A User and Entity Behavior Analytics Log Data Set for Anomaly Detection in Cloud Computing

User behavior visualization as a powerful tool can help analysts better understand user behavior data. Previous research such as Information Architecture believes that a user journey map is the regular way to visualize user behavior patterns. The technique is a diagram that shows the steps, touch-points, and emotions that users go through when using their products or services. Current studies mainly focus on the procedure and framework of applying user behavior data into visualization tools and its benefit to business development. Studies also found that visualization in user behavior analysis has difficulty in dealing with complex data. However, few studies have discussed the possible solution for the current limitation of visualization in user behavior analysis. With the popularization of the Internet and mobile devices, the scale and scale of user behavior data and the complexity will continue to increase, and user behavior visualization will be an important tool for analyzing this data, helping to tap potential business opportunities and improve the efficiency of data analysis. Therefore, we would make the review on visualization in user behavior analysis. We will first provide the background of the user behavior and its visualization and then introduce the definition of user behavior. Afterward, we will discuss the current process of visualization in user behavior analysis and its current limitations. Next, we will propose the possible solution of introducing machine learning, deep learning, predicting the data with a sequence model, and using cluster analysis with usability analysis tools to contribute to the future development of visualization in user behavior analysis. We believe that with data amount and scale keep increasing, the visualization tools need to develop themselves in diversity and data processing speed.

Analyzing user behavior characteristics in a complex power grid environment is essential for user behavior planning and resource coordination optimization. Traditional user behavior analysis methods based on model-driven and causal analysis have the disadvantages of strong subjectivity and physical models that are difficult to deal with the randomness and uncertainty of user behavior in complex grid environments. In this paper, we use unsupervised learning methods to analyze user behavior in complex power grid environments, and propose user behavior analysis methods based on stacked autoencoder and clustering. We first reduce the complexity of user behavior data by proposing adaptive feature selection algorithm of user behavior based on stacked autoencoder and unsupervised learning (AFS-SAEUL). Finally, we build a user behavior analysis model based on adaptive feature selection and improved clustering (UBA-AFSIC). The model improved the performance of unsupervised classification of user behavior by fusing the adaptive generation strategy of the initial cluster centers. The simulation experiment results on two real electricity datasets and one public electric vehicle charging dataset show that compared with the existing feature selection algorithm and clustering algorithm, the algorithms proposed in this paper have higher feature selection rate and better clustering performance.

With the rapid growth of e-commerce, consumer demand for personalized packaging solutions has grown significantly. To address this issue, this article constructs an intelligent recommendation model that combines user behavior data and sentiment analysis techniques. Firstly, a user behavior preference model is established by collecting potential preferences and behavioral characteristics of users, in order to explore their network behavior characteristics. Next, the Bidirectional Encoder Representations from Transformers (BERT) word vector is used to represent the comment text, and a bidirectional recurrent neural network is used to quantify the emotional information in the comment. Based on the emotional rating, the rating matrix is updated to map the shallow features of users and resources. Subsequently, by combining convolutional neural networks and self attention mechanisms, deep features of users and resources are extracted from comment texts, and shallow and deep features are fused through multi-layer neural networks to model the nonlinear interaction between users and resources and predict the rating values of recommended resources. In addition, this article proposes a hierarchical attention enhanced recommendation model incorporating User behavior and sentiment analysis (HAER-UBSA), which obtains feature information from comment texts through attention mechanisms and models comment level embedded representations of users and items. The experimental results show that compared with other baseline models, the mean absolute error (MAE) and root mean squared error (RMSE) indicators of our model have improved by 10.34% and 10.00%, respectively.

This paper presents a continuous authentication system based on user behavior analysis that makes use of environmental context information, users' behavior analysis and Neuro-Fuzzy Logic. This system must be able to acquire information in context, making them into a computational environment. This information is the basis of user behavior. The System, based on the evidences of the behavior, establishes if it can trust the user or not. According to the user behavior, levels of trust are released, to access the application software. Weights are attributed in the fuzzyfication process, according to the rules that were previously established for the parameters which help to establish the evidences of behavioral trust, in its different degrees. The neuro-fuzzy logic allows that the user behavioral database be continuously updated, interacting with the fuzzyfication mechanism, so as to keep trust levels updated according to the user behavior, in a more accurate and faithful way.

With the constant development of socialist modernization construction in our country, our country's computer information technology has made effective progress, and makes the world into an information age. People's production and life style experience a series of change. However, with the emergence of information diversity and multi-user mode, traditional computer information technology already can't satisfy people's needs, and the development and application of cloud computing technology come to the stage. This article focuses on analyzing the big data user behavior engine design under cloud computing technology, covering big data flow management, multi-user system design, and other aspects to explore the effect of the system test. In recent years, China's mobile Internet technology has made full development, which makes the Internet operators in our country face a new development opportunity and began from traffic management to flow management. The law of user behavior is analyzed to explore the real demand of market and the users. In order to fully meet the changing needs of users, operators must constantly develop and launch new products and strengthen the function of computer technology. Cloud computing technology is such kind of engine system which can satisfy the analysis and processing of huge amounts of data. RESEARCH ON CLOUD COMPUTING SYSTEM OVERALL DESIGN The overall architecture of cloud computing system This study mainly uses cloud computing technology’s huge amounts of data computing to set up complete mobile Internet data mining analysis system. Realize the analysis for Internet user behavior engine, and according to the user's preferences online habits and behavior, provide users with targeted personalized service and form a unified organic whole of data collection, analysis and service type and marketing strategy to improve enterprise's marketing efficiency. In addition, cloud computing system mainly realizes data collection with the help of FTP server, and makes distributed computing and data batch processing at the system interface. Large data shall be deposited in Hbase database. The system can not only realize mass data storage, but those unstructured data storage. Then through Hive integration layer and summary EIL processing, use MapReduce data analysis model to transmit the processing result into the database. The system’s overall structure is shown in Figure 1: 3rd International Conference on Materials Engineering, Manufacturing Technology and Control (ICMEMTC 2016) © 2016. The authors Published by Atlantis Press 1351 Figure 1 The overall architecture of cloud computing system System topology and function distribution System topology mainly refers that a server is taken as Hapdoop platform’s master node server, and the others are Hapdoop platform’s node servers. In general, node server can be dynamically extended according to the actual need, and the master node server not only assigns tasks and flow to node server, but also monitors the work execution of the node server. More than one node server participates to complete the task, and it can improve data processing efficiency. The master node server’s software running status is shown in Figure 2: Figure 2 Master node server function structure

Modern mass customization production allows user interaction activities to be distributed in the full product life cycle via multiple information systems. Investigating user behaviors across the boundaries of different domains helps to deeply integrate isolated fragmental profiles into a comprehensive one, and therefore can provide multi-dimensional, high-quality and valuable services. However, traditional user behavior analysis models are based on individual user profile information derived from separate domains, such as requirement analysis, design, supply chain, logistics, marketing, etc., which have not considered the whole complexity of mass customization manufacturing. In this paper, we introduce the concept of multi-dimensional semantic activity space, where user behavior features are merged and represented as combined vectors. User behavior patterns are discovered by mining action data extracted from log files in different subsystems in the corresponding domains. We also identify distinct categories of user behaviors in various modules and subsystems in the context of an intelligent manufacturing environment. Experiment results show a strong indication that the proposed approach can be applied to reveal variations in typical behavioral aspects of cross-domain participants, in terms of patterns in resource access, operation tasks, performance assessment, etc.

In today's world of digitization, anomaly detection has become one of the most important issues in our lives. User and Entity Behavior Analytics (UEBA) is a security solution for anomaly detection. UEBA minimizes the impact of attacks on data and decreases the risk of privacy breeches by keeping track of normal user and entity behaviors. Organizations need to deploy security control through UEBA in order to achieve security for their data by detecting attacks in their early stages, but the cost of such security control is often out of reach for many small and medium enterprises. One possible solution for such organizations is to outsource UEBA to third parties. However, such an approach can result in disclosure of private data. In this paper, we propose a novel scheme that integrates differential privacy into UEBA. We will show that by introducing noise to the data before publishing it to the third party for anomaly detection, we can ensure that the privacy of the data remains intact, while allowing the third party to preform accurate UEBA analysis. Our experimental results and security analysis will show the validity of our approach.

Enterprise systems often produce a large volume of logs to record runtime status and events. Anomaly detection from system logs is crucial for service management and system maintenance. Most existing log-based anomaly detection methods use log event indexes parsed from log data to detect anomalies. Those methods cannot handle unseen log templates and lead to inaccurate anomaly detection. Some recent studies focused on the semantics of log templates but ignored the information of parameter values. Therefore, their approaches failed to address the abnormal logs caused by parameter values. In this article, we propose HitAnomaly, a log-based anomaly detection model utilizing a hierarchical transformer structure to model both log template sequences and parameter values. We designed a log sequence encoder and a parameter value encoder to obtain their representations correspondingly. We then use an attention mechanism as our final classification model. In this way, HitAnomaly is able to capture the semantic information in both log template sequence and parameter values and handle various types of anomalies. We evaluated our proposed method on three log datasets. Our experimental results demonstrate that HitAnomaly has outperformed other existing log-based anomaly detection methods. We also assess the robustness of our proposed model on unstable log data.

The e-commercial systems have experienced a rapid development and have brought great benefit to people's daily life. Many e-commercial systems use recommend algorithm to filter irrelative information and recommend items to users. One of the most popular recommend algorithms is Collaborative Filtering Algorithm. However, there are some shortcomings in Collaborative Filtering Algorithm, causing that the algorithm cannot be well applied when the data set is sparse. This paper proposes a new item recommend algorithm which based on the analysis and prediction of user behavior by pattern recognition and statistic model that can be applied on sparse user behavior data set, avoiding the problems Collaborative Filtering Algorithm faced when the data set is sparse.

Japanese language instruction at universities has gained new life thanks to the quick growth of online learning via the Internet, microclasses, flipped classrooms, and other innovative teaching methods. This is the direction of future educational reform in colleges and universities. After gathering and pre-processing behavioral data of Japanese learners, this study builds a data analysis model of Japanese online learning user behavior based on the Japanese online education platform. In this model, user behavioral features are extracted and classified using the RFG-SVM model, which is based on SVM. Users with similar user behaviors are then clustered together using the FCM algorithm, and the association rule algorithm is utilized to explore the intricate relationship between user online learning behaviors and learning effects. Lastly, the FSQCA approach is used to investigate the optimization path of Japanese online education platforms after combining with example analysis. The most significant aspect of Japanese online learning is its online learning rate (0.7499). Users can be categorized into three groups: close cooperation (52.3%), active participation (6.1%), and weak participation (41.6%). Japanese online learners also exhibit better user behavior. The consistency indexes for Grouping H1: SI*LA*CC*HA, Grouping H2: ~SQ*CQ*SI*~LA*CC, and Grouping H3: SQ*CQ*SI*~LA*HA were 0.929, 0.959, and 0.965, respectively, with both social influence (SI) and habit (HA) serving as important requisites. This study helps to form a mature mechanism that prompts Japanese online education users to develop a continuous willingness to use the program, which contributes to the development of inclusive education to a certain extent.

Online news streaming services have been one of the major information acquisition resources for mobile users. In many cases, users click an article but find it cannot satisfy or even annoy them. Intuitively, these negative experiences will affect users' behaviors and satisfaction, but such effects have not been well understood. In this work, a retrospective analysis is conducted using real users' log data, containing user's explicit feedback of negative experiences, from a commercial news streaming application. Through multiple intra-session comparison experiments, we find that in current session, users will spend less time reading the content, lose activeness and leave sooner after having negative experiences. Later return and significant changes of user behaviors in the next session are also observed, which demonstrates the existence of inter-session effects of negative experiences. Since users' negative experiences are generally implicit, we further investigate the possibility and the approach to automatically identify them. Results show that using changes of both users' intra-session and inter-session behaviors achieves significant improvement. Besides the effects on user behaviors, we also explore the effects on user satisfaction by incorporating a laboratory user study. Results show that negative experiences reduce user satisfaction in the current session, and the impact will last to the next session. Moreover, we demonstrate users' negative feedback helps on the meta-evaluation of online metrics. Our research has comprehensively analyzed the impacts of users' item-level negative experiences, and shed light on the understanding of user behaviors and satisfaction.

As computer and network intrusions become more and more of a concern, the need for better capabilities to assist in the detection and analysis of intrusions also increases. We propose a methodology for analyzing network and computer log information visually based on the analysis of user behavior. Each user's behavior is the key to determining their intent and overriding goals, whether they attempt to hide their actions or not. Proficient hackers will attempt to hide their ultimate goal, which hinders the reliability of log file analysis. Visually analyzing the user's behavior, however, is much more adaptable and difficult to counteract. This paper will discuss how user behavior can be exhibited within the visualization techniques, the capabilities provided by the environment, typical characteristics users should look out for (i.e., how unusual behavior exhibits itself), and exploration paradigms effective for identifying the meaning behind the user's behavior.

The analysis of user behavior in online communities is a prominent topic in social media research. As such, user behavior is often analyzed using a set of metrics that describe the user’s participation behavior and structural position in the social network. Yet, for Enterprise Social Networks (ESN), i.e. internally used social networking platforms, such research is lacking. While prior studies have found users to engage in knowledge-intensive interactions, e.g. discussing and developing new ideas, little is known about how to conceptualize and measure ESN user behavior. Being able to measure user behavior, however, is an important prerequisite for the identification of knowledge management-related roles in the context of ESN.

We investigate the use of biased sampling according to the density of the data set to speed up the operation of general data mining tasks, such as clustering and outlier detection in large multidimensional data sets. In density-biased sampling, the probability that a given point will be included in the sample depends on the local density of the data set. We propose a general technique for density-biased sampling that can factor in user requirements to sample for properties of interest and can be tuned for specific data mining tasks. This allows great flexibility and improved accuracy of the results over simple random sampling. We describe our approach in detail, we analytically evaluate it, and show how it can be optimized for approximate clustering and outlier detection. Finally, we present a thorough experimental evaluation of the proposed method, applying density-biased sampling on real and synthetic data sets, and employing clustering and outlier detection algorithms, thus highlighting the utility of our approach.

Social media provides valuable resources to analyze user behaviors and capture user preferences. This article focuses on analyzing user behaviors in social media systems and designing a latent class statistical mixture model, named temporal context-aware mixture model (TCAM), to account for the intentions and preferences behind user behaviors. Based on the observation that the behaviors of a user in social media systems are generally influenced by intrinsic interest as well as the temporal context (e.g., the public's attention at that time), TCAM simultaneously models the topics related to users' intrinsic interests and the topics related to temporal context and then combines the influences from the two factors to model user behaviors in a unified way. Considering that users' interests are not always stable and may change over time, we extend TCAM to a dynamic temporal context-aware mixture model (DTCAM) to capture users' changing interests. To alleviate the problem of data sparsity, we exploit the social and temporal correlation information by integrating a social-temporal regularization framework into the DTCAM model. To further improve the performance of our proposed models (TCAM and DTCAM), an item-weighting scheme is proposed to enable them to favor items that better represent topics related to user interests and topics related to temporal context, respectively. Based on our proposed models, we design a temporal context-aware recommender system (TCARS). To speed up the process of producing the top- k recommendations from large-scale social media data, we develop an efficient query-processing technique to support TCARS. Extensive experiments have been conducted to evaluate the performance of our models on four real-world datasets crawled from different social media sites. The experimental results demonstrate the superiority of our models, compared with the state-of-the-art competitor methods, by modeling user behaviors more precisely and making more effective and efficient recommendations.