Mobile Context Data Mining: Methods and Applications

Mobile Context Data Mining: Methods and Applications

Enhong Chen (University of Science and Technology of China, China), Tengfei Bao (University of Science and Technology of China, China) and Huanhuan Cao (Nokia Research Center, China)
Copyright: © 2013 |Pages: 20
DOI: 10.4018/978-1-4666-2806-9.ch002


The mobile devices, such as iPhone, iPad, and Android are becoming more popular than ever before. Many mobile-based intelligent applications and services are emerging, especially those location-based and context-aware services, e.g. Foursquare and Google Latitude. The mobile device is important since it can detect a user’s rich context information with its in-device sensors, e.g. GPS, Cell ID, and accelerometer. With such data and suitable data mining methods better understanding of users is possible; smart and intelligent services thus can be provided. In this chapter, the authors introduce some mobile context mining applications and methods. To be specific, they first show some typical mobile context data types with a mobile phone which can be detected. Then, they briefly introduce mining methods that are related to two mostly used types of mobile context data, location, and accelerometer. In the following, we illustrate in detail two context data mining methods that process multiple types of context data and can deal with the more general problem of user understanding: how to mine users’ behavior patterns and how to model users’ significant contexts from the users’ mobile context log. In each section, the authors show some state-of-the-art works.
Chapter Preview

1. Introduction

In recent years, the mobile device market is continuously increasing in a very fast speed. Users are more likely to use their mobile devices to play games, listen to music, and browse Internet anywhere and anytime. Mobile device is becoming important than ever before. A special feature of mobile device is its personality to user. Unlike traditional devices, e.g., PC, mobile device can detect user’s rich mobile contexts, such as, user location and physical activity. More importantly, with the mobile context data, user understanding is becoming possible, e.g., mining habits of user in usual days, mining significant locations that user has visited, mining the contexts that the user likes to play games or listen to music, and inferring transportation mode when user is on the way, and so on. When we can understand user better, corresponding intelligent services would be provided naturally. Especially, in recent days, location based services are becoming popular. For example, users vote their favorite places with the mobile devices, which can detect location and publish the voting results through the social network. This has benefits both for those locations that users advertise for them and for users to let them have many funny things. Such application has attracted billions of users to engage in it.

Mobile devices are usually equipped with many sensors. Users may be familiar with two types of sensors, GPS sensor and Accelerometer sensor. With the GPS sensor, user can query where they are and with the Accelerometer sensor, user’s movement and activity can be detected, this function is needed by many games. Besides these, there are also other sensors, like WiFi, Light, and Temperate sensors, they all play part in the mobile context sensor system and detect the user’s context in different perspectives. Not at least, the camera and microphone can also be looked as sensors by which we can get the user’s surrounding pictures and sounds. Table 1 shows the common sensors that the state-of-the-art mobile phones have. In addition, Cell ID which represents the location of cell site can also be used to reflect user’s location, but with low accuracy.

Table 1.
Mobile phone sensors
Sensor NameDescription
GPSDetect location in out-door, have high accuracy.
WiFiDetect network connection, can also detect location.
AccelerometerDetect movement and gravity change.
CompassDetect the direction change.
BluetoothDetect nearby Bluetooth devices.
LightDetect light change.
ProximityDetect the presence of nearby objects without physical contact.
BarometerDetect change of atmospheric pressure.
GyroscopeDetect orientations.
CameraDetect nearby pictures.
MicrophoneDetect nearby sounds.

Complete Chapter List

Search this Book: