Data Sparsity Problem Research Articles

Data Augmentation with Nearest Neighbor Classifier for Few-Shot Named Entity Recognition.

Few-shot learning (FSL) is a category of machine learning models that are designed with the intent of solving problems that have small amounts of labeled data available for training. FSL research progress in natural language processing (NLP), particularly within the medical domain, has been notably slow, primarily due to greater difficulties posed by domain-specific characteristics and data sparsity problems. We explored the use of novel methods for text representation and encoding combined with distance-based measures for improving FSL entity detection. In this paper, we propose a data augmentation method to incorporate semantic information from medical texts into the learning process and combine it with a nearest-neighbor classification strategy for predicting entities. Experiments performed on five biomedical text datasets demonstrate that our proposed approach often outperforms other approaches.

Geography-aware Heterogeneous Graph Contrastive Learning for Travel Recommendation

Recommendation system concentrates on quickly matching products to consumer’s needs which plays a major role in improving user experiences and increase conversion rate. Travel recommendation has become a hot topic in both industry and academia with the development of the tourism industry. Nevertheless, the selection of travel products entails careful consideration of various geographical factors, such as departure and destination. Meanwhile, due to the limitation of finance and time, users browse and purchase travel products less frequently than they do for traditional products, which leads to data sparsity problem in representation learning. To solve these challenges, a novel model named GHGCL (short for G eography-aware H eterogeneous G raph C ontrastive L earning) is proposed for recommending travel products. Concretely, we model the travel recommender system as an heterogeneous information network with geographical information, and capture diverse user preferences from local and high-order structures. Especially, we design two kinds of contrastive learning tasks for better user and travel product representation learning. The multi-view contrastive learning aims to bridge the gap between network schema and meta-path view representations. The meta-path contrastive learning focuses on modeling the coarse-grained commonality between different meta-paths from the perspective of different geographical factors, i.e., departure and destination. We assess the performance of GHGCL by performing a series of experiments on a real-world dataset and the results clearly verify its superiority as compared to the baseline methods.

MKNBL: Joint multi-channel knowledge-aware network and broad learning for sparse knowledge graph-based recommendation

To address data sparsity and cold start problems in recommender systems, existing studies have utilized knowledge graph (KG) as side information to enhance the performance of recommendation. However, existing methods often explore side information only from one perspective, leading to limited expressive power in the representations of users and items. Some researchers have also combined broad learning system (BLS) with collaborative filtering (CF) to achieve efficient training. Yet the improvement of these approaches is influenced by the availability of data. These approaches perform poorly in cases of data sparsity. To solve these limitations, this study proposes a two-stage KG-based recommendation method (MKNBL) that integrates multi-channel knowledge-aware network (MKN) and BLS. The first stage focuses on exploring valuable side information. We adopt MKN to learn multi-view knowledge from KG. This enables recommender systems to widely utilize KG as side information to mitigate the limitations of data sparsity and cold start problems. The second stage focuses on efficient data enhancement learning. We employ BLS to enhance the valuable information discovered in the previous step, aiming to achieve better recommendation accuracy. Finally, we validate the high diversity and accuracy of MKNBL through experiments. Compared with other state-of-the-art (SOTA) methods, MKNBL achieves improvements of 7.50%, 3.06%, 5.35% and 3.07% in mean absolute error (MAE) in the Movielens-1M, Last.FM, Book-Crossing and Amazon datasets, respectively.

Explicit Behavior Interaction with Heterogeneous Graph for Multi-behavior Recommendation

Multi-behavior recommendation systems exploit multi-type user–item interactions (e.g., clicking, adding to cart and collecting) as auxiliary behaviors for user modeling, which can alleviate the problem of data sparsity faced by traditional recommendation systems. The key point of multi-behavior recommendation systems is to make full use of the auxiliary behavior information for the learning of user preferences. However, there are two challenges in existing methods that need to be explored: (1) capturing personalized user preferences based on multiple auxiliary behaviors, especially for negative feedback signals; and (2) explicitly modeling the semantics between auxiliary and target behaviors, and learning the explicit interactions between multiple behaviors. To tackle the two problems described above, we propose a novel model, called explicit behavior interaction with heterogeneous graph for multi-behavior recommendation (MB-EBIH). In particular, we first construct a heterogeneous behavior graph, including both positive and negative behaviors. A pre-trained model based on graph neural network (GNN) is then used to generate explicit behavior interaction values as the edge weights for the heterogeneous behavior graph. These weights reflect the importance of each of the auxiliary behaviors in an explicit manner. Finally, the extracted explicit behavior interaction information is incorporated into the multi-behavior user–item bipartite graphs to learn better representations. Experimental results on four real-world datasets demonstrate the effectiveness of our model in terms of exploring multi-behavioral data; and ablation and analysis experiments further demonstrate the effectiveness of explicit behavior interaction information.

A Metric Learning Perspective on the Implicit Feedback-Based Recommendation Data Imbalance Problem

Paper recommendation systems are important for alleviating academic information overload. Such systems provide personalized recommendations based on implicit feedback from users, supplemented by their subject information, citation networks, etc. However, such recommender systems face problems like data sparsity for positive samples and uncertainty for negative samples. In this paper, we address these two issues and improve upon them from the perspective of metric learning. The algorithm is modeled as a push–pull loss function. For the positive sample pull-out operation, we introduce a context factor, which accelerates the convergence of the objective function through the multiplication rule to alleviate the data sparsity problem. For the negative sample push operation, we adopt an unbiased global negative sample method and use an intermediate matrix caching method to greatly reduce the computational complexity. Experimental results on two real datasets show that our method outperforms other baseline methods in terms of recommendation accuracy and computational efficiency. Moreover, our metric learning method that introduces context improves by more than 5% over the element-wise alternating least squares method. We demonstrate the potential of metric learning in addressing the problem of implicit feedback recommender systems with positive and negative sample imbalances.

Causal inference of diachronic semantic maps from cross-linguistic synchronic polysemy data

Semantic maps are used in lexical typology to summarize cross-linguistic implicational universals of co-expression between meanings in a domain. They are defined as networks which, using as few links as possible, connect the meanings so that every isolectic set (i.e., set of meanings that can be expressed by the same word in some language) forms a connected component. Due to the close connection between synchronic polysemies and semantic change, semantic maps are often interpreted diachronically as encoding potential pathways of semantic extension. While semantic maps are traditionally generated by hand, there have been attempts to automate this complex and non-deterministic process. I explore the problem from a new algorithmic angle by casting it in the framework of causal discovery, a field which explores the possibility of automatically inferring causal structures from observational data. I show that a standard causal inference algorithm can be used to reduce cross-linguistic polysemy data into minimal network structures which explain the observed polysemies. If the algorithm makes its link deletion decisions on the basis of the connected component criterion, the skeleton of the resulting causal structure is a synchronic semantic map. The arrows which are added to some links in the second stage can be interpreted as expressing the main tendencies of semantic extension. Much of the existing literature on semantic maps implicitly assumes that the data from the languages under analysis is correct and complete, whereas in reality, semantic map research is riddled by data quality and sparseness problems. To quantify the uncertainty inherent in the inferred diachronic semantic maps, I rely on bootstrapping on the language level to model the uncertainty caused by the given language sample, as well as on random link processing orders to explore the space of possible semantic maps for a given input. The maps inferred from the samples are then summarized into a consensus network where every link and arrow receives a confidence value. In experiments on cross-linguistic polysemy data of varying shapes, the resulting confidence values are found to mostly agree with previously published results, though challenges in directionality inference remain.

Download Behavior Analysis Based on Google Play Store Data

As mobile apps continue to grow, app stores, as the main channel for users to download apps, are becoming increasingly important for developers and platforms. Understanding users 'download behavior and accurately predicting users' preferences and needs can effectively improve the effect of the recommendation system in the application mall, improve user experience and improve the download conversion rate. However, traditional rule-based recommender systems often face the problems of data sparsity and model complexity. Therefore, it is an urgent and valuable topic to analyze user download behavior combined with machine learning technology and to provide personalized recommendations and services. This study uses the download behavior information of Google Play Store users in 2018, and use three classic machine algorithms (linear regression, random forest and SVM) to model and predict the software rating, dig deep into various factors affecting the rating, and gain deep insight into users' preferences and behavior patterns. This will provide more accurate recommendation results for the application mall, improve the application quality and popularity, and improve the user satisfaction and loyalty, and provide an important reference for optimizing the recommendation system and personalized service of the application mall.

Application Analysis of English Personalized Learning Based on Large-scale Open Network Courses

In the context of Big data, large-scale open online courses increase learning paths for learners, but in the face of countless high-quality curriculum resources, it is easy for derivative learners to face the dilemma of rich curriculum resources but difficult to choose resources, which leads to information maze for learners. How to help learners quickly and accurately find their own learning resources in the explosive growth of MOOC resources is an urgent problem in the field of education Big data. However, the traditional Collaborative filtering recommendation technology does not perform well when dealing with sparse data and cold start. The recommendation content is repeated and can not effectively deal with high-dimensional and nonlinear data of online learning users, resulting in low efficiency of resource recommendation. Therefore, the study adopts a deep belief network (DBN) to construct a personalized resource recommendation model. The model combines the learner behavior characteristics with the curriculum resource content attribute characteristics to form the learner feature vector. The parameters of the model are adjusted according to the characteristics of learners. Through experiments, the proposed model has shown good performance. The experiment explored the effects of training set size, learner characteristics, and GPU on model performance. The experimental results show that when the training set proportion is 100%, the RMSE, Accuracy, Recall, and F1 values of the model are 0.76, 0.946, 0.957, and 0.951, respectively. When the model is trained using a training set containing learner features, the RMSE, Accuracy, Recall, and F1 values of the model are 0.75, 0.962, 0.908, and 0.958, respectively. After using GPU to accelerate the model, the running time of the model decreased from 360 minutes to 90 minutes. The results indicate that the model cannot effectively mine data information when the degree of correlation between sample information is low. The richer the relationships between samples, the better the performance of the model. Simultaneously learning hunger feature vectors and learner behavior feature vectors for training can significantly improve the recommendation accuracy of the model. The main contribution of this study is to propose a recommendation method based on DBN classification to replace traditional similarity calculation methods, using DBN's efficient feature abstraction and feature extraction capabilities to fully explore learners' interest and preference for course resources. In addition, in view of the common problems of cold start and data sparsity in traditional Collaborative filtering recommendation methods, the research deeply mines the characteristics of learners' Demographics and curriculum resources' content attributes, and constructs a learner interest model based on DBN combined with learners' behavior characteristics, which effectively solves the problems of cold start and data sparsity, as well as the inaccurate expression of learners' interest preferences for curriculum resources.

INCORPORATING SUBJECTIVE FUNCTION IN A USER-BASED COLLABORATIVE FILTERING RECOMMENDER SYSTEM

The existing recommender systems faced a lot of challenges as a result of data sparsity problem, which decreases the recommendations accuracy to the target audience. Though there are numerous studies conducted to reduce the data sparsity problem. However, these studies fail to consider the subjectivity of human beings’ ratings based on lenient and stringent, which lead to unsuitable recommendations. Similarly, it is unable to compute the similarity between pair of users when the ratings among them did not overlap leading to inaccurate recommendations. This work is intended to enhance the adaptability of the system to sparse data by proposing a User Based Collaborative Filtering Recommender System incorporated with data refinement method that utilised a subjective function to identify lenient and stringent ratings in a rating matrix. The function identifies the user’s lenient and stringent ratings to take care of human subjectivity with regard to rating of the items. It also employed a subjective similarity function that used the subjective function for computing the similarity between users. The top similar neighbors obtained from the similarity computations are used to perform prediction for the target user. The experimental results show that compared with other representative systems the proposed system has better recommendations quality.

Practices and Challenges of Artificial Intelligence-Assisted Teaching in Vocational Undergraduate Public English Classrooms

Abstract This paper first introduces the architecture of the natural language processing system and then analyzes how natural language processing decomposes the probability distribution function by multi-factor form and Bayesian formula to improve the efficiency of function description. Then, NLM is applied to enhance the statistical coefficients of the N-gram model, which solves the problems of data sparsity and dimensionality disaster. By using natural language understanding in English teaching practice and experimenting with the teaching effect, it was found that the overall English achievement of the experimental group increased by 5.4 points, the average Z-score reached more than 3 points, and 96.6% of the students were interested in the teaching method. The new teaching method is demonstrated to have a significant impact on the improvement of English scores.

Exploring the Communication Path of Civic Education in Cyberspace Using Data Mining Technology

Abstract A recommendation algorithm is a specific application of data mining technology, which is an important means to expand the dissemination of Civics education in cyberspace. The elbow method is employed in this paper to determine the best number of clusters to cluster the Civics course evaluation dataset and finish preprocessing large-scale data. Aiming at the problem of different importance of rating bias between items, the traditional Slope One algorithm is optimized by comprehensively considering the relationship between the number of common ratings and the similarity between items and the prediction results. The optimized Slope One algorithm is used to perform one round of prediction and fill the scoring matrix to solve the data sparsity problem and improve prediction accuracy and stability. The optimized algorithm reduces by 8.56% compared to the user-based collaborative filtering algorithm, 5.83% compared to the item-based collaborative filtering algorithm, and 5.25% compared to the content-based recommendation algorithm, with a significant improvement in accuracy, recall, and F1 value. The four hypothesis validations indicate that the diversity, personalization, and accuracy of resource recommendations all have positive effects on the dissemination effect of civic education. The recommendation algorithm’s dissemination effect score in this paper is 3.781, which is significantly superior to the dissemination effect of Civic Education.

An autoencoder-based deep learning model for solving the sparsity issues of Multi-Criteria Recommender System

In recent times, recommender systems have acquired significant popularity as a solution to the issue of information overload. These systems offer personalised recommendations to users. The superiority of multi-criteria recommender systems over their single-criterion counterparts has been demonstrated, as the former are able to provide more precise recommendations by taking into account multiple dimensions of user preferences when rating items. The prevalent recommendation technique, matrix factorization of collaborative filtering, is hindered by the data sparsity problem of the user-item matrix. On the other hand, it is noteworthy that deep learning techniques have demonstrated significant potential in various research domains, including but not limited to image processing, pattern recognition, computer vision, and natural language processing. In recent times, there has been a surge in the utilisation of deep learning techniques in recommender systems, yielding promising outcomes. This study presents a novel approach to multi-criteria recommender systems through the utilisation of deep learning algorithms to mitigate the data sparsity issue. Specifically, deep autoencoders are utilised to uncover complex, non-linear, and latent relationships between users' multi-criteria preferences followed by matrix factorization technique, ultimately leading to more precise recommendations. The proposed model is evaluated by conducting the experiments on the multi-criteria dataset of Yahoo! Movies. According to the outcomes, the proposed approach outperforms the state of the art recommendation methods by generating more accurate and personalized recommendations. Also, it reduces the data sparsity up to 11% from the multi-criteria dataset.

An E-Commerce Personalized Recommendation Algorithm Based on Multiple Social Relationships

Environmental e-commerce is a sustainability-oriented e-commerce model. To address the problem of data sparsity and the lack of diversity in traditional e-commerce recommendation algorithms, a new collaborative filtering recommendation algorithm based on multiple social relationships is proposed in environmental e-commerce. In real social networks, there were many relationships between users. On the basis of the traditional matrix decomposition model, the proposed algorithm integrates multiple social relationships between users into the user feature matrix, and then the multiple social relationships between users and the user rating preference similarity were used to jointly predict the user’s rating value for commodity, thus the personalized recommendation for users was achieved. In order to verify the superiority of the proposed algorithm, in this paper, two open datasets were used to compare the performance of several recommendation algorithms. The experimental results show that compared with the traditional social recommendation algorithms, the proposed algorithm improves recommendation accuracy and diversity. In real environmental e-commerce recommendation systems, the proposed algorithm can provide users with more personalized recommendation results, and reduce the arbitrariness of customer purchases and frequent returns in reality.

Contrastive Multi-view Interest Learning for Cross-domain Sequential Recommendation

Cross-domain recommendation (CDR), which leverages information collected from other domains, has been empirically demonstrated to effectively alleviate data sparsity and cold-start problems encountered in traditional recommendation systems. However, current CDR methods, including those considering time information, do not jointly model the general and current interests within and across domains, which is pivotal for accurately predicting users’ future interactions. In this article, we propose a Contrastive learning-enhanced Multi-View interest learning model (CMVCDR) for cross-domain sequential recommendation. Specifically, we design a static view and a sequential view to model uses’ general interests and current interests, respectively. We divide a user’s general interest representation into a domain-invariant part and a domain-specific part. A cross-domain contrastive learning objective is introduced to impose constraints for optimizing these representations. In the sequential view, we first devise an attention mechanism guided by users’ domain-invariant interest representations to distill cross-domain knowledge pertaining to domain-invariant factors while reducing noise from irrelevant factors. We further design a domain-specific interest-guided temporal information aggregation mechanism to generate users’ current interest representations. Extensive experiments demonstrate the effectiveness of our proposed model compared with state-of-the-art methods.

Community Preserving Social Recommendation with Cyclic Transfer Learning

Transfer learning-based recommendation mitigates the sparsity of user-item interactions by introducing auxiliary domains. Social influence extracted from direct connections between users typically serves as an auxiliary domain to improve prediction performance. However, direct social connections also face severe data sparsity problems that limit model performance. In contrast, users’ dependency on communities is another valuable social information that has not yet received sufficient attention. Although studies have incorporated community information into recommendation by aggregating users’ preferences within the same community, they seldom capture the structural discrepancies among communities and the influence of structural discrepancies on users’ preferences. To address these challenges, we propose a community-preserving recommendation framework with cyclic transfer learning, incorporating heterogeneous community influence into the rating domain. We analyze the characteristics of the community domain and its inter-influence on the rating domain, and construct link constraints and preference constraints in the community domain. The shared vectors that bridge the rating domain and the community domain are allowed to be more consistent with the characteristics of both domains. Extensive experiments are conducted on four real-world datasets. The results manifest the excellent performance of our approach in capturing real users’ preferences compared with other state-of-the-art methods.

A kinetic model of iron trafficking in growing Saccharomyces cerevisiae cells; applying mathematical methods to minimize the problem of sparse data and generate viable autoregulatory mechanisms.

Iron is an essential transition metal for all eukaryotic cells, and its trafficking throughout the cell is highly regulated. However, the overall cellular mechanism of regulation is poorly understood despite knowing many of the molecular players involved. Here, an ordinary-differential-equations (ODE) based kinetic model of iron trafficking within a growing yeast cell was developed that included autoregulation. The 9-reaction 8-component in-silico cell model was solved under both steady-state and time-dependent dynamical conditions. The ODE for each component included a dilution term due to cell growth. Conserved rate relationships were obtained from the null space of the stoichiometric matrix, and the reduced-row-echelon-form was used to distinguish independent from dependent rates. Independent rates were determined from experimentally estimated component concentrations, cell growth rates, and the literature. Simple rate-law expressions were assumed, allowing rate-constants for each reaction to be estimated. Continuous Heaviside logistical functions were used to regulate rate-constants. These functions acted like valves, opening or closing depending on component "sensor" concentrations. Two cellular regulatory mechanisms were selected from 134,217,728 possibilities using a novel approach involving 6 mathematically-defined filters. Three cellular states were analyzed including healthy wild-type cells, iron-deficient wild-type cells, and a frataxin-deficient strain of cells characterizing the disease Friedreich's Ataxia. The model was stable toward limited perturbations, as determined by the eigenvalues of Jacobian matrices. Autoregulation allowed healthy cells to transition to the diseased state when triggered by a mutation in frataxin, and to the iron-deficient state when cells are placed in iron-deficient growth medium. The in-silico phenotypes observed during these transitions were similar to those observed experimentally. The model also predicted the observed effects of hypoxia on the diseased condition. A similar approach could be used to solve ODE-based kinetic models associated with other biochemical processes operating within growing cells.

Contrastive Graph Prompt-tuning for Cross-domain Recommendation

Recommender systems commonly suffer from the long-standing data sparsity problem where insufficient user-item interaction data limits the systems’ ability to make accurate recommendations. This problem can be alleviated using cross-domain recommendation techniques. In particular, in a cross-domain setting, knowledge sharing between domains permits improved effectiveness on the target domain. While recent cross-domain recommendation techniques used a pre-training configuration, we argue that such techniques lead to a low fine-tuning efficiency, especially when using large neural models. In recent language models, prompts have been used for parameter-efficient and time-efficient tuning of the models on the downstream tasks—these prompts represent a tunable latent vector that permits to freeze the rest of the language model’s parameters. To address the cross-domain recommendation task in an efficient manner, we propose a novel Personalised Graph Prompt-based Recommendation (PGPRec) framework, which leverages the efficiency benefits from prompt-tuning. In such a framework, we develop personalised and item-wise graph prompts based on relevant items to those items the user has interacted with. In particular, we apply Contrastive Learning to generate the pre-trained embeddings, to allow an increased generalisability in the pre-training stage, and to ensure an effective prompt-tuning stage. To evaluate the effectiveness of our PGPRec framework in a cross-domain setting, we conduct an extensive evaluation with the top- k recommendation task and perform a cold-start analysis. The obtained empirical results on four Amazon Review datasets show that our proposed PGPRec framework can reduce up to 74% of the tuned parameters with a competitive performance and achieves an 11.41% improved performance compared to the strongest baseline in a cold-start scenario.

Watchog: A Light-weight Contrastive Learning based Framework for Column Annotation

Relational Web tables provide valuable resources for numerous downstream applications, making table understanding, especially column annotation that identifies semantic types and relations of columns, a hot topic in the field of data management. Despite recent efforts to improve different tasks in table understanding by using the power of large pre-trained language models, existing methods heavily rely on large-scale and high-quality labeled instances, while they still suffer from the data sparsity problem due to the imbalanced data distribution among different classes. In this paper, we propose the Watchog framework, which employs contrastive learning techniques to learn robust representations for tables by leveraging a large-scale unlabeled table corpus with minimal overhead. Our approach enables the learned table representations to enhance fine tuning with much fewer additional labeled instances than in prior studies for downstream column annotation tasks. Besides, we further proposed optimization techniques for semi-supervised settings. Experimental results on popular benchmarking datasets illustrate the superiority of our proposed techniques in two column annotation tasks under different settings. In particular, our Watchog framework effectively alleviates the class imbalance issue caused by a long-tailed label distribution. In the semi-supervised setting, Watchog outperforms the best-known method by up to 26% and 41% in Micro and Macro F1 scores, respectively, on the task of semantic type detection.

Heterogeneous information fusion based graph collaborative filtering recommendation

Nowadays, with the application of 5G, graph-based recommendation algorithms have become a research hotspot. Graph neural networks encode the graph structure information in the node representation through an iterative neighbor aggregation method, which can effectively alleviate the problem of data sparsity. In addition, more and more information graph can be used in collaborative filtering recommendation, such as user social information graph, user or item attributed information graph, etc. In this paper, we propose a novel heterogeneous information fusion based graph collaborative filtering method, which models graph data from different heterogeneous graph, and combines them together to enhance presentation learning. Through information propagation and aggregation, our model can learn the latent embeddings effectively and enhance the performance of recommendation. Experimental results on different datasets validate the outperformance of the proposed framework.

A many objective commercial recommendation algorithm via Game-Based core node extraction

The development of the recommendation system (RS) has been focused on improving the accuracy of algorithms. As the number of new users grows, RS suffers from the cold start problem, data sparsity and long-tail problem. In this paper, we proposed a many objective commercial recommendation algorithm that extracts the core node through the pay-off function of game theory and makes it guide the evolutionary computation. Firstly, the core aim of commercial recommendations is to maximize total profit. Profits can vary wildly with the same accuracy. Thus, the profit value as one of the goals is considered in attention to accuracy, coverage, and novelty. Besides, the current user state value is updated according to user features and rating scores matrix with the attention mechanism at different time stamps. The nodes are divided into different communities, forming a dynamic community, and recommendations are made within each community. Meanwhile, to reduce the computational complexity and runtime, the pay-off function of game theory is used to extract the core user in each community. Each contribution value and user interaction as criteria to guide the current community of users. Experiments on real datasets show that selecting core nodes can improve the performance of business RS. Compared with other existing algorithms in many experiments, our algorithm performs better.