly outperforms other methods in standard metric evaluations.

The objective of sequential recommendation is to predict user preferences for items based on historical interaction sequences. This process often leads to a phenomenon known as popularity bias, where popular items are excessively recommended. Conformity, the tendency of users to follow popular items, is a significant factor contributing to this issue. Previous methods have not adequately disentangled conformity and interest, failing to accurately model users’ true intent. To address this, we propose a novel Disentangled Interest and Conformity Sequential Recommendation method (DICSRec) to mitigate the popularity bias. Specifically, we first design an Intent Encoding Module (IEM), which includes two independent encoders for conformity and interest to model their representations. To better disentangle these two factors, we design a disentangling task with proxy-based self-supervised learning and orthogonal regularization. Furthermore, to provide the Intent Encoding Module with more global information, we design a Global Conformity-aware Module (GCM), which supplies item popularity information and aids in enhancing user conformity representation. Lastly, recognizing the varying significance of user conformity and interest, we propose an adaptive Fusion Prediction Module (FPM) that adaptively aggregates user conformity and interest representations for final prediction. Experiments on four real-world datasets consistently demonstrate the superiority of our method over advanced sequential recommendation models. Code implementation is available at: https://github.com/lyra0611/DICSRec.

Emotion Recognition in Conversations (ERC) is a widely applicable task, whether in emotional chatbots or recommender systems for dialog scenarios. The ability of systems to accurately identify human emotions is a crucial component in ERC success. In everyday conversations, a speaker’s emotional expression is closely tied to their individual characteristics and personality traits. Furthermore, individuals exhibit significant variations in their expressions of the same emotion. Traditional methods for ERC often overlook these personality differences and dynamic changes, leading to unreliable results. To address this issue, we propose a conversation emotion recognition model based on dynamic modeling of speaker’s personality (ERC DMSP). In our approach, to extract a speaker’s personality profile, we design a personality capture module to model the personalities of individuals based on their historical utterances. Moreover, a bidirectional influence exists between a speaker’s personality tendencies and their linguistic behavior. The speaker’s personality profile is thus continually updated based on the ongoing dialog. Additionally, since the personalities of other speakers in the conversation can also affect the form and intensity of a speaker’s emotional expression, we construct a conversation emotion representation utilizing group personality traits. We then comprehensively incorporate the speaker’s personality into the emotion representation of each utterance at both the group and individual levels through the personality engagement module. Throughout this process, the representation of dialog and the speaker’s personality undergo hierarchical updates, leading to more accurate conversation emotion recognition. We conducted extensive experiments on four commonly used public benchmark datasets for ERC to evaluate the effectiveness of our proposed ERC DMSP model. The results demonstrate its efficacy, as it exhibits a significant improvement over other methods in popular evaluation metrics. Our code is available for reference at https://github.com/Tars-is-a-robot/ERC-DMSP.

Sequential Recommender Systems (SRS), leveraging the temporal information from users' behaviors, have noticeably improved user experience against traditional systems. However, these behaviors often follow long-tail distribution, making the systems biased towards popular items (i.e., popularity bias). Moreover, popularity bias would amplify the neglect of long-tail recommendations, thereby sharpening the long-tail problem. Previous researches usually address these challenges independently, focusing on reducing the over-recommendation of popular items or enhancing the representation quality of tail items. Indeed, it is possible to incorporate their merits to achieve the best of both worlds. Thus, we propose a novel and unified framework, named Collaborative Solution to Long tailed problem and Popularity bias (CSLP), to tackle both the long-tail problem and popularity bias simultaneously. To achieve this, we first introduce a representation enhancement module featuring dual generators to enhance user and item representations, particularly for those in the tail. On the other hand, a debiasing module incorporating an Inverse Propensity Score (IPS) with a clipping strategy is introduced to further alleviate the popularity bias. Specifically, this clipping strategy demonstrates a clear decrease in the original IPS method's variance, effectively improving the recommendation for stability and accuracy. Experiments on three widely-used datasets show CSLP's effectiveness in solving both issues. CSLP surpasses all baselines (traditional, popularity bias, and long-tail problem) in overall performance, significantly enhancing recommendation accuracy for both tail users and items, and achieving a more balanced ratio of recommendations between popular and tail items. Code is available at https://github.com/Echohuangyan/CSLP.

Personalized News Recommendation (PNR) can enhance user experience by alleviating information overload. Traditional news recommendation methods consider all clicking behaviors as user interests, resulting in biased user modeling that fails to accurately capture user interests. In addition, although there are methods to reduce the impact of low-quality news at the representation level by simply filtering it through the attention mechanism. However, this only implicitly models the news in the interaction sequence without specifically considering the quality of each news, and thus has very limited effect in identifying noise. To address these issues, this paper proposes News Recommendation method based on News Quality-aware modeling (NQNR). We attempt to explicitly model the news in the click sequence and candidate ranking one by one to visually assess the quality of each news. Specifically, we design a detection module to detect whether the input news is low-quality news. Then, by reducing the influence of low-quality news in user modeling and candidate ranking, user interests are modeled more accurately, while recommendations of such news are reduced for users. In addition, to capture the similarity of vectors more accurately, we also design a similarity computation method based on the multiple attention mechanism in the detection module. Experiments on a large real-world Microsoft News Dataset (MIND) show that our model significantly outperforms previous models. Our code is posted at the following URL: https://github.com/xxbbjj/NQNR-.

Emotion recognition in conversation (ERC) is a widely used technology in both affective dialogue bots and dialogue recommendation scenarios, where motivating a system to correctly recognize human emotions is crucial. Uncovering as much contextual information as possible with a limited amount of dialogue information is essential for eventually identifying the correct emotion of each sentence. The integration of contextual information using the existing approaches often results in inadequate access to information or information redundancy. Deeply integrating the different knowledge behind utterances is also difficult. Therefore, to address these problems, we propose a dynamic complementary graph convolutional network (DCGCN) for conversational emotion recognition. Our approach uses commonsense knowledge to complement the contextual information contained in utterances and enrich the extracted conversation information. We creatively propose the concept of utterance density to prevent redundancy and the loss of utterance information in context-dependent contextual information modeling cases. An utterance dependency structure is dynamically determined by the utterance density, and the contextual information is fully integrated into each sentence representation. We evaluate our proposed model in extensive experiments conducted on four public benchmark datasets that are commonly used for ERC. The results demonstrate the effectiveness of the DCGCN, which achieves competitive results in terms of well-known evaluation metrics. Our code is available athttps://github.com/Tars-is-a-robot/Conversational-emotion-recognition.git.

In recent years, aspect sentiment quadruple prediction (ASQP) has become popular in aspect-based sentiment analysis (ABSA). Its purpose is to decode a given sentence into aspect sentiment quadruples (aspect category, aspect term, opinion term, and sentiment polarity). When trying to efficiently extract aspect sentiment quadruples, the following problems are often encountered: Firstly, the intrinsic relationships between aspect terms and opinion terms are usually ignored, thus failing to address the correlation between establishing aspect-opinion pairs and ignoring the mutual interference between different sentiment quadruples; Secondly, the semantic information contained in the sentiment elements of comment utterances is often underutilized, thus increasing the risk of obtaining inaccurate predictions. We propose a two-stage framework to address these issues by enhancing the correlations between aspects and opinions and fully utilizing the semantic information of sentiment elements. Specifically, in the first stage, we treat the extraction task as a machine reading comprehension (MRC) problem, employ a span-based labeling scheme, and construct a question-and-answer-based MRC task to efficiently extract aspect-opinion pairs. In the second stage, we view the classification of aspect categories and sentiment polarities as a text generation task, where the semantics of sentiment elements can be leveraged by learning to generate them in natural language form. Finally, the two stages are combined with our proposed template generator, and the aspect sentiment quadruples can be decoded. We conducted experiments on two datasets, and the experimental results were superior to those of the comparison approaches, with our model demonstrating excellent performance in terms of processing complex sentences containing multiple quaternary groups. Additionally, in sub-task experiments, our model achieved good results, further proving its effectiveness. We will upload the specific code to https://github.com/SlienceLZJ/ASQP.

The different news clicked by users reflects the diverse interests of users. Most of the existing news recommendation methods do not consider the interaction with candidate news in the process of modeling user interest representation. This method makes it challenging to precisely match candidate news to specific user interests. We propose a user interest activation recommendation method that fuses multichannel information—MIAR. It utilizes the word embedding of the user’s historical clicked news and the news title embedding generated by aggregation and interacts with the candidate news, respectively, to better match the candidate news with the user’s interests. Our proposed method contains two frameworks (interactive framework and distributed framework). In the interactive framework, we propose a user multichannel interest modeling framework MIF from the word embedding level of news headlines to capture more semantic cues related to user interests. In the distributed framework, we design a candidate-aware interest activation module TAR from the news embedding representation level obtained by attention aggregation. It uses different candidate news vectors to adjust the user representations learned from the user’s historical reading records. This allows the model to build candidate-guided user representations to accurately match candidate news to parts of user interests that are relevant to the candidate news. Finally, we effectively assign the weights of the two frame scores so that the models can fuse better. Extensive experiments on the MIND news recommendation dataset demonstrate the effectiveness of our method.

In rating prediction research, how to capture user and item features from review text is a key to improving model prediction accuracy. The sparsity of review text and the accuracy of the description of items in the review text make it difficult to obtain accurate feature representations of users and items by modeling the text content alone. Therefore, it is important to evaluate the usefulness of the reviews at first because not all review texts are valuable. How to analyze the usefulness of a review is a key to modeling the review text. The way previous models use attention to inscribe semantic weights on the review text is not sufficient to indicate the degree of usefulness of a review, so we suggest adding property information to model reviews. Based on this, we propose an interaction recommendation model that is based on enhanced graph convolution and fuses review properties (PGIR), which incorporates property information into text modeling by different activations and matches useful property feature interaction pairs for review text in a self-supervised manner. This allows the model to obtain an accurate feature representation of the review text. In addition, we analyze the high-order connectivity among user–item pairs. Then, we design an enhanced graph convolution method to capture the collaborative signals between users and items and model the dynamic features of users and items on this basis. After extensive experiments conducted on five standard datasets based on Amazon, the results show that the PGIR model achieves a substantial improvement over existing state-of-the-art models in terms of rating prediction. In addition, we experimentally demonstrate the superiority of our proposed property activation method, which further improves the rating prediction performance of the PGIR model.

The goal of the few-shot image classification is to identify the category based on a very small number of labeled samples. Two of the key problems are the insufficient amount of labeled data and the unknown category (the inconsistency between the training category and the test category). For these two problems, we propose a new few-shot classification model VE-HKF. First, we introduce a Virtual label expansion mechanism (VE). This mechanism expands the support set data by using the unlabeled data in the query set, thus increasing the number of samples in the support set and making the extracted features more robust. Second, we introduced a Highlighted key features mechanism (HKF). The mechanism first generates a mask through operations such as class average vector and dimensionality reduction and uses this mask to shield and support some irrelevant features, to highlight the important features between samples of each category in a disguised manner, and then focus on the support set and the features of the query set samples to highlight the common features between the support set and the query set, making the extracted features more conducive to classification. On the three data sets of Mini-ImageNet, Omniglot, and Tiered-ImageNet, our model has achieved good results.