The Retrieval-Augmented Language Model (RALM) has demonstrated remarkable performance on knowledge-intensive tasks by integrating external knowledge during inference, which mitigates the factual hallucinations inherited in large language models (LLMs). Despite these advancements, challenges persist in the implementation of RALMs, particularly in terms of reliability and traceability. Specifically, the irrelevant document retrieval may result in unhelpful responses or even deteriorate the performance of LLMs, while the lack of appropriate citations in outputs complicates efforts to verify the trustworthiness of the models. To this end, we propose a novel self-reasoning framework aimed at improving the reliability and traceability of RALMs, whose core idea is to leverage reasoning trajectories generated by the LLM itself. The framework involves constructing self-reasoning trajectories through three processes: a relevance-aware process, an evidence-aware selective process, and a trajectory analysis process. We evaluated our framework across four public datasets (two short-form QA datasets, one long-form QA dataset, and one fact verification dataset) to demonstrate its superiority. Our method can outperform existing state-of-the-art models and achieve performance comparable with GPT-4, using only 2,000 training samples.

Leveraging users' long engagement histories is essential for personalized content recommendations. The success of pretrained language models (PLMs) in NLP has led to their use in encoding user histories and candidate items, framing content recommendations as textual semantic matching tasks. However, existing works still struggle with processing very long user historical text and insufficient user-item interaction. In this paper, we introduce a content-based recommendation framework, SPAR, which effectively tackles the challenges of holistic user interest extraction from the long user engagement history. It achieves so by leveraging PLM, poly-attention layers and attention sparsity mechanisms to encode user's history in a session-based manner. The user and item side features are sufficiently fused for engagement prediction while maintaining standalone representations for both sides, which is efficient for practical model deployment. Moreover, we enhance user profiling by exploiting large language model (LLM) to extract global interests from user engagement history. Extensive experiments on two benchmark datasets demonstrate that our framework outperforms existing state-of-the-art (SoTA) methods.

Many real-world networks have been evolving and are finely modeled as temporal graphs from the viewpoint of the graph theory. A temporal graph is informative and always contains two types of features, i.e., the temporal feature and topological feature, where the temporal feature is related to the establishing time of the relationships in the temporal graph, and the topological feature is influenced by the structure of the graph. In this paper, considering both these two types of features, we perform time-topology analysis on temporal graphs to analyze the cohesiveness of temporal graphs and extract cohesive subgraphs. Firstly, a new metric named $$\mathbb {T}$$ -cohesiveness is proposed to evaluate the cohesiveness of a temporal subgraph from the time and topology dimensions jointly. Specifically, given a temporal graph $$\mathcal {G}_s = (V_s, \mathcal {E}_s)$$ , cohesiveness in the time dimension reflects whether the connections in $$\mathcal {G}_s$$ happen in a short period of time, while cohesiveness in the topology dimension indicates whether the vertices in $$V_s$$ are densely connected and have few connections with vertices out of $$\mathcal {G}_s$$ . Then, $$\mathbb {T}$$ -cohesiveness is utilized to perform time-topology analysis on temporal graphs, and two time-topology analysis methods are proposed. In detail, $$\mathbb {T}$$ -cohesiveness evolution tracking traces the evolution of the $$\mathbb {T}$$ -cohesiveness of a subgraph, and combo searching finds out cohesive subgraphs containing the query vertex, which have $$\mathbb {T}$$ -cohesiveness values larger than a given threshold. Moreover, since combo searching is NP-hard, a pruning strategy is proposed to estimate the upper bound of the $$\mathbb {T}$$ -cohesiveness value, and then improve the efficiency of combo searching. Experimental results demonstrate the efficiency of the proposed time-topology analysis methods and the pruning strategy. Besides, four more definitions of $$\mathbb {T}$$ -cohesiveness are compared with our method. The experimental results confirm the superiority of our definition.

Chronic obstructive pulmonary disease (COPD) is a serious chronic respiratory disease. Improving the ability to identify patients with COPD in primary medical institutions is important to prevent and treat the disease. With the continuous development of medical digitization, the application of big data informatization in the medical and health fields has become possible. Recently, applying innovative technologies such as big data analysis, machine learning, and artificial intelligence-assisted decision-making in the medical field has become an interdisciplinary research hotspot. Based on the identification and diagnosis of COPD in the high-risk population, this study proposes a convenient and effective clinical decision support system to help identify patients with COPD in primary health institutions. The results of the preliminary experiments show that the proposed method is convenient and effective compared with the existing methods.

Many real-world networks have been evolving, and are finely modeled as temporal graphs from the viewpoint of the graph theory. A temporal graph is informative, and always contains two types of information, i.e., the temporal information and topological information, where the temporal information reflects the time when the relationships are established, and the topological information focuses on the structure of the graph. In this paper, we perform time-topology analysis on temporal graphs to extract useful information. Firstly, a new metric named T-cohesiveness is proposed to evaluate the cohesiveness of a temporal subgraph. It defines the cohesiveness of a temporal subgraph from the time and topology dimensions jointly. Specifically, given a temporal graph G s = ( Vs , ε Es ), cohesiveness in the time dimension reflects whether the connections in G s happen in a short period of time, while cohesiveness in the topology dimension indicates whether the vertices in V s are densely connected and have few connections with vertices out of G s . Then, T-cohesiveness is utilized to perform time-topology analysis on temporal graphs, and two time-topology analysis methods are proposed. In detail, T-cohesiveness evolution tracking traces the evolution of the T-cohesiveness of a subgraph, and combo searching finds out all the subgraphs that contain the query vertex and have T-cohesiveness larger than a given threshold. Moreover, a pruning strategy is proposed to improve the efficiency of combo searching. Experimental results confirm the efficiency of the proposed time-topology analysis methods and the pruning strategy.

Importance. The last decade has witnessed the advances of cognitive computing technologies that learn at scale and reason with purpose in medicine studies. From the diagnosis of diseases till the generation of treatment plans, cognitive computing encompasses both data-driven and knowledge-driven machine intelligence to assist health care roles in clinical decision-making. This review provides a comprehensive perspective from both research and industrial efforts on cognitive computing-based CDSS over the last decade. Highlights. (1) A holistic review of both research papers and industrial practice about cognitive computing-based CDSS is conducted to identify the necessity and the characteristics as well as the general framework of constructing the system. (2) Several of the typical applications of cognitive computing-based CDSS as well as the existing systems in real medical practice are introduced in detail under the general framework. (3) The limitations of the current cognitive computing-based CDSS is discussed that sheds light on the future work in this direction. Conclusion. Different from medical content providers, cognitive computing-based CDSS provides probabilistic clinical decision support by automatically learning and inferencing from medical big data. The characteristics of managing multimodal data and computerizing medical knowledge distinguish cognitive computing-based CDSS from other categories. Given the current status of primary health care like high diagnostic error rate and shortage of medical resources, it is time to introduce cognitive computing-based CDSS to the medical community which is supposed to be more open-minded and embrace the convenience and low cost but high efficiency brought by cognitive computing-based CDSS.

Text-based diagnosis classification is a critical problem in AI-enabled healthcare studies, which assists clinicians in making correct decision and lowering the rate of diagnostic errors. Previous studies follow the routine of sequence based deep learning models in NLP literature to deal with clinical notes. However, recent studies find that structural information is important in clinical contents that greatly impacts the predictions. In this paper, a novel sequence-to-subgraph framework is introduced to process clinical texts for classification, which changes the paradigm of managing texts. Moreover, a new classification model under the framework is proposed that incorporates subgraph convolutional network and hierarchical diagnostic attentive network to extract the layered structural features of clinical texts. The evaluation conducted on both the real-world English and Chinese datasets shows that the proposed method outperforms the state-of-the-art deep learning based diagnosis classification models.

In the industry of medical intelligence, classification is one of the most common tasks. It appears in various medical jobs, such as triage, diagnosis, and pathologic analysis. Many classification algorithms studied in machine learning can be chosen to help solve these tasks. However, due to the special nature of the medical industry, its data sets show a character of imbalance. Namely, the data are skewed distributed in different classes. Unfortunately, the classification problem of imbalanced data has a reputation of classic and hard-to-solve in data mining and artificial intelligence research community. What's worse, most proposed classification methods are designed to deal with binary classification case, while the common scenario in medical intelligence applications is multi-classification. To deal with this, a pre-processing structure called Cost-Sensitive Variable Neighbour Search (CSVNS) is proposed in this paper. It combines the ideas of sampling and cost-sensitive, which are two most commonly used strategies for multi-class imbalanced data classification tasks. As for the sampling process, a double-stack Variable Neighbour Search (VNS) structure is introduced and 15 different neighborhood structures are designed to help optimizing the process. Also, the classes are allocated different weights to improve the classifier's classification capacity. In the experiment part, the proposed method is evaluated on 4 medical data sets. $G$ - mean and mAUC are selected to represent the method's performance in medical classification tasks. Experimental results show the proposed method outperforms the classic methods in most situations. In the end, 3 extra data sets are tested to demonstrate the algorithms' scalability.

The lesion detection on medical images has been a challenging problem in computer vision due to the lack of high-quality annotations at the bounding box level and the difficulty to identify the lesion on gray-scale medical images which are commonly used in radiology tests like X-ray, CT and MRI. In this paper, we propose a novel framework of medical image enhancement based on the class-aware attention weight extraction and the deep colorization of medical images with generative adversarial networks motivated by human visual characteristics and cell staining. The evaluation conducted on the real public medical image datasets proves that, the performance of lesion detection based on the existing detectors is improved after enhancing the medical images with the proposed enhancement framework.

Training a classifier based on imbalanced data set is considered a great challenge in classification tasks, as classifiers are often "biased" due to highly skewed data distribution and overlapping borderline between different classes. When the imbalanced data appears in the multi-class classification scenario, the classification difficulty increases exponentially. In this paper, we propose an integrated approach to handle imbalanced multi-class classification by combining the population-based sampling method and a multi-expert classifier. In the implementation, we choose the Ant Colony Optimization to realize the sampling process. As for the classifier, the voting mechanism is applied to intensify the weak classifiers. To test the algorithm’s performance, we choose 10 representative imbalanced multi-class data sets from the UCI Machine Learning Repository. G – mean and mAUC are chosen as the metrics. According to the experimental results, the proposed algorithm dominates in 8 data sets and gets a second place in 1 data set when evaluated by G – mean, and ranks first in 3 data sets and top-3 among the most for mAUC.

The automatic text-based diagnosis remains a challenging task for clinical use because it requires appropriate balance between accuracy and interpretability. In this paper, we attempt to propose a solution by introducing a novel framework that stacks Bayesian Network Ensembles on top of Entity-Aware Convolutional Neural Networks (CNN) towards building an accurate yet interpretable diagnosis system. The proposed framework takes advantage of the high accuracy and generality of deep neural networks as well as the interpretability of Bayesian Networks, which is critical for AI-empowered healthcare. The evaluation conducted on the real Electronic Medical Record (EMR) documents from hospitals and annotated by professional doctors proves that, the proposed framework outperforms the previous automatic diagnosis methods in accuracy performance and the diagnosis explanation of the framework is reasonable.

The automatic diagnosis has been suffering from the problem of inadequate reliable corpus to train a trustworthy predictive model. Besides, most of the previous deep learning based diagnosis models adopt the sequence learning techniques (CNN or RNN), which is difficult to extract the complex structural information, e.g. graph structure, between the critical medical entities. In this paper, we propose to build the diagnosis model based on the high-standard EMR documents from real hospitals to improve the accuracy and the credibility of the resulting model. Meanwhile, we introduce the Graph Convolutional Network into the model that alleviates the sparse feature problem and facilitates the extraction of structural information for diagnosis. Moreover, we propose the mutual attentive network to enhance the representation of inputs towards the better model performance. The evaluation conducted on the real EMR documents demonstrates that the proposed model is more accurate compared to the previous sequence learning based diagnosis models. The proposed model has been integrated into the information systems in over hundreds of primary health care facilities in China to assist physicians in the diagnostic process.

Social networks facilitate communication among Internet users while generating large volumes of online short-text conversations every day. This leads to a huge number of free-style asynchronous conversations where multiple users are involved and multiple topics are discussed at the same time in the same place, e.g., an instant group chat in WeChat. Here emerges an interesting problem: as a result of a large number of users and topics, the conversation structure may get into a mess, which often interferes with the acquisition of messages users are interested in. For example, when a user enters a conversation, (s)he usually does not want to read all the historical messages, but just hope to get the messages that are the most relevant to some messages (s)he cares about. Therefore, it is an essential task to understand the logical correlations among messages, which benefits text mining, natural language processing, and web intelligence techniques. In this paper, we focus on “reply-to” relations, such as Q&A between messages in group chats. At first, a model called LSTM-RT is presented to predict the “reply-to” relations between messages, which is based on deep bidirectional LSTM networks. Then, three versions of the LSTM-RT model are proposed. In detail, the first version is based on a non-siamese architecture, which processes ordered message pairs; The other two versions are end-to-end models, which are based on the word level and the sentence level, respectively. Finally, experimental results conducted on two real-world group chat data sets demonstrate the effectiveness of the proposed model.

Medical Question Answering (medical QA), which studies the problem of automatically answering patients' medical questions online, is one of the major applications of bioinformatics. Though many efforts have been made before, the medical QA system still deserves delicate algorithm optimization due to the serious application scenario and strict requirement for the answer quality. In this paper, we introduce a novel Knowledge Abstraction Matching (KAM) method for the medical QA problem. The intuition of KAM is that there are many frequent repeat text segments appearing in the answers across different questions. From this view, we propose a new method that consists of frequent segment N-gram mining, medical knowledge abstraction, medical segment matching and answer re-retrieval. KAM has been incorporated into Baidu's enterprise medical QA system MelodyQA deployed on the backend of Muzhi Doctor. The evaluation shows that the proposed method can generate more quality answers for MelodyQA with a significant improvement of question coverage under acceptable accuracy.

To stretch a music piece to a given length is a common demand in people's daily lives, e.g., in audio-video synchronization and animation production. However, it is not always guaranteed that the stretched music piece is acceptable for general audience since music stretching suffers from people's perceptual artefacts. Over-stretching a music piece will make it uncomfortable for human psychoacoustic hearing. The research on music stretching resistance attempts to estimate the maximum stretchability of music pieces to further avoid over-stretch. It has been observed that musical genres can significantly improve the accuracy of automatic estimation of music stretching resistance, but how musical genres are related to music stretching resistance has never been explained or studied in detail in the literature. In this paper, the characteristics of music stretching resistance are compared across different musical genres. It is found that music stretching resistance has strong intra-genre cohesiveness and inter-genre discrepancies in the experiments. Moreover, the ambiguity and the symmetry of music stretching resistance are also observed in the experimental analysis. These findings lead to a new measurement on the similarity between different musical genres based on their music stretching resistance. In addition, the analysis of variance (ANOVA) also supports the findings in this paper by verifying the significance of musical genre in shaping music stretching resistance.

Social networks enjoy great popularity among Internet users while generating large volumes of online short-text conversations every day. It leads to a huge number of free-style asynchronous conversations where multiple users are involved and multiple topics are discussed at the same time in the same place, e.g., an instant group chat in WeChat. Here emerges an interesting problem: As a result of a large number of users and topics, the conversation structure may get into a mess, which interferes with our access to the messages we are interested in. For example, when we open the chat records, we do not want to read all the historical messages. We just want to get the messages that are the most relevant with the messages we care about. Therefore, it is an essential task to understand the logical correlations among messages, which benefits the text mining, the natural language processing and the web intelligence techniques. In this paper, we present the concept of "reply-to" relations to capture most kinds of logical correlations between messages, such as Q&A or complement. Also, we propose a model called LSTM-RT to predict the "reply-to" relations between messages, which is based on the high-quality vector representations of words and LSTM networks. In addition, we give two versions of LSTM-RT based on word level and sentence level, respectively. Experiments conducted on two real-world group chat datasets demonstrate the effectiveness of our proposed models.

Among the previous studies in modeling users' preference on images, most of them assume there is a consistent ranking of images, and users' preference is transitive. That is, if a user likes image A over B and B over C, it must have A over C for this user. This condition holds when user compares images from a single angle. However, if there are multiple angles to consider, users' preference may not be transitive at all. Thus, it is interesting to know whether users' pairwise preference on images can be intransitive, and how can such personalized intransitivity be modeled. In this paper, the problem of modeling the intransitive pairwise image preference is studied, and a family of multi-angle preference models based on pairwise image comparisons are proposed to tackle this problem. These models separately consider users' most-like angles, least-like angles and the balance between two sides towards different images. This study attempts to find the angles that each user selects to evaluate images as well as the angles that each image attracts general users. The experimental results on two real-world data sets show superior performance of the proposed models.

The online social networks have embraced huge success from the crowds in the last two decades. Now, more and more people get used to chat with friends online via instant messaging applications on personal computers or mobile devices. Since these conversations are sequentially organized, which fails to show the logical relations between messages, they are called asynchronous conversations in previous studies. Unfortunately, the sequential layouts of messages are usually not intuitive to see how the conversation evolves as time elapses. In this paper, we propose to learn the structures of online asynchronous conversations by predicting the "reply-to" relation between messages based on text similarity and latent semantic transferability. A heuristic method is also brought forward to predict the relation, and then recover the conversation structure. We demonstrate the effectiveness of the proposed method through experiments on a real-world web forum comment data set.

Most of the previous studies on the user preferences assume that there is a personal transitive preference ranking of the consumable media like images. For example, the transitivity of preferences is one of the most important assumptions in the recommender system research. However, the intransitive relations have also been widely observed, such as the win/loss relations in online video games, in sport matches, and even in rock-paper-scissors games. It is also found that different subjects demonstrate the personalized intransitive preferences in the pairwise comparisons between the applicants for college admission. Since the intransitivity of preferences on images has barely been studied before and has a large impact on the research of personalized image search and recommendation, it is necessary to propose a novel method to predict the personalized intransitive preferences of images. In this paper, we propose the novel Multi-Criterion preference (MuCri) models to predict the intransitive relations in the image preferences. The MuCri models utilize different kinds of image content features as well as the latent features of users and images. Meanwhile, a new data set is constructed in this paper, in order to evaluate the performance of the MuCri models. The experimental evaluation shows that the MuCri models outperform all the baselines. Due to the interdisciplinary nature of this topic, we believe it would widely attract the attention of researchers in the image processing community as well as in other communities, such as machine learning, multimedia, and recommender system.

Similarity join is one of the hottest topics in the field of data management,and it has been widely applied in many fields.However,existing similarity join methods cannot meet the increasing demands in the real world.This paper define generalized bisimilarity join as a new similarity join to expend the applications of the similarity join research by introducing the satisfaction operator on various data types with individual thresholds.Two efficient methods,SJS (sub-join set) and MFV (mapping-filtering-verification),are proposed to solve this problem.A large amount of experiments conducted on both real-world and synthetic datasets demonstrate the correctness and the effectiveness of the proposed methods.