Search

Data integration, which aims to solve problems and create new services by combining datasets, has attracted considerable attention. The discovery of similar datasets that can be combined is critical. In the literature on similar dataset discovery, it is important to select an appropriate discovery method for each information need, such as the domain. However, conventional studies have evaluated discovery methods in different ways, such as domains, test datasets, and evaluation metrics. This factor prevents the appropriate method selection for each situation. Furthermore, the specific effects of the combination of different methods are not well known despite conventional studies arguing the importance of the combination. This study attempts to understand (1) the similarity indicators that should be employed for each domain and (2) the effects of a combination of different indicators on performance. We evaluated 16 inter-dataset clustering models based on different metadata-based similarity indicators, using unified evaluation metrics and datasets for 15 domains. Our results (1) suggest that similarity indicators should be used for each domain and (2) demonstrate that most of the combinations of different methods can improve clustering performance.

The proliferation of online social networking websites has resulted in the growth of online communities and increased communication among users. However, sharing personal information can pose a privacy threat to users, leaving them vulnerable to attacks like blackmail and fraud. Also, with the rise in scandals such as Geofeedia and Cambridge Analytica, as well as Facebook’s admission of sharing user data with top mobile phone manufacturers, worries about privacy on social networking platforms have become increasingly palpable. This paper discusses the privacy risks of sharing personal information on social networking websites. The proposed approaches offer a comprehensive protection of user privacy and prevents unauthorized access to user’s data and information by social network providers and unauthorized users. The study employs a preferential selective encryption approach that encompasses not only tweets but also other personally identifiable components, including images, memes, and metadata. This approach serves as a comprehensive solution to encrypt all elements within a social network, making it a one-stop destination for privacy protection in online social networks (OSN). When compared to other investigated schemes, the proposed approach significantly reduces privacy leakage to 0.48 units while achieving a high utility score of 0.8116 units for non-sensitive attributes in tweets. Moreover, it demonstrates remarkable performance in maintaining sensitive information coverage with notably minimal information loss: 0.4 units for images, 0.3 units for memes, and 12.7 units for metadata. Thus, in essence these findings, backed by quantifiable measures of privacy and utility, strongly validate the effectiveness of the proposed methods in achieving a superior balance between privacy preservation and utility when compared to its contemporaries, particularly in applications like online social networks (OSNs). Furthermore, the suggested method’s architecture makes it platform-neutral; it can be customized for any social networking website, such as Facebook, Instagram, and LinkedIn, and it may include any attribute that the user chooses to keep private.

Large scale enterprise networks often use Enterprise Key-Management (EKM) platforms for unified management of cryptographic keys. Monitoring access and usage patterns of EKM Systems (EKMS) may enable detection of anomalous (possibly malicious) activity in the enterprise network that is not detectable by other means. Analysis of enterprise system logs has been widely studied (for example at the operating system level). However, to the best of our knowledge, EKMS metadata has not been used for anomaly detection. In this article we present a framework for anomaly detection based on EKMS metadata. The framework involves automated outlier rejection, normal heuristics collection, automated anomaly detection, and system notification and integration with other security tools. This is developed through investigation of EKMS metadata, determining characteristics to extract for dataset generation, and looking for patterns from which behaviors can be inferred. For automated labeling and detection, a deep learning-based model is applied to the generated datasets: Long Short-Term Memory (LSTM) auto-encoder neural networks with specific parameters. This generates heuristics based on categories of behavior. As a proof of concept, we simulated an enterprise environment, collected the EKMS metadata, and deployed this framework. Our implementation used QuintessenceLabs EKMS. However, the framework is vendor neutral. The results demonstrate that our framework can accurately detect all anomalous enterprise network activities. This approach could be integrated with other enterprise information to enhance detection capabilities. Further, our proposal can be used as a general-purpose framework for anomaly detection and diagnosis.

Community Question Answering (CQA) forums such as Yahoo! Answers and Stack Overflow have become popular. The main goal of a CQA is to provide the most suitable answer in the shortest possible time. Since there is a reach archive of answered questions, similar question retrieval has received much attention intending to answer questions immediately after asking. One of the main challenges in this task is the lexical gap between questions, which refers to the discrepancies between the terminologies used by users asking questions. In this paper, we use metadata and two transformer-based techniques to improve the translation-based language model as a traditional technique addressing the lexical gap in retrieval systems. To overcome the lexical gap problem, additional context and information about the questions can help. Metadata is a rich source of information that refers to supplementary data associated with each question. Subject, category, and answer are metadata used in this article. To leverage these metadata, two transformer-based methods are employed. First, to utilize category information, we build category-specific dictionaries to obtain more accurate translation probabilities. A BERT model predicts the categories of the questions. Second, to utilize answer information, we propose a question expansion technique. Expansion is done by a transformer-based model using a retrieval-augmented generation (RAG) model to generate answers and expand new questions with corresponding answers. Finally, candidate questions are ranked according to their similarity to the expanded new question. Our proposed method achieves 51.47 in terms of MAP, outperforming all state-of-the-art approaches in question retrieval.

Large scale enterprise networks often use Enterprise Key-Management (EKM) platforms for unified management of cryptographic keys. In such a system, requests and responses commonly use the Key Management Interoperability Protocol (KMIP) format. The KMIP client and server use Transport Layer Security (TLS) to negotiate a mutually-authenti cated connection. Although KMIP traffic is encrypted, monitoring traffic and usage patterns of EKM Systems (EKMS) may enable detection of anomalous (possibly malicious) activity in the enterprise network that is notdetectable by other means. Metadata analysis of enterprise system traffic has been widely studied (for example at the TLS protocol level). However, KMIP metadata in EKMS has not been used for anomaly detection. In this paper, wepresent a framework for automated outlier rejection and anomaly detection. This involves investigati on of KMIP metadata, determining characteristics to extract for dataset generation, and looking for patt erns from which behaviors can be inferred. For automated labeling and detection, a deep learning-based model is applied to thegenerated datasets: Long Short-Term Memory (LSTM) auto-encoder neural networks with specific parameters. As aproof of concept, we simulated an enterprise environment, collected relevant KMIP metadata, and deployed this framework. Although our implementati on used Quintessence Labs EKMS, the framework we proposed is vendorneutral. The experimental results (Precision, Recall, F1 = 1.0) demonstrate that our framework can accurately detectall anomalous enterprise network activities. This approach could be integrated with other enterprise information toenhance detection capabilities. Our proposal can be used as a general-purpose framework for anomaly detecti on and diagnosis.

Crimes result in not only loss to individuals but also hinder national economic growth. While crime rates have been reported to decrease in developed countries, underdeveloped and developing nations still suffer from prevalent crimes, especially those undergoing rapid expansion of urbanization. The ability to monitor and assess trends of different types of crimes at both regional and national levels could assist local police and national-level policymakers in proactively devising means to prevent and address the root causes of criminal incidents. Furthermore, such a system could prove useful to individuals seeking to evaluate criminal activity for purposes of travel, investment, and relocation decisions. Recent literature has opted to utilize online news articles as a reliable and timely source for information on crime activity. However, most of the crime monitoring systems fueled by such news sources merely classified crimes into different types and visualized individual crimes on the map using extracted geolocations, lacking crucial information for stakeholders to make relevant, informed decisions. To better serve the unique needs of the target user groups, this paper proposes a novel comprehensive crime visualization system that mines relevant information from large-scale online news articles. The system features automatic crime-type classification and metadata extraction from news articles. The crime classification and metadata schemes are designed to serve the need for information from law enforcement and policymakers, as well as general users. Novel interactive spatiotemporal designs are integrated into the system with the ability to assess the severity and intensity of crimes in each region through the novel Criminometer index. The system is designed to be generalized for implementation in different countries with diverse prevalent crime types and languages composing the news articles, owing to the use of deep learning cross-lingual language models. The experiment results reveal that the proposed system yielded 86%, 51%, and 67% F1 in crime type classification, metadata extraction, and closed-form metadata extraction tasks, respectively. Additionally, the results of the system usability tests indicated a notable level of contentment among the target user groups. The findings not only offer insights into the possible applications of interactive spatiotemporal crime visualization tools for proactive policymaking and predictive policing but also serve as a foundation for future research that utilizes online news articles for intelligent monitoring of real-world phenomena.

With the rapid spread of Internet of Things (IoT) services, the number of sensor devices has exploded, and the complexity of managing sensor devices has become a problem. To solve this problem, a metadata-based approach that uses the unique environmental information associated with each device for its management is being developed. This paper focuses on metadata collection for device management and control, and proposes a new collection method that uses low-layer communication while not modifying the existing protocol. Our proposal utilizes the extended area of the probe request (PRQ) frame of IEEE 802.11, which is a layer-2 protocol, to collect metadata. This makes it possible to achieve stable operation even on inexpensive and resource-limited IoT devices, and to realize metadata collection with low communication overhead and power consumption. It is shown to reduce the load on the central processing unit (CPU) and reduce power consumption compared to Internet Protocol (IP) -based metadata collection (layer-3 protocol and above). In addition, in terms of time sensitivity, the collection delay at the time of rising from deep sleep is reduced by 89.8% compared to IP-based techniques. Furthermore, as a benefit of low-layer collection, it enables periodic metadata collection in the background regardless of network connection above the IP layer. To demonstrate this benefit, an example of applying metadata collection to device management is prototyped and its feasibility is experimentally confirmed.

Skin cancer is a dangerous form of cancer that develops slowly in skin cells. Delays in diagnosing and treating these malignant skin conditions may have serious repercussions. Likewise, early skin cancer detection has been shown to improve treatment outcomes. This paper proposes DeepMetaForge, a deep-learning framework for skin cancer detection from metadata-accompanied images. The proposed framework utilizes BEiT, a vision transformer pre-trained as a masked image modeling task, as the image-encoding backbone. We further propose merging the encoded metadata with the derived visual characteristics while compressing the aggregate information simultaneously, simulating how photos with metadata are interpreted. The experiment results on four public datasets of dermoscopic and smartphone skin lesion images reveal that the best configuration of our proposed framework yields 87.1% macro-average F1 on average. The empirical scalability analysis further shows that the proposed framework can be implemented in a variety of machine-learning paradigms, including applications on low-resource devices and as services. The findings shed light on not only the possibility of implementing telemedicine solutions for skin cancer on a nationwide scale that could benefit those in need of quality healthcare but also open doors to many intelligent applications in medicine where images and metadata are collected together, such as disease detection from CT-scan images and patients’ expression recognition from facial images.

The aim of this study is to introduce a virtual laboratory platform with metadata-based API for the creation of dynamic software components. Also, the study seeks to assess the platform’s validity and technological acceptance among the community of system developers. The platform is referred to as Esfinge Virtual Lab, and it has been developed utilizing the Java programming language. The platform has a general purpose and is demonstrated in this work in the context of system prototyping. Using declarative programming techniques and a metadata-based APIs, a software developer may focus on business principles and processes rather than expend energy on the retrieval of information or the programming of the user interface. The Esfinge Virtual Lab platform enables developers to efficiently produce diverse forms of data visualization with minimal coding effort. This includes the rapid generation of tables, graphs, maps, and personalized outputs. The components are packaged as Java Archive (JAR) files and uploaded onto the platform. The components that have been loaded can be accessed for execution and can be either injected or dynamically invoked for a new component, thereby facilitating code reuse. This study provides a comprehensive explanation of the platform, including its internal architecture and features. The deployment of the Esfinge Virtual Lab was demonstrated through two practical applications in Brazilian institutions, namely INPE and CEMADEN. A survey was performed to determine the validity of the platform, consisting of open-ended and closed-ended questions. The research used the Technology Acceptance Model (TAM) as a quantitative analytical instrument and thematic analysis as a qualitative analytical approach.

Machine learning (ML) practitioners and organizations are building model repositories of pre-trained models, referred to as model zoos. These model zoos contain metadata describing the properties of the ML models and datasets. The metadata serves crucial roles for reporting, auditing, ensuring reproducibility, and enhancing interpretability. Despite the growing adoption of descriptive formats like datasheets and model cards, the metadata available in existing model zoos remains notably limited. Moreover, existing formats have limited expressiveness, thus constraining the potential use of model repositories, extending their purpose beyond mere storage for pre-trained models. This paper proposes a unified metadata representation format for model zoos. We illustrate that comprehensive metadata enables a diverse range of applications, encompassing model search, reuse, comparison, and composition of ML models. We also detail the design and highlight the implementation of an advanced model zoo system built on top of our proposed metadata representation.

WebAssembly (Wasm), a technology enabling efficient native code execution in web browsers, has seen a significant rise in adoption as a popular compilation target. This has led to the emergence of lightweight web services powered by Wasm, characterized by their small binary size and reduced data transfer overhead, thanks to the inherent efficiency of Wasm. Despite their lightweight nature, these services can deliver powerful features like image/video processing, AI and graphical application that surpass the capabilities of JavaScript. To ensure lightweight web services and enhance the overall web experience, Wasm has been extensively optimized. However, these optimizations have raised concerns about memory safety, leading to memory-related vulnerabilities. Wasm’s characteristic memory structure, linear memory, has vulnerabilities that provide various attack vectors to attackers. In particular, it presents various attack possibilities through metadata modification containing memory structure information. Attackers can exploit heap memory overflow in Wasm applications, allowing them to target arbitrary memory addresses, modify data, or execute arbitrary code. Such overflows can corrupt memory metadata, resulting in incorrect memory behavior. While research has mitigate memory-related weaknesses in languages such as C and C++ and architectures like X86 in recent decades, the direct application of security solutions designed for different domains to Wasm is not a practical approach. Consequently, allocators in Wasm remain vulnerable to issues like heap overflow and metadata corruption. Thus, there is a pressing need for tailored memory safety techniques and solutions that accommodate Wasm’s architecture-agnostic and linear memory structures. In this paper, we propose metaSafer as a solution. By shadowing metadata from Wasm linear memory to JavaScript virtual machine memory and conducting metadata verification, metaSafer effectively blocks attack attempts and vectors. Notably, our solution achieves fast memory shadowing and validation while maintaining a small code size. Through various verification processes, we measured the performance and code size of metaSafer, revealing that it is a software-only security solution with no additional hardware requirements. metaSafer demonstrates robust metadata protection for Wasm applications with an acceptable performance overhead of up to 8% in SQLite speed tests and Polybench benchmarks.

Finding relevant research papers can be a challenging task due to the enormous number of scientific publications released each year. Recently, the scientific community has been diving deep into citation analysis, specifically examining the content of papers to identify more crucial documents. Citations serve as potential parameters for establishing connections between research articles. They have been widely utilized for various academic purposes, including calculating journal impact factors, determining researchers’ h-index, allocating research grants, and pinpointing the latest research trends. However, researchers have argued that not all citations carry equal weight in terms of influence. Consequently, alternative techniques have been proposed to identify significant citations based on content, metadata, and bibliographic information. Nonetheless, the current state-of-the-art approaches still require further refinement. Additionally, the application of deep learning models and word embedding techniques in this context has not been extensively studied.In this research work, we present an approach consisting of two primary modules: 1) Section-wise citation count, and 2) metadata-based analysis of citation intent. Our study involves conducting several experiments using deep learning models in conjunction with FastText, word2vec, and BERT-based word embeddings to perform citation analysis. These experiments were carried out using two benchmark datasets, and the results were compared with a contemporary study that employed a rich set of content-based features for classification. Our findings reveal that the deep learning CNN model, coupled with FastText word embeddings, achieves the best results in terms of accuracy, precision, and recall. It outperforms the existing state-of-the-art model, achieving a precision score of 0.97

Learning to detect objects, such as humans, in imagery captured by an unmanned aerial vehicle (UAV) usually suffers from tremendous variations caused by the UAV’s position towards the objects. In addition, existing UAV-based benchmark datasets do not provide adequate dataset metadata, which is essential for precise model diagnosis and learning features invariant to those variations. In this paper, we introduce Archangel, the first UAV-based object detection dataset composed of real and synthetic subsets captured with similar imagining conditions and UAV position and object pose metadata. A series of experiments are carefully designed with a state-of-the-art object detector to demonstrate the benefits of leveraging the metadata during model evaluation. Moreover, several crucial insights involving both real and synthetic data during model optimization are presented. In the end, we discuss the advantages, limitations, and future directions regarding Archangel to highlight its distinct value for the broader machine learning community.

Forecasting research disciplines associated with research projects is a significant challenge in research information systems. It can reduce the administrative effort involved in entering research project-related metadata, eliminate human errors, and enhance the quality of research project metadata. It also enables the calculation of the degree of interdisciplinarity of these projects. However, predicting scientific research disciplines and measuring interdisciplinarity in a research endeavor remain difficult. In this paper, we propose a framework for predicting the research disciplines associated with a research project and measuring the degree of interdisciplinarity based on associated metadata to address these issues. The proposed framework consists of several components to improve the performance of research disciplines prediction and interdisciplinarity measurement systems. These include a feature extraction component that utilizes a topic model to extract the most appropriate features. Further, the framework proposes a discipline encoding component that applies a data mapping strategy to lower the dimensionality of the output variables. Furthermore, a distance matrix creation component is proposed to recommend the most appropriate research disciplines and compute interdisciplinarity associated with research projects. We implemented the suggested framework on two separate research information systems databases for research projects, Dimensions and the Flemish Research Information Space. Experimental results demonstrate that the proposed framework predicts the research disciplines associated with research projects more accurately than related work.

With the accelerating development of science and technology, the academic papers being published in various fields are increasing rapidly. Academic papers specially in science and technology fields are a crucial media for researchers who develop new technologies by identifying knowledge regarding the latest technological trends and conduct derivative studies in science and technology. Therefore, the continual collection of extensive academic papers, structuring of metadata, and construction of databases are significant tasks. However, research on automatic metadata extraction from Korean papers is not being actively conducted currently owing to insufficient Korean training data. We automatically constructed the largest labeled corpus in South Korea to date from 315,320 PDF papers belonging to 503 Korean academic journals and this labeled corpus can be used for training the models of automatic extraction for 12 metadata types from PDF papers. This labeled corpus is available at https://doi.org/10.23057/48. Moreover, we developed inspection process and guidelines for the automatically constructed data and performed a full inspection of the validation and testing data. The reliability of the inspected data was verified through the inter-annotator agreement measurement. Using our corpus, we trained and evaluated the BERT based transfer learning model to verify its reliability. Furthermore, we proposed new training methods that can improve the metadata extraction performance of Korean papers, and through these methods, we developed KorSciBERT-ME-J and KorSciBERT-ME-J+C models. The KorSciBERT-ME-J showed the highest performance with an F1 score of 99.36%, as well as robust performance in automatic metadata extraction from Korean academic papers in various formats.

Data ingestion tools are critical component of Data Lake. Existing data ingestion tools face challenges of handling large variety, formats, sources of data. There exists void for unified data ingestion interface to handle the above research problems. This study proposes an innovative and integrated framework for data ingestion in a data lake, addressing the challenges posed by heterogeneous data sources, formats, and metadata management. The framework comprises three novel modules: First Unified Data Integration Connectors (UDIC), which provide seamless connectivity and data retrieval capabilities from diverse sources including databases, data warehouses, file systems, cloud storage, and APIs; Second, Adaptive Data Variety Transformation (ADVT), a module that intelligently handles the transformation and processing of structured, semi-structured, and unstructured data types, ensuring efficient ingestion into the data lake; and third, Intelligent Metadata Management (IMM), a module that captures, stores, and manages metadata associated with the ingested data, offering advanced search, discovery, and enrichment functionalities. Comparative study corroborates features offered by the service with existing data ingestion tools to evaluate the novelty and significance of the study. Performance validation shows varying ingestion latencies across different data types: approximately 148.1 microseconds per record for structured data, 234.2 microseconds per record for semi-structured data, 65.6 microseconds per kilobyte (KB) for video data, and 42.7 microseconds per KB for image data. These results underscore the importance of considering data structure and size in optimizing ingestion processes. Overall, this research aims to revolutionize data ingestion in data lake environments by providing a unified solution for handling diverse data sources, formats, and metadata management.

Automated camera-based sensors are widely used in vessel-based research to monitor plankton and marine particles. However, current methods suffer from the costly and time-consuming requirement of annotating data for fully supervised learning, especially in plankton grouping tasks characterized by long-tailed datasets. In response, we propose a novel self-supervised learning framework that significantly reduces reliance on expensive human annotations by leveraging crucial metadata such as water depth and location. The method comprises three major steps: self-supervised training, innovative sampling, and final classification. It identifies key sample subsets from an unlabeled dataset using a hierarchical clustering approach and incorporates an innovative balancing representative subsampling strategy that addresses the challenge of dataset imbalance and enhances generalizability across diverse plankton classes. Our approach prioritizes discerning representation features observed in images that exhibit correlations with the patterns found in their associated metadata. Furthermore, our method introduces a novel grouping based on the visual perspective selection method, enabling the identification of balanced subset views that depart from traditional class-based categorization. Our experimental results showcase a significant enhancement in image classification accuracy, with a 23% improvement over methods that do not utilize metadata, and attains a macro F1-score of 54% for ten populated species from a severely long-tailed dataset. This is achieved with a mere 0.3% of the entire dataset used for annotation.

The increasing number of people are living with disability in the World and their access to formal education is considered a challenge for the development of the online education and educational resources. This problem is considered one of the 17 sustainable development goals that are focused on inclusive and equitable quality education. Nevertheless, the existing proposals for mainstream accessibility in virtual education are still complex to apply. However, the models, standards, and good practices to contribute to the virtual educational process and the design of learning for all are identified. For these reasons, in this paper, we describe an accessibility evaluation proposal based on 4 interaction domains: user analysis and interaction, intelligent systems, knowledge databases, and evaluation. In the same way, we describe a set of tools that constitute a Repository of Accessible Learning Objects (RALO) from the perspective of accessibility and adaptability metadata. In this line, the knowledge database follows the regulation and educational models focused on the students with disabilities needs and preferences from the conception of universal design. The validation of the proposal is based on the interaction study and analysis of regular and disabled students and teachers who developed the Learning Objects (LO). To determine whether there was consensus among the teacher’s scores, we used Kendall’s Coefficient of Concordance W.

The commercial success of online retailing platforms relies on sophisticated technological advancements of recommender systems. As online marketplaces gain popularity it becomes crucial to provide accurate product suggestions to make personalized shopping experience more exciting for the online customer. The evolution of recommendation systems from using collaborative and content-based approaches to more advanced deep learning-based approaches provide motivation as well the necessity to leverage more advanced tools for creating recommendation systems more powerful. Among variety of e-commerce products, this work concentrates on the apparels section, specifically coming under men’s shirt category. In this work, we introduce a Hybrid Recommender System that combines content based and item-based collaborative filtering strategies. To implement the content-based filtering approach we employ text-representation techniques. These methods carefully analyze how users interact with product specifications to understand their nuanced preferences. We also propose a cascading transformer mechanism for generating contextually aware representations using sentence transformers derived from Large Language Models (LLMs) T5 and BERT. To gauge the emotional tone and sentiment associated with product reviews, we make use of rule-based and embedding based sentiment analysis models. A weighted score mechanism has been used to determine the final recommendations by the proposed system. Finally, we discuss the efficiency of the proposed filtering technique and compare various text representation methods along with popular baseline transformer models. The introduced hybrid recommendation system with the proposed cascaded model was evaluated and found to give recommendations with 94.76% similarity score while FastText achieved 66.90%. Doc2Vec and Word2Vec resulted in recommendations with 46.62% and 47.70% similarity respectively.

In the telecommunications field, the problem of data island is caused by the separation and isolation of data. They are distributed in different systems including the business support system (BSS), management support system (MSS), and operation support system (OSS). The common idea is to use global ID mapping to break data barriers. However, using the direct global ID mapping of raw data has the problems of large data scale and the inability to guarantee privacy and security. With this in mind, constructing and completing a metadata knowledge graph to fabric the data is a feasible approach. Considering the particularity of the telecom metadata knowledge graph and the need for hierarchical distinction in business and semantic abstraction, we propose a deep learning method and framework based on the spherical coordinate system. It can be extended to a poly-spherical coordinate system and add a pre-training process composed of word2vec and a clusterer. Experimental results show that our method achieves state-of-the-art on our dataset, MDCT and has excellent performance on two public datasets, FB15k-237 and WN18RR.

File system forensics is one of the most important elements in digital forensic investigations. To date, various file system forensic methods, such as analysis of tree structure and the recovery of deleted file data, have been studied. Among these file system forensic methods, the recovery of file system metadata is a key technique that makes digital forensic investigations possible by recovering metadata when it is not possible to obtain metadata in a regular manner because the file system structure is damaged due to an accident/disaster or cyber terrorism. Previous studies mainly focused on recovering record or entry data, which are the basic units of metadata, using carving techniques via a fixed values or values capable of range prediction at the beginning of the data. However, no studies have been conducted on metadata without such fixed values or values capable of range prediction. $\$ $ LogFile, which is a metadata file of the New Technology File System (NTFS) that is one of the most used file systems at present, contains very important metadata that provide a history of all file system operations during a specific period. However, since there is no fixed value or a value capable of range prediction at the start position of the record, which is the basic unit of $\$ $ LogFile, there have been no studies on recovery using record units, and only recovery by file and page have been possible. If the file header or page header of $\$ $ LogFile is damaged, existing recovery methods cannot properly recover the metadata; in such cases, a record-level recovery method is required to recover the metadata. In this context, we investigated the mechanisms of record storage through a detailed analysis of the $\$ $ LogFile structure and proposed a recovery method for records without fixed values. Our proposed method was implemented as a tool and verified through comparative experiments with existing forensic tools that recover $\$ $ LogFile data. The experimental results showed that the proposed recovery method was able recover all the data that existing tools are unable to recover in situations where the $\$ $ LogFile data were damaged. The implemented tools are released free of charge to contribute digital forensic community. Finally, we explained what important role $\$ $ LogFile played in solving real-world cases and confirm the importance of recovering $\$ $ LogFile data in situations where file systems may be damaged due to accidents and disasters.

The need for adaptive e-learning environments that respond to learning variability is now a fundamental requirement in education, as it helps to ensure that students learn and pass their courses within a set time frame. Although guidelines, techniques and methods have been established in recent years to contribute to the development of accessible and adaptable e-learning environments that promote digital inclusion, their implementation is challenging due to the lack of knowledge of an adequate way to do it and because it is considered more of a technological competence for scholars in the area. In this context, automated support for adapting material that responds to the correct use of accessibility metadata not only provides a way to improve the description of adapted educational resources, but also facilitates their search according to the needs and preferences of students, particularly those with disabilities. In this article, we carry out a multilevel methodological proposal for the automatic adaptation of open educational resources, in order to provide a tool that contributes to the accessibility and correct use of their metadata in e-learning environments. A research is conducted with students with disabilities to establish their real needs and preferences, highlighting the need to strengthen the adequate description and coherent alternative text in images, the correct subtitling in videos and the conversion of audio to text, data that are relevant to our proposal. The research conducted aims to contribute with an automated support tool in the generation of accessible educational resources that are correctly labeled for search and reuse. This research also aims to support researchers in artificial intelligence applications to address challenges and opportunities in the field of virtual education, in addition to providing an overview that could help those who generate educational resources and maintain their interest in making them accessible.

Existing nonvolatile memory (NVM)-based file systems can fully leverage the characteristics of NVM to obtain better performance than traditional disk-based file systems. It has the potential capacity to efficiently manage metadata and perform fast metadata operations. However, most NVM-based file systems mainly focus on managing file metadata (inode), while pay little attention to directory metadata (dentry), which also has a noticeable impact on the file system performance. Besides, the traditional journaling technique that guarantees metadata consistency may not yield satisfactory performance on NVM-based file systems. To solve these problems, in this article we propose a fast and low overhead metadata operation mechanism, called FLOMO. It first adopts a novel slotted-paging structure in NVM to reorganize dentry for efficiently performing dentry operations, and utilizes the red–black tree in DRAM to accelerate dentry lookup and the search process of dentry deletion. Moreover, FLOMO presents a selective journaling scheme for metadata updates, which partially logs the changes related to dentry in the proposed slotted page, thereby, mitigating the redundant journaling overhead. To verify FLOMO, we implement it in a typical NVM-based file system, the persistent memory file system (PMFS). Experimental results show that FLOMO accelerates the metadata operations in PMFS by 34.4% $\sim 59$ %, and notably reduces the journaling overhead for metadata, shortening the latency by 59% on average. For real-world applications, FLOMO has higher throughput compared with PMFS, PMFS without journal, and NOVA, achieving up to $2.1\times $ , $1.1\times $ , and $1.3\times $ performance improvement, respectively. [ABSTRACT FROM AUTHOR]

During the last decade, numerous governmental, educational or cultural institutions have launched Open Data initiatives that have facilitated the access to large volumes of datasets on the web. The main way to disseminate this availability of data has been the deployment of Open Data catalogs exposing metadata of these datasets, which are easily indexed by web search engines. Open Source platforms have facilitated enormously the labor of institutions involved in Open Data initiatives, making the setup of Open Data portals almost a trivial task. However, few approaches have analyzed how precisely metadata describes the associated datasets. Taking into account the existing approaches for analyzing the quality of metadata in the Open Data context and other related domains, this work contributes to the state of the art by extending an ISO 19157 based method for checking the quality of geographic metadata to the context of Open Data metadata. Focusing on metadata models compliant with the Data Catalog Vocabulary proposed by W3C, the proposed extended method has been applied for the evaluation of the Open Data catalog of the Spanish Government. The results have been also compared with those obtained by the Metadata Quality Assessment methodology proposed at the European Data Portal.

Recent advancements in the domain of recommender systems have stemmed from the inspiration of representing the user-item interaction into graphs. These heterogeneous graphs comprehensively capture the non-linear relationships between users and items alongwith features and emneddings. Graph Convolution Networks (GCNs) are state-of-the-art graph-based learning models that learn and represent the graph structures by recursively stacking layers of convolution and non-linear activation operations. GCNs are augmented with the strength of deep learning paradigms resulting in achieving better performance as compared to traditional CF (Collaborative Filtering) methods. Despite modern improvements in the domain of recommender systems, cold-start users are a daunting challenge in the design of recommender systems since the conventional recommendation services are based on solely one data source. During the recent years, cross-domain recommendation methods have gained popularity because of the availability of information in multiple domains for cold start users. We supplement this information by utilizing the data contained in the metadata of users alongwith the strength of modelling graphs using GCNs. Our proposed algorithm seams the strength of GCNs with cross-domain paradigm utilizing the richness in metadata in user’s feedback to overcome the sparsity in user-item rating matrix. The combined advantages of GCNs and cross-domain approaches alleviated the issues of cold-start users by transferring user preferences from a source domain to a target domain.

Temporal prefetchers are powerful because they can prefetch irregular sequences of memory accesses, but temporal prefetchers are commercially infeasible because they store large amounts of metadata in DRAM. This article presents Triage, the first temporal data prefetcher that does not require off-chip metadata. Triage builds on two insights: (1) Metadata are not equally useful, so the less useful metadata need not be saved, and (2) for irregular workloads, it is more profitable to use portions of the LLC to store metadata than data. We also introduce novel schemes to identify useful metadata, to compress metadata, and to determine the fraction of the LLC to dedicate for metadata. Using an industrial-strength simulator running irregular workloads on a single-core system, we show that at a prefetch degree of 4, Triage improves performance by 41.1 percent compared to a baseline with no prefetching, whereas BO, a state-of-the-art prefetcher that uses only on-chip metadata, sees only 10.9 percent improvement. Compared with MISB, a temporal prefetcher that uses off-chip metadata, Triage provides a design alternative that reduces memory traffic by an order of magnitude (260.8 percent extra traffic for MISB at degree 1 versus 56.9 percent for Triage), while reducing coverage by 20 percent. [ABSTRACT FROM AUTHOR]

Understanding system performance metrics ensures better utilization of the radiology resources with more targeted interventions. The images produced by radiology scanners typically follow the DICOM (Digital Imaging and Communications in Medicine) standard format. The DICOM images consist of textual metadata that can be used to calculate key timing parameters, such as the exact study durations and scanner utilization. However, hospital networks lack the resources and capabilities to extract the metadata from the images quickly and automatically compute the scanner utilization properties. Thus, they resort to using data records from the Radiology Information Systems (RIS). However, data acquired from RIS are prone to human errors, rendering many derived key performance metrics inadequate and inaccurate. Hence, there is motivation to establish a real-time image transfer from the Picture Archiving and Communication Systems (PACS) to receive the DICOM images from the scanners to research clusters to conduct such metadata processing to evaluate scanner utilization metrics efficiently and quickly. This paper analyzes the scanners' utilization by developing a real-time monitoring framework that retrieves radiology images into a research cluster using the DICOM networking protocol and then extracts and processes the metadata from the images. Our proposed approach facilitates a better understanding of scanner utilization across a vast healthcare network by observing properties such as study duration, the interval between the encounters, and the series count of studies. Benchmarks against using the RIS data indicate that our proposed framework based on real-time PACS data estimates the scanner utilization more accurately. Furthermore, our framework has been running stable and performing its computation for more than two years on our extensive healthcare network in pseudo real-time.

Rapid increase in scholarly publications on the web has posed a new challenge to the researchers in finding highly relevant and important research articles associated with a particular area of interest. Even a highly relevant paper is sometimes missed especially for novice researchers due to lack of knowledge and experience in finding and accessing the most suitable articles. Scholarly recommender system is a very appropriate tool for this purpose that can enable researchers to locate relevant publications easily and quickly. However, the main downside of the existing approaches is that their effectiveness is dependent on priori user profiles and thus, they cannot recommend papers to the new users. Furthermore, the system uses both public and non-public metadata and therefore, the system is unable to find similarities between papers efficiently due to copyright restrictions. Considering the above challenges, in this research work, a novel hybrid approach is proposed that separately combines a Content Based Filtering (CBF) recommender module and a Collaborative Filtering (CF) recommender module. Unlike previous CBF and CF approaches, public contextual metadata and paper-citation relationship information are effectively incorporated into these two approaches separately to enhance the recommendation accuracy. In order to verify the effectiveness of the proposed approach, publicly available datasets were employed. Experimental results demonstrate that the proposed approach outperforms the baseline approaches in terms of standard metrics (precision, recall, F1-measure, mean average precision, and mean reciprocal rank), indicating that the proposed approach is more efficient in recommending scholarly publications.

Scientific applications often require high-bandwidth shared storage to perform joint simulations and collaborative data analytics. Shared memory pools provide a chance to satisfy such needs. Recently, a high-speed network such as Gen-Z utilizing persistent memory (PM) offers an opportunity to create a shared memory pool connected to compute nodes. However, there are several challenges to use scientific applications on the shared memory pool directly such as scalability, failure-atomicity, and lack of scientific metadata-based search and query. In this paper, we propose MOSIQS, a persistent memory object storage framework with metadata indexing and querying for scientific computing. We design MOSIQS based on the key idea that memory objects on PM pool can live beyond the application lifetime and can become the sharing currency for applications and scientists. MOSIQS provides an aggregate memory pool atop an array of persistent memory devices to store and access memory objects to accelerate scientific computing. MOSIQS uses a lightweight persistent memory key-value store to manage the metadata of memory objects, which enables memory object sharing. To facilitate metadata search and query over millions of memory objects resident on memory pool, we introduce Group Split and Merge (GSM), a novel persistent index data structure designed primarily for scientific datasets. GSM splits and merges dynamically to minimize the query search space and maintains low query processing time while overcoming the index storage overhead. MOSIQS is implemented on top of PMDK. We evaluate the proposed approach on many-core server with an array of real PM devices. Experimental results show that MOSIQS gains a 100% write performance improvement and executes multi-attribute queries efficiently with $2.7\times $ less index storage overhead offering significant potential to speed up scientific computing applications.

The existing plethora of document classification techniques exploits different data sources either from the content or metadata of research articles. Various journal publishers like Springer, Elsevier, IEEE, etc., do not provide open access to the content of research articles, whereas metadata is freely available there. Metadata like title, keyword, and abstract can serve as a better alternative to the content in various scenarios. In the current literature, researchers have assessed the role of some of the metadata individually. We believe that the collective contribution of metadata parameters can play a significant role in classifying research papers. This paper presents a comprehensive evaluation of the role of metadata, individually as well as in combinations to achieve the objective of research paper classification. Moreover, we have classified the research articles into ACM hierarchy root categories (e.g. general literature, hardware, software, etc.). In this comprehensive evaluation, we have assessed all the possible combinations of metadata features against different classifiers such as Random Forest, K Nearest Neighbor, and Decision Tree. The results of this research reveal that the title & keywords combination outperforms other combinations with an F-measure score of 0.88.

Recent trends in the Web of Things (WoT) have led to data explosion. Data lake (DL), as a flexible on-demand heterogeneous data management architecture, has become a feasible solution in data management. Metadata modeling for DLs is the key basis for smart analysis and processing. However, the varieties in structures and semantics of industrial WoT data hinder metadata modeling and maintenance. Moreover, the lack of textual descriptions and the semantics hidden in value streams make it hard to automatically construct semantic metadata. The dynamic nature of WoT requires on-time evolution on metadata. To overcome these challenges, we propose an automated bottom-up metadata generation approach for DL of WoT applications. Applying a data-driven framework, raw data are notated as linked data and self-organizing map-based online clustering is applied to real timely extract data characteristics. To recognize entities, concepts and relations, semantics-based entity discovery approach from short texts is proposed according to the feature of WoT data. The numerical analysis is performed to find the hidden relations from raw values. Full-dimensional metadata with rich semantic knowledge are finally built. Experiments on a real-world dataset are conducted to verify the effectiveness of methods and a case study on an energy WoT system is provided to demonstrate the feasibility of the approach. [ABSTRACT FROM AUTHOR]

Current distributed file systems are designed to support PB-scale even EB-scale data storage. Metadata service, which manages file attribute information and the global namespace tree, is crucial to system performance. Distributed metadata management, using multiple metadata servers (MDS's) to store metadata, provides effective approaches to alleviate the workload of a single server. However, maintaining good metadata locality and keeping load balancing among MDS's at the same time is a nontrivial problem. To better take advantage of the current distribution of the metadata, in this article, we present the first machine learning based model called DeepHash, which leverages the neural network to learn a locality preserving hashing (LPH) mapping scheme. DeepHash first converts the metadata nodes to feature vectors by the network embedding technology. Due to the absence of training labels, i.e., the hash values of metadata nodes, we design a pair loss function with distinctive characters to train DeepHash, and introduce the sampling strategy to improve the training efficiency. Besides, we propose an efficient algorithm to dynamically balance the workload and adopt the cache model to improve query efficiency. The experiments on the Amazon EC2 platform demonstrate that the DeepHash can preserve the metadata locality meanwhile maintaining a high load balancing, which denotes the effectiveness and efficiency of DeepHash compared with traditional and state-of-the-art schemes. [ABSTRACT FROM AUTHOR]

With the emergence of global platforms for trading and buying/selling data, data have become a profitable commodity. The growth of such platforms has necessitated the further expansion of the scope of data in digital economies. To this end, understanding the nature of available data and their relationships between them has become an important challenge for expanding their use. Thus, in this study, we assumed data on the platforms as a population and metadata as the samples. Thus, we quantitatively investigated the structural characteristics of data platforms, while avoiding the risk of lost business opportunities and privacy issues by not sharing the data themselves. By observing the characteristics of data and variables, we found that the data network had a structure that was locally dense and globally sparse, which is quite similar to networks of human relationships. Moreover, we found that data play different roles on the platforms when divided into sharing conditions, namely, shareable data and sensitive data. Finally, we discussed the potential tactics for individuals who create/use data platforms based on our findings. The contributions of this study include a new framework for data platform observation and a method that uses metadata and a network approach to analyze structural characteristics of data.

The unprecedented growth of the research publications in diversified domains has overwhelmed the research community. It requires a cumbersome process to extract this enormous information by manually analyzing these research documents. To automatically extract content of a document in a structured way, metadata and content must be annotated. Scientific community has been focusing on automatic extraction of content by forming different heuristics and applying different machine learning techniques. One of the renowned conference organizers, ESWC organizes state-of-the-art challenge to extract metadata like authors, affiliations, countries in affiliations, supplementary material, sections, table, figures, funding agencies, and EU funded projects from PDF files of research articles. We have proposed a feature centric technique that can be used to extract logical layout structure of articles from publishers with diversified composition styles. To extract unique metadata from a research article placed in logical layout structure, we have developed a four-staged novel approach “FLAG-PDFe”. The approach is built upon distinct and generic features based on the textual and the geometric information from the raw content of research documents. At the first stage, the distinct features are used to identify different physical layout components of an individual article. Since research journals follow their unique publishing styles and layout formats, therefore, we develop generic features to handle these diversified publishing patterns. We employ support vector classification (SVC) in the third stage to extract the logical layout structure (LLS)/ sections of an article, after performing comprehensive evaluation of generic features and machine learning models. Finally, we further apply heuristics on LLS to extract the desired metadata of an article. The outcomes of the study are obtained using the gold standard data set. The results yields 0.877 recall, precision 0.928 and 0.897 F-measure. Our approach has achieved a 16% gain on f-measure when compared to the best approach of the ESWC challenge.

People's usage of social networks, mobile applications, websites, sensor networks and other computer systems leads to a massive production of personal data about their behaviors and preferences. Personal data are used by organizations in business and marketing tasks. However, details about personal data usage are often not accessible or clear to data subject, raising concerns about privacy and security. Presentation of information about personal data usage needs improvement towards Personal Data Transparency. Thus, this paper aims to present the TR-Model, a Metadata Application Profile guideline that intends to propose a standardization on information to be considered minimally necessary to Personal Data Transparency as well as a set of specifications to guide developers on how to present this data. TR-Model elements are focused providing Personal Data Transparency in a user-friendly and high quality format. TR-Model presents a set of specification based on entities, metadata, metaevents and descriptions. The model evaluation was based on user testing in several scenarios of usage of personal data in a gym application tool. The information presented was created based on the TR-Model metadata, metaevents and descriptions. Participants evaluated transparency considering dimensions of Human-Computer Interaction and Information Quality. Participants' opinions were recorded in surveys and analyzed with descriptive statistics; the results indicate that the TR-Model was effective in supporting the production of friendly, understandable and relevant Transparency for data subjects, in compliance with regulations like GDPR.

The Adaptive Object Model (AOM) is an architectural style in which domain entity types are represented as instances that can be changed at runtime. It can be used to achieve higher flexibility in domain classes. Due to AOM entities having a distinct structure, they are not compatible with most popular frameworks, especially those that use reflection and code annotations. To solve such limitations, this study aims to propose a model that enables the reuse of frameworks designed for classic object-oriented domain models in an AOM application. The proposed model uses dynamically-generated adapters for AOM entities that encapsulate them in a class with the format expected by the frameworks. A reference implementation was developed in the Esfinge AOM RoleMapper framework to evaluate the viability of the proposed model. Initially, to evaluate the solution feasibility, a case study was carried out using the Hibernate framework. Further, an experiment was conducted to assess how the participants perceived the framework functionality reuse through the proposed model. The feasibility study revealed that the solution could be applied in a complex setting for the chosen object-relational mapping frame. It raised some difficulties that can be addressed in future studies. In the experiment, the development time did not present a significant difference compared to the competing approach. Despite the considerable learning curve, most participants considered that the proposed approach has more advantages than the alternative. Based on the evaluations, we can conclude that the proposed model can be successfully employed to use AOM entities with frameworks that were not designed for AOM applications. The possibility of reusing existing frameworks can reduce the effort required to adopt an AOM architecture and, consequently, be a facilitator in implementing more flexible and adaptive approaches.

Multiagent Systems (MASs) have multiple different characteristics, such as autonomy, and asynchronous and social features, which make these systems difficult to understand. Thus, there is a lack of procedures guaranteeing that multiagent systems once implemented would behave as desired. Determining the reliability of such systems is further complicated by the fact that current agent-based approaches may also involve non-deterministic characteristics, such as learning, self-adaptation and self-organization (SASO). Nonetheless, there is a gap in the literature regarding the testing of systems with these features. This paper presents an approach based on metadata and the publish-subscribe paradigm to develop test applications that address the process of failure diagnosis in a self-organizing MAS. The novelty of the proposed approach involves its ability to test self-organizing MAS systems in the context of local and global behavior. To illustrate the use of this approach, we developed a self-organizing MAS system based on the Internet of Things (IoT), which simulates a set of smart street lights, and we performed functional ad-hoc tests. The street lights need to interact with each other in order to achieve the global goals of reducing energy consumption and maintaining the maximum value of visual comfort in illuminated areas. To achieve these global behaviors, the street lights develop local behaviors automatically through a self-organizing process based on machine-learning algorithms.

Encrypted deduplication combines encryption and deduplication in a seamless way to provide confidentiality guarantees for the physical data in deduplicated storage, yet it incurs substantial metadata storage overhead due to the additional storage of keys. We present a new encrypted deduplication storage system called ${\sf Metadedup}$ Metadedup , which suppresses metadata storage by also applying deduplication to metadata. Its idea builds on indirection, which adds another level of metadata chunks that record metadata information. We find that metadata chunks are highly redundant in real-world workloads and hence can be effectively deduplicated. We further extend ${\sf Metadedup}$ Metadedup to incorporate multiple servers via a distributed key management approach, so as to provide both fault-tolerant storage and security guarantees. We extensively evaluate ${\sf Metadedup}$ Metadedup from performance and storage efficiency perspectives. We show that ${\sf Metadedup}$ Metadedup achieves high throughput in writing and restoring files, and saves the metadata storage by up to 93.94 percent for real-world backup workloads. [ABSTRACT FROM AUTHOR]

The proposes a high dynamic range (HDR) tone mastering system which dynamically corrects the picture quality based on creative intent metadata to preserve content providers’ creative intent in different ambient light levels/ranges. Luminance percentile information based creative intent metadata is used to compensate for imagery degradation caused by ambient light. Multiple sections of tone mapping curves with multiple adjustment points along explicit Bezier curve are modified for more flexible tone mapping curve control than simply single section tone mapping curves. Simulation results show that the proposed method can adapt to both ambient light level and the scene content to keep the creative intent in different ambient conditions. [ABSTRACT FROM AUTHOR]

Environmental Internet of Things (EIoT) has been widely applied in environmental monitoring. However, unified management, which is important for improving the operability and interoperability of EIoT systems, is difficult to perform due to the ubiquitous heterogeneity of EIoT systems. This paper puts forward an analysis of EIoT system management patterns and then proposes a storage layer monitoring (SLM) method to understand the state of EIoT systems, which is the basis for systems' management. A unified management framework for EIoT systems based on SLM is proposed, which includes event detection and fault diagnosis assistance. The framework uses a database to store metadata pertaining to each EIoT system, which are mostly registered as ontologies. Based on pre-registered metadata, users can create listeners for events, which is applicable for detecting events occurring in specific EIoT storage layers through the SLM method. The fault diagnosis assistance function can provide ontology-based evidence and other useful information about the participating objects of the event in the system, which is helpful for identifying the cause of each event. This framework could satisfy the requirements for multiple EIoT systems' management.

Continuous Data Protection (CDP) is an emerging scheme of data backup and recovery for Cyber Security. Compared with traditional data protection techniques, it can provide much higher reliability and allow the data state to be reverted to any point-in-time in the past. CDP continuously captures and logs every disk update, producing large amount of information, thus desperately requires an efficient mechanism of organizing such information (about the disk I/Os) in order to guarantee the performance to be acceptable. However, there are few works which address the organization of history data to improve the recovery efficiency in case of the disaster occurrences. In this paper, three methods of metadata management for Cyber Security are presented, of which two are simple implementations based on MySQL database (named DIR_MySQL and OPT_MySQL respectively) and the other one is designed with consideration of its characteristics. Experimental results show that META_CDP method is far more efficient than the other two and its performance is reasonably acceptable. Furthermore, we discuss in-depth two recovery algorithms, which are full-recovery and incremental recovery. Users can choose the optimal recovery algorithm based on the required recovery time point.

Massive music data and diverse listening behaviors have caused great difficulties for existing methods in user-personalized recommendation scenarios. Most previous music recommendation models extract features from temporal relationships among sequential listening records and ignore the utilization of additional information, such as music’s singer and album. Especially, a piece of music is commonly created by a specific musician and belongs to a particular album. Singer and album information, regarded as music metadata, can be utilized as important auxiliary information among different music pieces and may considerably influence the user’s choices of music. In this paper, we focus on the music sequential recommendation task with the consideration of the additional information and propose a novel Graph-based Attentive Sequential model with Metadata (GASM), which incorporates metadata to enrich music representations and effectively mine the user’s listening behavior patterns. Specifically, we first use a directed listening graph to model the relations between various kinds of nodes (user, music, singer, album) and then adopt the graph neural networks to learn their latent representation vectors. After that, we decompose the user’s preference for music into long-term, short-term and dynamic components with personalized attention networks. Finally, GASM integrates three types of preferences to predict the next (new) music in accordance with the user’s music taste. Extensive experiments have been conducted on three real-world datasets, and the results show that the proposed method GASM achieves better performance than baselines.

Extensive research on persistent memory (PM)-aware file systems has led to the development of numerous methods for improving read/write throughput. In particular, accessing or modifying file contents in a similar manner to the memory operations through mmap is a common approach. We designed a file system, CFFS (Contiguous File Allocation with Fine-Grained Metadata File System), to rapidly allocate PM pages to upper layer applications for mmap and to alleviate page fault overheads due to mmap. We optimized the physical contiguity of files in PM to reduce file fragmentation and increase fragment alignment with the goal of reducing software overhead. To achieve this goal, we implemented greedy-based buddy systems and implicit preallocation with a not-most-recently-used (NMRU) policy based on our overall page allocation strategy of considering not only the spatial but also the temporal locality of file access patterns. Furthermore, for efficient and atomic metadata operations, we fully leveraged the byte-addressable property of PM to design fine-grained metadata. CFFS adopts persistent doubly linked lists for directory operations to identify and recover from inconsistencies caused by system failures, doing so without using traditional log mechanisms. In experiments, CFFS showed superior page allocation performance to EXT4-DAX and NOVA did under different PM fragmentation levels. Our allocation algorithm also reduced the cost of page faults for frequently appended files. Finally, CFFS’s lightweight directory operations performed excellently in creating and deleting files of various quantities. In summary, the main contribution of the paper is proposing an efficient page allocation algorithm to improve the performance of subsequent mmaps, based on the strategy of considering not only the spatial but also the temporal locality of file access patterns in PM file systems. Another contribution was the fine-grained and log-free method for atomic directory operations.