Recommendation systems aim to learn user interests from historical behaviors and deliver relevant items. Recent methods leverage large language models (LLMs) to construct and integrate semantic representations of users and items for capturing user interests. However, user behavior theories suggest that truly understanding user interests requires not only semantic integration but also semantic reasoning from explicit individual interests to implicit group interests. To this end, we propose an Iterative Semantic Reasoning Framework (ISRF) for generative recommendation. ISRF leverages LLMs to bridge explicit individual interests and implicit group interests in three steps. First, we perform multi-step bidirectional reasoning over item attributes to infer semantic item features and build a semantic interaction graph capturing users' explicit interests. Second, we generate semantic user features based on the semantic item features and construct a similarity-based user graph to infer the implicit interests of similar user groups. Third, we adopt an iterative batch optimization strategy, where individual explicit interests directly guide the refinement of group implicit interests, while group implicit interests indirectly enhance individual modeling. This iterative process ensures consistent and progressive interest reasoning, enabling more accurate and comprehensive user interest learning. Extensive experiments on the Sports, Beauty, and Toys datasets demonstrate that ISRF outperforms state-of-the-art baselines. The code is available at https://github.com/htired/ISRF.

Fine-grained emotion recognition aims to identify the emotional type in queries through reasoning and decision-making processes, playing a crucial role in various systems. Recent methods use In-Context Learning (ICL), enhancing the representation of queries in the reasoning process through semantically similar examples, while further improving emotion recognition by explaining the reasoning mechanisms. However, these methods enhance the reasoning process but overlook the decision-making process. This paper investigates decision-making in fine-grained emotion recognition through prototype theory. We show that ICL relies on similarity matching between query representations and emotional prototypes within the model, where emotion-accurate representations are critical. However, semantically similar examples often introduce emotional discrepancies, hindering accurate representations and causing errors. To address this, we propose Emotion In-Context Learning (EICL), which introduces emotionally similar examples and uses a dynamic soft-label strategy to improve query representations in the emotion reasoning process. A two-stage exclusion strategy is then employed to assess similarity from multiple angles, further optimizing the decision-making process. Extensive experiments show that EICL significantly outperforms ICL on multiple datasets.

Imbalanced Time-Series Classification is a critical, yet challenging task across a spectrum of real-world applications. Previous oversampling and generative approaches primarily target the minority class and often rely on static decision boundaries or similarity-based heuristics. However, these methods overlook the underlying causal factors that govern the distinction between majority and minority classes, particularly in scenarios with ambiguous class boundaries. As a result, the generated samples may fail to enhance class separability, thereby limiting improvements in classification performance. To this end, we propose a CounterFactual Augmentation Minority Generation (CFAMG) method based on generative models that aims to discover the causal factors that determine different classes from a causality perspective. Specifically, our method first utilizes a disentangled classifier to distinguish between causal and non-causal factors. Next, we perform counterfactual intervention by replacing the causal factors of majority class samples with those from minority class samples, creating an intervened latent representation that reflects minority characteristics while preserving essential structures. Finally, the trained minority class decoder generates counterfactual minority samples that resemble real minority instances yet remain distinguishable from the original majority class. Extensive experiments demonstrate that our method outperforms state-of-the-art methods in both univariate and multivariate imbalanced time-series classification tasks. The code is published at https://github.com/WangLei-CQU/CFAMG.

Multimodal emotion recognition in conversations aims to accurately detect emotions by integrating audio, text, and video modalities, playing an important role in various systems. Existing approaches utilize convolutional and recurrent networks to learn short-term emotional information from individual modalities, or employ graph and attention mechanisms to integrate long-term emotional information from multiple modalities. These methods effectively combine emotional information within the conversational content in the time domain.However, psychological research shows that emotional information are not only conveyed in the time domain but also in the frequency domain (e.g., pitch and speech rate). To capture emotions from a more comprehensive perspective, we propose TF-MERC, a framework that integrates both time and frequency domains.TF-MERC uses a multi-domain alignment module to learn modality information within the time or frequency domains. It then employs FATransformer to deeply integrate the multimodal associations between the time and frequency domains, providing a more comprehensive approach for emotion prediction.Experimental results show that TF-MERC outperforms state-of-the-art methods, achieving superior performance across multiple datasets.

Running has emerged as an alternative to traditional meditation practices that involve sitting with closed eyes. The rhythmic nature of running establishes a connection between mind and body to promote mindfulness and improve the overall experience and benefits of running. The use of technology-supported meditation is gaining attention in the fields of mental and physical wellness. However, existing meditation technologies primarily focus on one-way stimuli for sitting or walking meditation, with little emphasis on adaptive design and development for running meditation. To address this issue, we present the design framework and system development of RunMe, an adaptive sound system specifically designed for running meditation. RunMe integrates stimulation and regulation mechanisms to enhance adaptive data interactions between the sounds and runners’ biodata. We compared the significance of the RunMe group with three other groups: use of non-adaptive sound, use of favorite music, and use of no music/sound. The results show that the RunMe group outperforms the other groups in attention regulation, body awareness, exercise motivation, and mindfulness. Importantly, RunMe has the potential to allow users to engage in running meditation without specialized equipment, making it accessible for daily practice. We also discuss the design framework and practical distinctions of RunMe, as well as the design implications and future directions for advancing running meditation.

Zhou Yang, Zhaochun Ren, Wang Yufeng, Haizhou Sun, Chao Chen, Xiaofei Zhu, Xiangwen Liao. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Entity representation plays a central role in building effective entity retrieval models. Recent works propose to learn entity representations based on entity-centric contexts, which achieve SOTA performances on many tasks. However, these methods lead to poor representations for unseen entities since they rely on a multitude of occurrences for each entity to enable accurate representations. To address this issue, we propose to learn enhanced descriptional representations for unseen entities by distilling knowledge from distributional semantics into descriptional embeddings. Specifically, we infer enhanced embeddings for unseen entities based on descriptions by aligning the descriptional embedding space to the distributional embedding space with different granularities, i.e., element-level, batch-level and space-level alignment. Experimental results on four benchmark datasets show that our approach improves the performance over all baseline methods. In particular, our approach can achieve the effectiveness of the teacher model on almost all entities, and maintain such high performance on unseen entities.

Temporal web graphs have been attracting much attention recently due to their important applications in web search, data mining, and social network analysis. Accumulated over long periods, those graphs have grown gigantic in size and rich in temporal evolution, which poses tough challenges for data storage and management. Though a few temporal graph management systems were previously proposed, none of them can simultaneously satisfy both essential requirements when retrieving on temporal web graphs: very large data scalability and very low querying latency.In this work, we address the above gap in existing works by developing a highly efficient temporal graph management system which is dedicated to web graphs. To this end, we greatly extend the most efficient framework for managing large static web graphs to handle temporal information using the property matrix while preserving most of the outstanding features of the base framework. Ultimately, our proposed system can achieve a nearly instant response for vertex-centric temporal retrieval while still being scalable to huge datasets. Experiments on a real-world dataset with more than 43B nodes and 317B links show that using a small non-dedicated cluster, our system can reach a reduction of data storage space up to 88% of raw data size and reduce the retrieval time by 20%, compared to the baselines. We also demonstrate that our system also yields a significant reduction of computational costs for many graph ranking algorithms.

Publicly available social media archives facilitate research in the social sciences and provide corpora for training and testing a wide range of machine learning and natural language processing methods. With respect to the recent outbreak of the Coronavirus disease 2019 (COVID-19), online discourse on Twitter reflects public opinion and perception related to the pandemic itself as well as mitigating measures and their societal impact. Understanding such discourse, its evolution, and interdependencies with real-world events or (mis)information can foster valuable insights. On the other hand, such corpora are crucial facilitators for computational methods addressing tasks such as sentiment analysis, event detection, or entity recognition. However, obtaining, archiving, and semantically annotating large amounts of tweets is costly. In this paper, we describe TweetsCOV19, a publicly available knowledge base of currently more than 8 million tweets, spanning October 2019 - April 2020. Metadata about the tweets as well as extracted entities, hashtags, user mentions, sentiments, and URLs are exposed using established RDF/S vocabularies, providing an unprecedented knowledge base for a range of knowledge discovery tasks. Next to a description of the dataset and its extraction and annotation process, we present an initial analysis and use cases of the corpus.

A fast algorithm is proposed to tackle the constrained total generalised variation (TGV)‐based image‐restoration and reconstruction problems. The proposed algorithm proceeds by splitting: the non‐smooth constrained TGV model is first decomposed into several sub‐problems easier to solve, and then the linear gradient or proximity operators, including projections and shrinkages, of the sub‐problems are individually called without inner iteration. The algorithm is highly parallel since most of its steps can be executed simultaneously. Image‐restoration and reconstruction experiments demonstrate that the proposed algorithm outperforms several state‐of‐the‐art TV‐based methods both in accuracy and high‐speed efficiency. Besides, the proposed method efficiently suppresses staircase effects and presents better visual impression.

Multi-view clustering has received an increasing attention in many applications, where different views of objects can provide complementary information to each other. Existing approaches on multi-view clustering mainly focus on extending Non-negative Matrix Factorization (NMF) by enforcing the constraint over the coefficient matrices from different views in order to preserve their consensus. In this paper, we argue that it is more reasonable to utilize the high-level manifold consensus rather than the low-level coefficient matrix consensus to better capture the underlying clustering structure of the data. Moreover, it is also effective to utilize the sparse coding framework, instead of the NMF framework, to deal with the sparsity issue. To this end, we propose a novel approach, named Multiple Manifold Regularized Sparse Coding (MMRSC). Experimental results on two publicly available real-world image datasets demonstrate that our proposed approach can significantly outperform the state-of-the-art approaches for the multi-view image clustering task.

Due to their first-hand, diverse and evolution-aware reflection of nearly all areas of life, web archives are emerging as gold-mines for content analytics of many sorts. However, supporting search, which goes beyond navigational search via URLs, is a very challenging task in these unique structures with huge, redundant and noisy temporal content. In this paper, we address the search needs of expert users such as journalists, economists or historians for discovering a topic in time: Given a query, the top-k returned results should give the best representative documents that cover most interesting time-periods for the topic. For this purpose, we propose a novel random walk-based model that integrates relevance, temporal authority, diversity and time in a unified framework. Our preliminary experimental results on the large-scale real-world web archival collection shows that our method significantly improves the state-of-the-art algorithms (i.e., PageRank) in ranking temporal web pages.

A large number of images are continuously uploaded to popular photo sharing websites and online social communities. In this demonstration we show a novel application which automatically classifies images in a live photo stream according to their attractiveness for the community, based on a number of visual and textual features. The system effectively introduces an additional facet to browse and explore photo collections by highlighting the most attractive photographs and demoting the least attractive.

Social tags are known to be a valuable source of information for image retrieval and organization. However, contrary to the conventional document retrieval, rich tag frequency information in social sharing systems, such as Flickr, is not available, thus we cannot directly use the tag frequency (analogous to the term frequency in a document) to represent the relevance of tags. Many heuristic approaches have been proposed to address this problem, among which the well-known neighbor voting based approaches are the most effective methods. The basic assumption of these methods is that a tag is considered as relevant to the visual content of a target image if this tag is also used to annotate the visual neighbor images of the target image by lots of different users. The main limitation of these approaches is that they treat the voting power of each neighbor image either equally or simply based on its visual similarity. In this paper, we cast the social tag relevance learning problem as an adaptive teleportation random walk process on the voting graph. In particular, we model the relationships among images by constructing a voting graph, and then propose an adaptive teleportation random walk, in which a confidence factor is introduced to control the teleportation probability, on the voting graph. Through this process, direct and indirect relationships among images can be explored to cooperatively estimate the tag relevance. To quantify the performance of our approach, we compare it with state-of-the-art methods on two publicly available datasets (NUS-WIDE and MIR Flickr). The results indicate that our method achieves substantial performance gains on these datasets.

Ranking is an important problem in various applications, such as Information Retrieval (IR), natural language processing, computational biology, and social sciences. Many ranking approaches have been proposed to rank objects according to their degrees of relevance or importance. Beyond these two goals, diversity has also been recognized as a crucial criterion in ranking. Top ranked results are expected to convey as little redundant information as possible, and cover as many aspects as possible. However, existing ranking approaches either take no account of diversity, or handle it separately with some heuristics. In this paper, we introduce a novel approach, Manifold Ranking with Sink Points (MRSPs), to address diversity as well as relevance and importance in ranking. Specifically, our approach uses a manifold ranking process over the data manifold, which can naturally find the most relevant and important data objects. Meanwhile, by turning ranked objects into sink points on data manifold, we can effectively prevent redundant objects from receiving a high rank. MRSP not only shows a nice convergence property, but also has an interesting and satisfying optimization explanation. We applied MRSP on two application tasks, update summarization and query recommendation, where diversity is of great concern in ranking. Experimental results on both tasks present a strong empirical performance of MRSP as compared to existing ranking approaches.

Query recommendation plays a critical role in helping users' search. Most existing approaches on query recommendation aim to recommend relevant queries. However, the ultimate goal of query recommendation is to assist users to reformulate queries so that they can accomplish their search task successfully and quickly. Only considering relevance in query recommendation is apparently not directly toward this goal. In this paper, we argue that it is more important to directly recommend queries with high utility, i.e., queries that can better satisfy users' information needs. For this purpose, we propose a novel generative model, referred to as Query Utility Model (QUM), to capture query utility by simultaneously modeling users' reformulation and click behaviors. The experimental results on a publicly released query log show that, our approach is more effective in helping users find relevant search results and thus satisfying their information needs.

Query similarity calculation is an important problem and has a wide range of applications in IR, including query recommendation, query expansion, and even advertisement matching. Existing work on query similarity aims to provide a single similarity measure without considering the fact that queries are ambiguous and usually have multiple search intents. In this paper, we argue that query similarity should be defined upon search intents, so-called intent-aware query similarity. By introducing search intents into the calculation of query similarity, we can obtain more accurate and also informative similarity measures on queries and thus help a variety of applications, especially those related to diversification. Specifically, we first identify the potential search intents of queries, and then measure query similarity under different intents using intent-aware representations. A regularized topic model is employed to automatically learn the potential intents of queries by using both the words from search result snippets and the regularization from query co-clicks. Experimental results confirm the effectiveness of intent-aware query similarity on ambiguous queries which can provide significantly better similarity scores over the traditional approaches. We also experimentally verified the utility of intent-aware similarity in the application of query recommendation, which can suggest diverse queries in a structured way to search users.

Query recommendation has been considered as an effective way to help search users in their information seeking activities. Traditional approaches mainly focused on recommending alternative queries with close search intent to the original query. However, to only take relevance into account may generate redundant recommendations to users. It is better to provide diverse as well as relevant query recommendations, so that we can cover multiple potential search intents of users and minimize the risk that users will not be satisfied. Besides, previous query recommendation approaches mostly relied on measuring the relevance or similarity between queries in the Euclidean space. However, there is no convincing evidence that the query space is Euclidean. It is more natural and reasonable to assume that the query space is a manifold. In this paper, therefore, we aim to recommend diverse and relevant queries based on the intrinsic query manifold. We propose a unified model, named manifold ranking with stop points, for query recommendation. By turning ranked queries into stop points on the query manifold, our approach can generate query recommendations by simultaneously considering both diversity and relevance in a unified way. Empirical experimental results on a large scale query log of a commercial search engine show that our approach can effectively generate highly diverse as well as closely related query recommendations.