Mining smartphone data for app usage prediction and recommendations: A survey

Smartphones nowadays have become indispensable personal gadgets to support our activities in almost every aspect of our lives. Thanks to the tremendous advancement of smartphone technologies, platforms, as well as the enthusiasm of individual developers, numerous mobile applications (apps) have been created to serve a wide range of usage purposes, making our daily life more convenient. While these apps are used, data logs are typically generated and ambience context is recorded forming a rich data source of the smartphone users’ behaviors. In this paper, we survey existing studies on mining smartphone data for uncovering app usage patterns leveraging such a data source. Our discussions of the studies are organized according to two main research streams, namely app usage prediction and app recommendations alongside a few other related studies. Finally, we also present several challenges and opportunities in the emerging area of mining smartphone usage patterns.     

Output-oriented power saving mode for mobile devices

Fast battery discharge is still the most nerve wracking issue for smartphone users. Though many energy saving methods have been studied, still users are not satisfied with their phone’s battery. Power management system provides low power mode when the phone is not in use for a long time. When the user is interacting with the phone, current system assumes the user is interactive and should keep the device in active mode. However, this is not true. After the user’s interaction, the device processes the request and displays the result on the smartphone’s output device. During this period, the user cannot see any meaningful information from the phone. In this paper, we propose a new low power mode where we put smartphone’s output device into low power mode while phone is preparing result for the user. We named this as o-sleep, an output-oriented power saving mode. While a device is processing a user’s request, output from the device may require preparation time. We consider the situation as the device’s output idle time and put the phone’s user interfaces into sleep mode while maintaining other subsystems in active state. To prove our concept, we have applied our technique onto various smartphone applications with varying operation environment. From the experiment, we found that the smartphone entered the o-sleep mode up to 58% of its total usage time in various test scenarios. Usability study supported feasibility of our proposed method.     

Building a mobile movie recommendation service by user rating and APP usage with linked data on Hadoop

Movie recommendation systems are important tools that suggest films with respect to users’ choices through item-based collaborative filter algorithms, and have shown positive effect on the provider’s revenue. Given that mobile Apps are rapidly growing, the recommender is implemented to support web services in frontend Apps. Among those films recommended, users can give ratings and feedback, collecting film information from linked data concurrently. In order to solve cold-start problems, Cluster-based Matrix Factorization is adopted to model user implicit ratings related to Apps usage. Knowing that user rating data processing is a large-scale problem in producing high quality recommendations, MapReduce and NoSQL environments are employed in performing efficient similarity measurement algorithms whilst maintaining rating and film datasets. In this investigation, the system analyzes user feedbacks to evaluate the recommendation accuracy through metrics of precision, recall and F-score rates, while cold-start users make use the system with two MovieLens datasets as main rating reference in the recommendation system.     

Utilizing a Bluetooth remote lock system for a smartphone

With the advent of mobile devices and the convergence of wireless technologies and the Internet, both the content and the quality of research in this field are subject to regular change. A variety of state-of-the-art computing devices that are compatible with each other have been produced. These devices have the ability to interact with people. This is also known as pervasive computing. Particularly, as smartphones have recently become one of the most popular devices worldwide, various convenient applications are being released. Smartphones available today not only provide the ordinary internal processes such as dialing or receiving phone calls, sending text messages, and doing mobile banking, but also increasingly control various other devices that are part of our daily lives. In effect, this means that through smartphone applications, we can remotely control a variety of external devices such as televisions, projectors for presentations, computers, and even cars. The research in this paper is based on the evolving technological possibilities of using smartphone applications to control external devices. This paper presents the design and implementation of a remote lock system using wireless communication on a smartphone. In this context, remote lock system refers to a lock system that can be controlled remotely by a dedicated Android application. Every smartphone is equipped with Bluetooth which makes this technology possible. The application proposed in this paper uses the existing Bluetooth function on Android smartphones to open and manage locks. The users’ lock information can be stored and managed in real time in the database via a server that is built and managed by a server manager. Even if users forget the password of the lock, our proposed lock system can guide them to retrieve it easily, and a user manual is included to help users navigate the system. This system also provides a variety of management functions such as adding, deleting, modifying, and purchasing the user’s own locks.     

Context Management for Adaptive Information Systems

Modern technologies enable users accessing services using multiple channels. In the service design phase, this poses additional requirements for high software adaptivity along different technical requirements and different user expectations. During execution, services are usually dynamically selected; this service selection phase requires the identification of the most suitable service along the context that characterizes the users in the time instant in which they send the service request. This paper focuses on the selection phase and it aims at providing a framework to define and manage the context in a general environment characterized by adaptivity and multichannel access. An example to demonstrate the suitability and feasibility of the framework is provided referring to the MAIS (Multichannel Adaptive Information Systems) architecture and considering services related to the tourism domain. The MAIS architecture aims at providing automatically and efficiently services with the appropriate features by choosing among many provider offerings.     

Helpful-stressful cycle? Psychological links between type of mobile phone user and stress

The profusion of smartphones allows more people to have access to a telephone, a computer and the Internet, all via one device. Despite this convenience, excessive usage and habitual checking can cause significant stress for smartphone users. A helpful-stressful cycle is beginning to emerge as a pattern. With this in mind, we develop a model that examines the differences between smartphone and traditional mobile phone users with regard to the relationships between four psychological factors: locus of control (LOC), social interaction anxiety (SIA), need for touch (NFT) and materialism. Our examination explores this in the context of the level of stress these two types of users experience when using their chosen phone. The results from our empirical study suggest that user type moderates the aforementioned relationships. The effects of LOC, SIA and NFT on ‘technostress’ are stronger for smartphone users than for traditional mobile phone users. In contrast, materialism is positively related to stress for traditional mobile phone users but not for smartphone users.     

A personal visual analytics on smartphone usage data

The percentage of individuals frequently using their smartphones in work and life is increasing steadily. The interactions between individuals and their smartphones can produce large amounts of usage data, which contain rich information about smartphone owners’ usage habits and their daily life. In this paper, a personal visual analytic tool is proposed to develop insights and discover knowledge of personal life in smartphone usage data. Four cooperated visualization views and many interactions are provided in this tool to visually explore the temporal features of various interactive events between smartphones and their users, the hierarchical associations among event types, and the detailed distributions of massive event sequences. In the case study, plenty of interesting patterns are discovered by analyzing the data of two smartphone users with different usage styles. We also conduct a one-month user study on several invited volunteers from our laboratory and acquaintance circle to improve our prototype system based on their feedback.     

Intelligent healthcare service based on context inference using smart device

In the field of “U-Healthcare Service”, many studies have been actively conducted to develop “smart device”-based healthcare applications that enable healthcare providers and patients to be better served through interoperations among various kinds of sensors and wireless network interfaces. In particular, contemporary intelligent healthcare services not only recognize users’ context information through smart devices, computers, and so forth, but also acquire information from heterogeneous sensors to achieve context-aware inference services. Among such information, the weather information is tightly related to diseases such as asthma and allergies. Therefore, there is a high demand for research to utilize the weather information for healthcare services. In this paper, we propose a context inference-based intelligent healthcare service that exploits both the weather conditions information and the diverse healthcare ontologies available on the Internet. The proposed service aims at modeling a context ontology in users’ healthcare service environment and defining the inference rules, thereby accomplishing a satisfactory real-time healthcare service.     

Implications of smartphone user privacy leakage from the advertiser’s perspective

Many smartphone apps routinely gather various private user data and send them to advertisers. Despite recent study on protection mechanisms and analysis on apps’ behavior, the understanding of the consequences of such privacy losses remains limited. In this paper, we investigate how much an advertiser can infer about users’ social and community relationships. After one month’s user study involving about 190 most popular Android apps, we find that an advertiser can infer 90% of the social relationships. We further propose a privacy leakage inference framework and use real mobility traces and Foursquare data to quantify the consequences of privacy leakage. We find that achieving 90% inference accuracy of the social and community relationships requires merely 3 weeks’ user data. Finally, we present a real-time privacy leakage visualization tool that captures and displays the spatial–temporal characteristics of the leakages. The discoveries underscore the importance of early adoption of privacy protection mechanisms.     

Ontology driven interactive healthcare with wearable sensors

Ubiquitous healthcare is the service that offers health-related information and contents to users without any limitations of time and space. Especially, to offer customized services to users, the technology of acquiring context information of users in real time is the most important consideration. In this paper, we researched wearable sensors. We proposed the ontology driven interactive healthcare with wearable sensors (OdIH_WS) to achieve customized healthcare service. For this purpose, wearable-sensor-based smart-wear and methods of data acquisition and processing are being developed. The proposed system has potential value in healthcare. A smart wear using wearable sensors is fabricated as a way of non-tight and comfortable style fitting for the curves of the human body based on clothes to wear in daily life. The design sample of the smart wear uses basic stretch materials and is designed to sustain its wearable property. To offer related information, it establishes an environment-information-based healthcare ontology model needed for inference, and it is composed of inside-outside context information models depending on the users’ context. The modeling of the proposed system involved combinations of information streams, focusing on service context information. With the proposed service inference rules, customized information and contents could be drawn by the inference engine. In the established OdIH_WS, real-time health information monitoring was achieved. The results of system performance and users’ satisfaction evaluations confirmed that the proposed system is superior to other existing systems.     

Utilising the co-occurrence of user interface interactions as a risk indicator for smartphone addiction

The push to a connected world where people carry an always-online device which has been designed to maximise instant gratification and prompts users via notifications has lead to a surge of potentially problematic behaviour as a result. This has lead to a rising interest in addressing and understanding the addictiveness of smartphone usage, as well as for particular applications (apps). However, capturing addiction from usage involves not only assessment of potential addiction risk but also requires understanding of the complex interactions that define user behaviour and how these can be effectively isolated and summarised. In this paper, we examine the correlation of physical user interface (UI) interactions (e.g. taps and scrolls) and smartphone addiction risk using a large dataset of those smartphone events (65,093,343, N=301,024 sessions) collected from 64 users over an 8-week period with an accompanying smartphone addiction survey. Our novel method which reports on the probability of a users addiction risk and in a model case we show how it was be used to identify 57 of 64 users correctly. This supports our observations of UI events during sessions of usage being indicative of addiction risk while improving previous approaches which rely on summative data such as screen on time. Within this we also find that users only exhibit addictive behaviour in a subset of all sessions while using their smartphone.     

Designing personal attentive user interfaces in the mobile public safety domain

In the mobile computing environment, there is a need to adapt the information and service provision to the momentary attentive state of the user, operational requirements and usage context. This paper proposes to design personal attentive user interfaces (PAUI) for which the content and style of information presentation is based on models of relevant cognitive, task, context and user aspects. Using the police work environment as the application domain, relevant attributes of these aspects are identified based on literature and domain analyses. We present a user-centered design (UCD) method for the iterative development and validation of the proposed PAUI. Application of this approach provided requirements for (1) adaptation to users’ attentive state, (2) notification, (3) information processing and task switching support and (4) user modeling. We aim at refining and validating the models and requirements through continuing empirical evaluation.     

智能环境基于用户交互模态的情境感知服务

智能环境系统构成的复杂性和大量的情境信息增加了为用户提供和布局服务的难度。情境感知服务系统能感知系统运行和人机交互情境,根据相应的决策机制选择适当的设备为用户提供服务,从而简化环境设施的功能布局。在决策过程中引入用户交互模态信息有利于系统根据用户的交互能力和对人机界面的偏好提供服务。把情境细分为用户信息、设备信息、软件信息和环境拓扑信息,针对智能环境提出了一种基于用户交互模态的情境感知服务系统,同时详细说明了利用该技术选择服务的推理过程及其应用方法。通过智慧家庭环境中的场景案例验证了该系统的有效性。     

Design Strategies for Idle Screens of Smartphones Based on Service Usages

Linking specific smartphone services to user characteristics can inform design guidelines and improve the competitive edge of smartphone models. A key component in the design of the user interface is the idle screen, as it is the starting point for all smartphone services. The focus of this study is to elicit insights for establishing strategies for improving idle screen design by relating user behaviors and smartphone services. Benefit segmentation is employed for users’ responses to questionnaire items about service usages. Smartphone services are identified through a factor analysis and user segments are obtained by a cluster analysis. The findings can contribute to classifying smartphone services according to users’ behaviors and preferences for them and to suggesting a design alternative for idle screens, depending on the features of user groups for improved market segmentation.     

Detecting privacy leaks in the RATP App: how we proceeded and what we found

We analyzed the RATP App, both Android and iOS versions, using our instrumented versions of these mobile OSs. Our analysis reveals that both versions of this App leak private data to third-party servers, which is in total contradiction to the In-App privacy policy. The iOS version of this App doesn’t even respect Apple guidelines on cross-App user tracking for advertising purposes and employs various other cross-App tracking mechanisms that are not supposed to be used by Apps. Even if this work is illustrated with a single App, we describe an approach that is generic and can be used to detect privacy leaks from other Apps. In addition, our findings are representative of a trend in Advertising and Analytics (A&A) libraries that try to collect as much information as possible regarding the smartphone and its user to have a better profile of the user’s interests and behaviors. In fact, in case of iOS, these libraries even generate their own persistent identifiers and share it with other Apps through covert channels to better track the user, and this happens even if the user has opted-out of device tracking for advertising purposes. Above all, this happens without the user knowledge, and sometimes even without the App developer’s knowledge who might have naively included these libraries during the App development. Therefore this article raises many questions concerning both the bad practices employed in the world of smartphones and the limitations of the privacy control features proposed by Android/iOS Mobile OSs.     

Secure Encrypted Steganography Graphical Password scheme for Near Field Communication smartphone access control system

The revolutionary development of smartphone which offers compelling computing and storage capabilities has radically changed the digital lifestyles of users. The integration of Near Field Communication (NFC) into smartphone has further opened up opportunities for new applications and business models such as in industry for payment, electronic ticketing and access control systems. NFC and graphical password scheme are two imperative technologies that can be used to achieve secure and convenient access control system. One of the potential uses of such technologies is the integration of steganography graphical password scheme into NFC-enabled smartphone to transcend conventional digital key/tokens access control systems into a more secure and convenient environment. Smartphone users would have more freedom in customizing the security level and how they interact with the access control system. As such, this paper presents a secure two-factor authentication NFC smartphone access control system using digital key and the proposed Encrypted Steganography Graphical Password (ESGP). This paper also validates the user perception and behavioral intention to use NFC ESGP smartphone access control system through an experiment and user evaluation survey. Results indicated that users weigh security as a dominant attribute for their behavioral intention to use NFC ESGP smartphone access control system. Our findings offer a new insight for security scholars, mobile device service providers and expert systems to leverage on the two-factor authentication with the use of NFC-enabled smartphone.     

Anonymous smartphone data collection: factors influencing the users’ acceptance in mobile crowd sensing

Mobile crowd sensing (MCS) assumes a collaborative effort from mobile smartphone users to sense and share their data needed to fulfill a given MCS objective (e.g., modeling of urban traffic or wellness index of a community). In this paper, we investigate the user’s perception of anonymity in MCS and factors influencing it. We conducted a 4-week extensive smartphone user study to fulfill three main objectives. (1) Understand if users prefer to share data anonymously or not anonymously. (2) Investigate the possible factors influencing the difference between these two modalities, considering: (a) users’ sharing attitude, (b) shared data kind and (c) users’ intimacy when data are shared (we defined intimacy as the users’ perception of their context with respect to place, number and kind of people around them). (3) Identify further users’ personal factors influencing their perception of anonymity via multiple interviews along the user study. In the results, we show that data are shared significantly more when anonymously collected. We found that the shared data kind is the factor significantly contributing to this difference. Additionally, users have a common way to perceive anonymity and its effectiveness. To ensure the success of anonymization algorithms in the context of MCS systems, we highlight which issues the researchers developing these algorithms should carefully consider. Finally, we argue about new research paths to better investigate the user perception of anonymity and develop anonymous MCS systems that users are more likely to trust based on our findings.     

A user-centric evaluation of context-aware recommendations for a mobile news service

Traditional recommender systems provide personal suggestions based on the user’s preferences, without taking into account any additional contextual information, such as time or device type. The added value of contextual information for the recommendation process is highly dependent on the application domain, the type of contextual information, and variations in users’ usage behavior in different contextual situations. This paper investigates whether users utilize a mobile news service in different contextual situations and whether the context has an influence on their consumption behavior. Furthermore, the importance of context for the recommendation process is investigated by comparing the user satisfaction with recommendations based on an explicit static profile, content-based recommendations using the actual user behavior but ignoring the context, and context-aware content-based recommendations incorporating user behavior as well as context. Considering the recommendations based on the static profile as a reference condition, the results indicate a significant improvement for recommendations that are based on the actual user behavior. This improvement is due to the discrepancy between explicitly stated preferences (initial profile) and the actual consumption behavior of the user. The context-aware content-based recommendations did not significantly outperform the content-based recommendations in our user study. Context-aware content-based recommendations may induce a higher user satisfaction after a longer period of service operation, enabling the recommender to overcome the cold-start problem and distinguish user preferences in various contextual situations.     

Linking information and people in a social system for academic conferences

This paper investigates the feasibility of maintaining a social information system to support attendees at an academic conference. The main challenge of this work was to create an infrastructure where users’ social activities, such as bookmarking, tagging, and social linking could be used to enhance user navigation and maximize the users’ ability to locate two important types of information in conference settings: presentations to attend and attendees to meet. We developed Conference Navigator 3, a social conference support system that integrates a conference schedule planner with a social linking service. We examined its potential and functions in the context of a medium-scale academic conference. In this paper, we present the design of the system’s socially enabled features and report the results of a conference-based study. Our study demonstrates the feasibility of social information systems for supporting academic conferences. Despite the low number of potential users and the short timeframe in which conferences took place, the usage of the system was high enough to provide sufficient data for social mechanisms. The study shows that most critical social features were highly appreciated and used, and provides direction for further research.     

Frequency and recency context for the management and retrieval of personal information on mobile devices

As users store increasingly larger amounts of personal information on their mobiles, the task of retrieving such items (e.g., contacts) becomes more difficult. We show that users can be categorized by their communication patterns and that each category benefits differently from supporting contact management applications. By examining mobile user call logs, we show that it is possible to aid retrieval tasks using relatively simple heuristics and algorithms that describe usage context, using solely the dimensions of contact use frequency and recency. We compare and discuss the results of the proposed method applied on two different mobile datasets: a large dataset from NOKIA and a smaller dataset collected by ourselves.