Information Processing & Management最新文献

Most existing expert finding methods in Community Question Answering (CQA) websites typically determine an expert’s suitability for answering one question based on their past answered questions. However, experts’ interests evolve over time, and their abilities to address questions vary. Consequently, effectively capturing the diverse interests and expertise of experts from their historical records poses a challenge due to dynamic preferences and varying abilities. In this paper, we propose an expert finding framework, which aims to capture experts’ diverse expertise and temporal-aware interests from their past answered questions. Specifically, we encode the timestamp and vote score information of each question answered by the expert for modeling their interests and expertise. Then, we design a personalized transformer encoder to effectively learn the inherent representation based on the expert’s historical answering behaviors. We further design an additive attention-based interaction encoder to dynamically capture the relevance between a target question and an expert’s historical answered questions. We conduct experiments on six real-world CQA datasets from StackExchange, the largest of which contains 8921912 questions and 687213 answerers. Experimental results show that on the metric P@1, compared with the best baseline methods, our method has achieved 3.3%–15.6% performance improvement.

Sign language is a form of visual communication employing hand gestures, body movements, and facial expressions. The growing prevalence of hearing impairment has driven the research community towards the domain of Continuous Sign Language Recognition (CSLR), which involves identification of successive signs in a video stream without prior knowledge of temporal boundaries. This survey article conducts a review of CSLR research, spanning the past 25 years, offering insights into the evolution of CSLR systems. A critical analysis of 126 studies is presented and organized into a taxonomy comprising seven critical dimensions: sign language, data acquisition, input modality, sign language cues, recognition techniques, utilized datasets, and overall performance. Additionally, the article investigated the classification of deep-learning CSLR models, categorizing them based on spatial, temporal, and alignment methods, while identifying their advantages and limitations. The article also explored various research aspects including CSLR challenges, the significance of non-manual features in CSLR systems, and identified gaps in existing literature. This literature taxonomy serves as a resource aiding researchers in the development and positioning of novel CSLR techniques. The study emphasizes the efficacy of multi-modal deep learning systems in capturing diverse sign language cues. However, the examination of existing research uncovers numerous limitations, calling for continued research and innovation within the CSLR domain. The findings not only contribute to the broader understanding of sign language recognition but also lay the foundations for future research initiatives aimed at addressing the persistent challenges within this emerging field.

Software estimation rates are still suboptimal regarding efficiency, runtime, and the accuracy of model predictions. Graph Neural Networks (GNNs) are complex, yet their precise forecasting reduces the gap between expected and actual software development efforts, thereby minimizing associated risks. However, defining optimal hyperparameter configurations remains a challenge. This paper compares state-of-the-art models such as Long-Short-Term-Memory (LSTM), Graph Gated Neural Networks (GGNN), and Graph Gated Sequence Neural Networks (GGSNN), and conducts experiments with various hyperparameter settings to optimize performance. We also aim to gain the most informative feedback from our models by exploring insights using a post-hoc agnostic method like Shapley Additive Explanations (SHAP). Our findings indicate that the Taguchi orthogonal array optimization method is the most computationally efficient, yielding notably improved performance metrics. This suggests a compromise between computational efficiency and prediction accuracy while still requiring the lowest number of runnings, with an RMSE of 0.9211 and an MAE of 310.4. For the best-performing model, the GGSNN model, within the Constructive Cost Model (COCOMO), Function Point Analysis (FPA), and Use Case Points (UCP) frameworks, applying the SHAP method leads to a more accurate determination of relevance, as evidenced by the norm reduction in activation vectors. The SHAP method stands out by exhibiting the smallest area under the curve and faster convergence, indicating its efficiency in pinpointing concept relevance.

The considerable semantic information contained in multimodal data is increasingly appreciated by industry and academia. To effectively leverage multimodal information, existing multimodal recommendation methods mainly build multimodal auxiliary graphs to improve the representation of users and items. However, the weak value density of multimodal data inevitably leads to serious noise issues, making it difficult to effectively exploit valuable information from the multimodal contents. To address this issue, we propose a novel Fine-grained Self-supervised Mom entum Distillation Framework (FUMMER) for multimodal recommendations. Specifically, we propose a Transformer-based Fine-grained Feature Extractor (TFFE) and a Momentum Distillation (MoD) structure that incorporates intra- and inter-modal contrastive learning to fully pre-train TFFE for fine-grained feature extraction. Additionally, we design a structure-aware fine-grained contrastive learning module to fully exploit the self-supervised signals from fine-grained structural features. Extensive experiments on three real-world datasets show that our method outperforms state-of-the-art multimodal recommendation methods. Further experiments verify that the fine-grained feature extraction method we propose can serve as a pre-trained model, enhancing the performance of recommendation methods effectively by learning the fine-grained feature representations of items. The code is publicly available at https://github.com/BIAOBIAO12138/FUMMER.

Predicting users’ interests on social networks is gaining attention due to its potential to cater customized information and services to the end users. Although previous works have extensively explored how users’ interests can be modeled on social networks, there has been limited investigation into the prediction of users’ future interests. The objective of our work in this paper is to empirically study the effectiveness of different sets of features based on users’ past social interactions, historical interests and their temporal dynamics to predict their interests over a collection of future-yet-unobserved topics. More specifically, we introduce and formalize the features for interest prediction in four categories: user-based, topical, explicit user-topic engagement, and friends’ influence. We further explore the influence of temporality by augmenting features with information pertaining to users’ historical interests and social connections. We model the task of future interest prediction as a learning-to-rank problem where different features and their related categories are ranked based on their relevance and performance in interest prediction, and investigate the efficiency of different features individually and comparatively for predicting the future interest of users with different activity levels in social networks over on unobserved topics. After conducting experiments on a real-world dataset sourced from Twitter, we have identified several noteworthy findings: (1) relevance feature in the category of past explicit user-topic engagement is the strongest indicator for predicting user’s future interest across all user groups, with an observed 8.57% decrease in NDCG and an 8.95% decrease in MAP when it is removed in the ablation study. (2) the observation of an 8.06% decrease in NDCG and a 7.3% decrease in MAP, when topical features such as popularity, freshness, and coherence are removed in the ablation study, highlights their significance as among the strongest indicators for users’ future interest, particularly for low-active users. (3) although temporal features show a clear positive impact across user groups with varying levels of activity (resulting in a 4.5% decrease in NDCG and a 7.3% decrease in MAP when removed in the ablation study), the temporal topical features do not demonstrate a significant positive effect, and 4) The removal of user-specific characteristics such as influence and personality traits in the ablation study reveals their significant impact in predicting future interest over cold topics, reflected by a 5.49% decrease in NDCG and a 5.72% decrease in MAP. Our findings make significant contributions to the field of future interest prediction, offering valuable insights and practical implications for various applications in social network analysis.

Sorting any $m$ target suffixes of an input string $X$ of $n$ characters from a constant alphabet is a key task for building the sparse suffix array $SSA\left(X\right)$ for index construction. A number of probabilistic and deterministic algorithms have been proposed for sorting sparse suffixes with varying time and space complexities, but only some experimental results are available for performance evaluation of these algorithms. We design a divide-and-conquer algorithm called sSAIS for computing $SSA\left(X\right)$ in $O(nlogmlog(n/m))$ time and $O\left(m\right)$ workspace by using the induced sorting principle, and conduct an experimental performance study on real and artificial datasets. This work reveals that to design an efficient deterministic algorithm for sorting sparse suffixes is a tough challenge and the density of target suffixes might be considered as a critical design parameter.

Multi-modal image clustering focuses on exploring and exploiting related information across various modals of input images to obtain clear image cluster patterns. Recent multi-modal/view clustering methods have shown promising performance in solving the image clustering problem. However, most existing methods fail to properly deal with the multi-modal image data with massive number of clusters and high dimensionality of features in real-world applications, such as image retrieval, multi-modal autonomous driving perception and industrial automation. We call this problem “Big Clusters and Features” for short just as “Big Data” for large number of samples. To fix this challenging problem, we in this article design a general multi-modal image clustering framework which integrates cluster-wise weight learning, feature learning, and clustering structure learning. Under this framework, we further propose a new Cluster-weighted Multi-modal Information Bottleneck Co-clustering (CMIBC) method, and it could effectively measure image cluster importance information and discriminative features of each modal for obtaining satisfactory image clustering performance. Unlike existing cluster weight learning methods only considering intra-cluster similarity or cross-cluster dissimilarity, we design a novel cluster-wise weight learning strategy by jointly considering and enjoying the best of both worlds. Many carefully designed experiments on various multi-modal image datasets with big clusters and features reveal the competitive advantages of the CMIBC algorithm over lots of compared single/multi-modal clustering methods, particularly the notable improvement of 3.12% and 5.28% on the Leaves Plant Species dataset in terms of accuracy and normalized mutual information. Owing to the promising performance, the proposed CMIBC may be extended to many other practical applications, e.g., multi-modal medical analysis and video recognition.

In the domain of complex networks, the identification of influential nodes plays a crucial role in ensuring network stability and facilitating efficient information dissemination. Although the study of influential nodes has been applied in many fields such as suppression of rumor spreading, regulation of group behavior, and prediction of mass events evolution, current deep learning-based algorithms have limited input features and are incapable of aggregating neighbor information of nodes, thus failing to adapt to complex networks. We propose an influential node identification method in complex networks based on the Transformer. In this method, the input sequence of a node includes information about the node itself and its neighbors, enabling the model to effectively aggregate node information to identify its influence. Experiments were conducted on 9 synthetic networks and 12 real networks. Using the SIR model and a benchmark method to verify the effectiveness of our approach. The experimental results show that this method can more effectively identify influential nodes in complex networks. In particular, the method improves 27 percent compared to the second place method in network Netscience and 21 percent in network Faa.

Brand image assessment and maintenance are essential for branding. This paper constructs an online data-driven quantitative brand image assessment method using the classic brand image theory as a conceptual model. The method is organized as follows. First, with domain expert knowledge and deep learning, this paper constructs a task ontology to clearly describe the brand image constituent content, constituent relationship, properties, and property values. Then, using the task ontology as a priori knowledge, we identify the content of brand associations from User-generated Content (UGC) and Firm-generated content (FGC), respectively, and calculate the associations’ favorability, strength and uniqueness; classify brand associations into three categories: functional, experiential, and symbolic to achieve a dual-perspectives (consumer perceptions & corporate claims) brand image assessment. Finally, this study compares the dual-perspective brand images from the components and benefits to construct a brand image communication and maintenance strategy. The development and validation of the methodology take the Chinese New Energy Vehicle (NEV) market as the analysis object. The proposed dual-perspective brand image quantitative assessment model is a new development of brand image evaluation and maintenance theoretical method in the digital era. It is also a practical tool for brand management in enterprises.

Effective retrieval is essential for finding resources in demand handily amidst extensive data records in data warehouse. Mainstream map retrieval methods suffer from intention gap problem and are incapable to describe sophisticated user demands precisely due to the limits of low- and middle-level text or visual feature matching, resulting in unsatisfactory retrieval results. Such limitations are more marked when map retrieval demands were characterized with joint constraints of geographic concepts. To address this issue, we propose a map retrieval intention recognition method to perceive user demands with relevance feedback samples and geographic semantics guidance. Specifically, we construct a hierarchical intention expression model to describe retrieval goals and their multi-dimensional attribute constrains; incorporate geographic ontologies to provide semantic guidance and facilitate recognition; utilize the frequent itemset mining (FIM) algorithm Apriori to generate intention candidates from relevance feedback samples, and search for the optimal intention set by adopting the minimum description length (MDL) principle. The experiments verify the effectiveness of Apriori algorithm and MDL principle on intention recognition. The proposed method outperforms the FIM algorithm Gene Ontology (RuleGO) and the Decision Tree algorithm with Hierarchical Features (DTHF) with higher recognition accuracy and noise tolerance. Furthermore, through our sample augmentation strategy, the method yields promising recognition accuracy even when the feedback sample size is as low as ten, substantially reducing the feedback burden in human-computer interactions. We envision that the application of our method in spatial data infrastructures (SDIs), such as geoportals and catalogue services, could enhance the quality of service and user experience in geospatial data discovery.