User Behavior Prediction Research Articles

Applying text mining methods for data loss prevention

Currently, the greatest risks for information security of organizations are internal, rather than external, threats. Data loss prevention (DLP) systems are used for minimization of risks related to internal threats. The main function of the DLP systems is to prevent leak of confidential data; however, comparison of the DLP systems relies currently on their capabilities to analyze information captured and convenience of carrying out retrospective investigations of information security incident. In the paper, a new approach to retrospective analysis of user's text information is presented. The idea of the proposed approach consists in topic analysis of the text content processed by the user in the past and prediction of further user behavior with content. User text content can cover different categories, including confidential ones. The topic analysis of user text content assumes determination of main topics and their weights for given past time intervals. Based on deviations of behavior of user's operations with a content from the forecast, one can reveal time intervals when operation with documents of one or another category differs from normal (historical) work and when the user worked with documents of unusual categories. The proposed approach was experimentally verified on an example of actual corporate email correspondence created from the Enron data set.

모바일 기반 라이프로그를 이용한 사용자 행동 예측 기법

모바일 기반 라이프로그를 이용한 사용자 행동 예측 기법

Mining and Predicting Smart Device User Behavior

Three types of user behavior are mined in this paper: application usage, smart device usage and periodicity of user behavior. When mining application usage, the application installation, most frequently used applications and application correlation are analyzed. The application usage is long-tailed. When mining the device usage, the mean, variance and autocorrelation are calculated both for duration and interval. Both the duration and interval are long-tailed but only duration satisfies power-law distribution. Meanwhile, the autocorrelation of both duration and interval is weak, which makes predicting user behavior based on adjacent behavior not so reasonable in related works. Then DFT (Discrete Fourier Transform) is utilized to analyze the periodicity of user behavior and results show that the most obvious periodicity is 24 hours, which is in agreement with related works. Based on the results above, an improved user behavior predicting model is proposed based on Chebyshev inequality. Experiment results show that the performance is good in accurate rate and recall rate.

What drives purchase intention for paid mobile apps? – An expectation confirmation model with perceived value

While many reports predict huge growth potential for the mobile application (app) market, little is known about user intention to purchase paid apps. This study amends the expectation confirmation model and incorporates app rating, free alternatives to paid apps and habit as belief-related constructs to predict user behavior. The proposed model was empirically evaluated using a survey of 507 respondents about their perceptions of app usage. The results indicated that confirmation was positively related to perceived value and satisfaction. Value-for-money, app rating and free alternatives to paid apps were found to have a direct impact on intention to purchase paid apps. Specifically, there was a significant difference between potential users and actual users. The results may provide further insights into app marketing strategies.

Interesting User Behaviour Prediction in Mobile E-commerce Environment using Constraints

ABSTRACTMobile e-commerce has become a more emerging topic these days and is speedily increasing with the growth of internet. It is playing an essential role in our life as it helps us by using new trends and technologies to improve business and thus providing quicker information. Behaviour analysis and sequence pattern prediction provide scope for more advanced research topics. If a user finds various interesting sequences in mobile e-commerce environment, then she/he can predict lots of new information. This predicted information may be useful for defining various new services or improving on existing and may also be useful to manage service infrastructure to offer quicker responses. Today, user behaviour sequence mining is an emerging issue in mobile e-commerce environment which considers mobile user services, accessing sequences at various locations on different times. In the current scenario, the user may only be interested in the service sequences with some specific constraints. The constraint may define to the type of service or importance of a particular service. Here, we consider important or useful services for particular user in mobile e-commerce environment. We proposed an efficient framework, namely CIUBSM (Constraint based Interesting User Behaviour Sequence Mining) which deals with important constraint of services and generates an interesting behaviour sequence of mobile users. Experimental results show that the proposed framework is better than previous constraint-based frameworks.

Efficient Pattern Mining and Prediction of User Behavior in Mobile Commerce

Efficient Pattern Mining and Prediction of User Behavior in Mobile Commerce

English

Web prediction is a classification problem in which we try to predict the preceding set of Web pages in which a user may visit supported on the knowledge of the previously visited pages. While serving the Internet user's behavior prediction can be applied effectively in various critical applications. Such application has usual tradeoffs between modeling complexity and prediction accuracy. In this paper we proposed artificial neural network (ANN) for predicting web by the user. In addition modified Markov model has been analysed and presented in prediction of web. A prediction framework uses ANN based on the training samples. By doing this the proposed framework shows the improved prediction time without compromising prediction accuracy. Web prediction is a classification scheme in which the next set of Web pages is predicted that user may visit pages depends on the knowledge of the previously visited pages. This information of user's history of navigation within a period of time is said to be session. These sessions provides source of data for training which are extracted from the logs of the Web servers, and they hold series of pages that users have visited with the visit date and duration. The Web prediction problem (WPP) can be simplified and utilized in major industrial applications such as wireless applications, caching systems and search engines. Consequently, it is critical to look for practical solutions in which it improves both prediction and training processes. The prediction process can be improved to reduce the user's access times during browsing, and it can simplify the network traffic by neglecting visiting needless pages. Once a prediction model for a definite Web site is available, the search engine can utilize it to cache the next set of pages that the users might visit. Those caching eases the latency problem of out looking Web documents certainly at the time of Internet traffic congestion periods. An additional extensive application of Web prediction is personalization, wherein users are sorted based on their desires and interests. The interests and desires of users are confined along with the previously visited sorted Web pages. Moreover, in the wireless domain, Web prediction in mobile is utilized to minimize the number of clicks required in wireless devices such as PDAs and smart phones that assists to alleviate problems related to the display size restrictions. The challenges in web prediction of both preprocessing and prediction are an issue now-a-days. Preprocessing challenges comprise handling huge amount of data that cannot fit in the computer memory, selecting optimum sliding window size, recognizing sessions, and extracting domain information. Prediction challenges comprise long prediction/ training time, memory limitation and low prediction accuracy.

English

The mobile phones grow to be an essential part of our everyday life, with smart phone sales at present greater than before very much and also user demands to run lots of applications have enhanced. The victory of next invention mobile phone communication based on the capability of service suppliers to engineer innovative added worth to video services. The Streams are programmed by the Scalable Video Coding expansion of the H.264/AVC model. Adding or removing the layer is determined on the basis of user behaviour environment of the mobile system. The current advances in the mobile video streams greater than mobile networks have been souring more than these new trends, the wireless connection service not basically hold up with the rising traffic demand. The hole among the traffic load and the wireless connection facility, along with time-changeable link circumstances, results reduced superiority of mobile video streams more than mobile networks, such as extensive buffering delays and discontinuous disruptions. The innovative cloud computing technology, we suggest a novel framework to get enhanced quality of video services for the mobile users, which has two parts: A newly distributed web User Behaviour Prediction model is bring in to cognize and guess user behaviour and additional Adaptive Policy Pre- fetching and Caching method is addressed for a well-organized cloud supervision. We can able to apply a new framework model to illustrate significant development in terms of lesser loss rate, decrease delay and buffering time. For all mobile user, make a private agent in the cloud data centre to alter the mobile video streams flow by the Scalable Video Coding method based on the reply from the mobile customer and perform the video prefetching based on the social network examination. In addition, the essential tools for providing temporal, spatial, and quality are described in feature and experimentally analyse their difficulty and effectiveness.

User behavior learning and transfer in composite social networks

Accurate prediction of user behaviors is important for many social media applications, including social marketing, personalization, and recommendation. A major challenge lies in that although many previous works model user behavior from only historical behavior logs, the available user behavior data or interactions between users and items in a given social network are usually very limited and sparse (e.g., ⩾ 99.9% empty), which makes models overfit the rare observations and fail to provide accurate predictions. We observe that many people are members of several social networks in the same time, such as Facebook, Twitter, and Tencent’s QQ. Importantly, users’ behaviors and interests in different networks influence one another. This provides an opportunity to leverage the knowledge of user behaviors in different networks by considering the overlapping users in different networks as bridges, in order to alleviate the data sparsity problem, and enhance the predictive performance of user behavior modeling. Combining different networks “simply and naively” does not work well. In this article, we formulate the problem to model multiple networks as “adaptive composite transfer” and propose a framework called ComSoc . ComSoc first selects the most suitable networks inside a composite social network via a hierarchical Bayesian model, parameterized for individual users. It then builds topic models for user behavior prediction using both the relationships in the selected networks and related behavior data. With different relational regularization, we introduce different implementations, corresponding to different ways to transfer knowledge from composite social relations. To handle big data, we have implemented the algorithm using Map/Reduce. We demonstrate that the proposed composite network-based user behavior models significantly improve the predictive accuracy over a number of existing approaches on several real-world applications, including a very large social networking dataset from Tencent Inc.

Measurement and analysis of P2P IPTV program resource.

With the rapid development of P2P technology, P2P IPTV applications have received more and more attention. And program resource distribution is very important to P2P IPTV applications. In order to collect IPTV program resources, a distributed multi-protocol crawler is proposed. And the crawler has collected more than 13 million pieces of information of IPTV programs from 2009 to 2012. In addition, the distribution of IPTV programs is independent and incompact, resulting in chaos of program names, which obstructs searching and organizing programs. Thus, we focus on characteristic analysis of program resources, including the distributions of length of program names, the entropy of the character types, and hierarchy depth of programs. These analyses reveal the disorderly naming conventions of P2P IPTV programs. The analysis results can help to purify and extract useful information from chaotic names for better retrieval and accelerate automatic sorting of program and establishment of IPTV repository. In order to represent popularity of programs and to predict user behavior and popularity of hot programs over a period, we also put forward an analytical model of hot programs.

Web Usage Mining: A Concise Survey on Tools and Applications

Web usage mining focuses on techniques that might predict user behavior whereas the user interacts with the net. It tries to create sense of the info generated by the net surfer’s sessions or behaviors. There has been an effort to supply a summary of the state of the art within the analysis of internet usage mining, whereas discussing the foremost relevant tools obtainable within the sphere likewise because the niche needs that this form of tools lack. It offers an outlook on the prevailing tools, their specialized focus with reference to the practical objectives and also the would like for a additional comprehensive new entrant during this sphere within the light-weight of this state of affairs. In the end, the paper are finished by listing some challenges and future trends during this analysis space. Overall the main target of the paper are to gift a survey of the recent developments during this space that is obtaining an excessive amount of attention from internet development arena.

Pattern-growth based frequent serial episode discovery

Frequent episode discovery is a popular framework for pattern discovery from sequential data. It has found many applications in domains like alarm management in telecommunication networks, fault analysis in the manufacturing plants, predicting user behavior in web click streams and so on. In this paper, we address the discovery of serial episodes. In the episodes context, there have been multiple ways to quantify the frequency of an episode. Most of the current algorithms for episode discovery under various frequencies are apriori-based level-wise methods. These methods essentially perform a breadth-first search of the pattern space. However currently there are no depth-first based methods of pattern discovery in the frequent episode framework under many of the frequency definitions. In this paper, we try to bridge this gap. We provide new depth-first based algorithms for serial episode discovery under non-overlapped and total frequencies. Under non-overlapped frequency, we present algorithms that can take care of span constraint and gap constraint on episode occurrences. Under total frequency we present an algorithm that can handle span constraint. We provide proofs of correctness for the proposed algorithms. We demonstrate the effectiveness of the proposed algorithms by extensive simulations. We also give detailed run-time comparisons with the existing apriori-based methods and illustrate scenarios under which the proposed pattern-growth algorithms perform better than their apriori counterparts.

Scalable Influence Analysis in Mobile Social Networks

Influence is a complex and subtle force that governs the dynamics of social networks as well as the behaviors of involved users. In large social networks, nodes are influenced by others for various reasons. Understanding influence can benefit various applications such as viral marketing, recommendation, and information retrieval. However, most existing works on social influence analysis have focused on verifying the existence of social influence. In this paper, an algorithm is proposed which utilizes the heterogeneous link information and the textual content associated with each node in the network to mine micro and macro influence. Based on the direct and indirect influence, the MapReduce framework for social influence analysis algorithm is proposed to derive the indirect influence between nodes. We further study how the discovered topic-level influence can help the prediction of user behaviors. Empirical studies on a large real-world mobile social network show that our algorithm has a good scalability performance.

A Framework for Personal Mobile Commerce Pattern Mining and Prediction

Due to a wide range of potential applications, research on mobile commerce has received a lot of interests from both of the industry and academia. Among them, one of the active topic areas is the mining and prediction of users' mobile commerce behaviors such as their movements and purchase transactions. In this paper, we propose a novel framework, called Mobile Commerce Explorer (MCE), for mining and prediction of mobile users' movements and purchase transactions under the context of mobile commerce. The MCE framework consists of three major components: 1) Similarity Inference Model (SIM) for measuring the similarities among stores and items, which are two basic mobile commerce entities considered in this paper; 2) Personal Mobile Commerce Pattern Mine (PMCP-Mine) algorithm for efficient discovery of mobile users' Personal Mobile Commerce Patterns (PMCPs); and 3) Mobile Commerce Behavior Predictor (MCBP) for prediction of possible mobile user behaviors. To our best knowledge, this is the first work that facilitates mining and prediction of mobile users' commerce behaviors in order to recommend stores and items previously unknown to a user. We perform an extensive experimental evaluation by simulation and show that our proposals produce excellent results.

A Semantic and Environment Models-Based User Behavior Prediction Scheme in Smart Home

A Semantic and Environment Models-Based User Behavior Prediction Scheme in Smart Home

Learning influence from heterogeneous social networks

Influence is a complex and subtle force that governs social dynamics and user behaviors. Understanding how users influence each other can benefit various applications, e.g., viral marketing, recommendation, information retrieval and etc. While prior work has mainly focused on qualitative aspect, in this article, we present our research in quantitatively learning influence between users in heterogeneous networks. We propose a generative graphical model which leverages both heterogeneous link information and textual content associated with each user in the network to mine topic-level influence strength. Based on the learned direct influence, we further study the influence propagation and aggregation mechanisms: conservative and non-conservative propagations to derive the indirect influence. We apply the discovered influence to user behavior prediction in four different genres of social networks: Twitter, Digg, Renren, and Citation. Qualitatively, our approach can discover some interesting influence patterns from these heterogeneous networks. Quantitatively, the learned influence strength greatly improves the accuracy of user behavior prediction.

Static and Dynamic User Portraits

User modeling and profiling has been used to evaluate systems and predict user behaviors for a considerable time. Models and profiles are generally constructed based on studies of users’ behavior patterns, cognitive characteristics, or demographic data and provide an efficient way to present users’ preferences and interests. However, such modeling focuses on users’ interactions with a system and cannot support complicated social interaction, which is the emerging focus of serious games, educational hypermedia systems, experience, and service design. On the other hand, personas are used to portray and represent different groups and types of users and help designers propose suitable solutions in iterative design processes. However, clear guidelines and research approaches for developing useful personas for large-scale and complex social networks have not been well established. In this research, we reflect on three different design studies related to social interaction, experience, and cross-platform service design to discuss multiple ways of identifying both direct users and invisible users in design research. In addition, research methods and attributes to portray users are discussed.

A Classification Algorithm to Improve the Design of Websites

In very short time today web has become an enormously important tool for communicating ideas, conducting business and entertainment. At the time of navigation, web users leave various records of their action. This vast amount of data can be a useful source of knowledge for predicting user behavior. A refined method is required to carry out this task. Web usages mining (WUM) is the tool designed to do this task. WUM system is used to extract the knowledge based on user behavior during the web navigation. The extracted knowledge can be used for predicting the users’ future request when user is browsing the web. In this paper we advanced the online recommender system by using a Longest Common Subsequence (LCS) classification algorithm to classify users’ navigation pattern. Classification using the proposed method can improve the accuracy of recommendation and also proposed an algorithm that uses LCS method to know the user behavior for improvement of design of a website.

A review and analysis of deterrence theory in the IS security literature: making sense of the disparate findings

Deterrence theory is one of the most widely applied theories in information systems (IS) security research, particularly within behavioral IS security studies. Based on the rational choice view of human behavior, the theory predicts that illicit behavior can be controlled by the threat of sanctions that are certain, severe, and swift. IS scholars have used deterrence theory to predict user behaviors that are either supportive or disruptive of IS security, and other IS security-related outcome variables. A review of this literature suggests an uneven and often contradictory picture regarding the influence of sanctions and deterrence theory in general in the IS security context. In this paper, we set out to make sense of the discrepant findings in the IS deterrence literature by drawing upon the more mature body of deterrence literature that spans multiple disciplines. In doing so, we speculate that a set of contingency variables and methodological and theoretical issues can shed light on the inconsistent findings and inform future research in this area. The review and analysis presented in this paper facilitates a deeper understanding of deterrence theory in the IS security domain, which can assist in cumulative theory-building efforts and advance security management strategies rooted in deterrence principles.

Paying Attention to the Man behind the Curtain

In the push to develop smart energy systems, designers have increasingly focused on systems that measure and predict user behavior to effect optimal energy consumption. While such focus is an important component in these systems' success, designers have paid substantially less attention to the people on the other side of the energy system loop-the supervisors of power generation processes. Smart energy systems that leverage pervasive computing could add to these supervisory control operators' workload. They'll have to predict possible power plant load and production changes caused by environmental and plant events, as well as dynamic system adaptation in response to consumer behaviors. Contrary to many assumptions, inserting more automation, including distributed sensors and algorithms to postprocess data, won't necessarily reduce operators' workload or improve system performance.