2011年11月17日

MGC - Machine Generated Content



In this article, I'd like to mention a new Web field, and check its posibility through some examples, and also to define its key factors.
As we know, the key feature of Web2.0 is User Generated Content, which shares a common feature with Web1.0, which is that: all content is generated by humen intentionally, and the data has already got its meaning after being generated.
As now Android-Inside mobile phones and devices are getting popular, this new trend might lead us into a new Web field, which feature is that: the meta data was generated or collected by machines(mobile phones, devices, etc) originally, the data got its meaning only after being processed in remote data center. For this field, I call it MGC, which is Machine Generated Content.

Now let's take some examples to help to understand, although they're might not be possible in the next couple of years.
Example 1: Health application.
Scenario: John suspects that his heart is not funcitoning well, so he wears a special watch for a week. Actually this wathc is a Android-Inside device, which can monitor and record his pulse and blood pressure in real time. The data will be sent to hospital every night. The data center will analyze it and generate a report after all data being received. Doctor will check the report and take it as assitance for diagnosis.

Example 2: Traffic application.
Scenario: John's using Google navigator as other million drivers. The chip inside navigator collects car's data such as location and speed in real time, and send the data out to data center in real time. Data center will analyze all of data and then generate a summory traffic report, and then send the report into navigator's traffic layer. So while driving, John can get the road traffic information in real time, so that he can adjust his route according to it.

Example 3: Monitor application.
Scenario: John has a very yound child, so John concerns his safety when he's working in office, so he installed several smart cameras, which he can access through remotely, and each minute a snap shot saved into album automatically. If all of the cameras cannot find the kid in the visual angle, then an alert notificaiton will to send to him through email and short message.

Through above examples, we can find some features of MGC: 1. The collecting and processing of data can be isolated, data could be collected in local, while be processed remotely. 2. Data didn't have its meaning before analysis, so the data looks like device logs, most of them are kind of useless.

Now let's analyze some key factors of MGC.
Factor 1: data collecting.
Data collecting includes these methods:
1. Mobile APP - APP recall mobile's sensors such as GPS/Gravity/Atmospheric-Pressure to collect data.
2. Mobile Device - Device attached to mobile phone via USB/Audio ports, and convert data by algorithm.
3. Android-Inside Device.

Factor 2: data transaction.
I'd like to advise these 2 methods for small companies:
1. Use Twitter API as a protocal, to transfer data via Internet.
2. Use Amazon EC2, and install OpenSource MicroBlog software, so that you can setup your private transaction cloud.
With these 2 methods, we can transfer data through Internet in real time.

Factor 3: data processing.
Data processing includes core business logic.
Take the human body as a metaphor, the sensors inside the body surface and neurons collects data, the nervous system transfers data, and the brain processes data.
And as for a typical future MGC application, Android is its neuron, Twitter its nervous system, and its brain located in Goolge or Amazon's data center.

2011年11月16日

MGC - 机器生成内容



本文试图探讨一个新的互联网领域,通过几个例子来说明其可能性,并定义了其关键因素。
我们知道,WEB2.0的特征是用户生成内容,它和早期的WEB1.0,以及目前流行的SNS都有一个共同的特征,即所有的内容都是由人有意识地生成的,其数据在生成之时已经有了明确的含义。
而内嵌ANDROID的智能手机和设备的日渐流行,似乎将带领我们进入一个崭新的互联网领域,其特征是:数据的原始内容是由机器(手机和设备)自动生成、收集的,数据只有在经过远端的业务中心处理之后才有明确的含义。这个网络领域,我称之为MGC - Machine Generated Content - 机器生成内容。

下面我们举几个例子来帮助理解,请用科幻的眼光来看待这些场景。
例子一:健康应用。
场景:张三怀疑自己的心脏功能下降,于是他在手腕上戴了一周的"表"。这个"表"是一个内嵌ANDROID的设备,能够实时监测他的脉搏、血压等数据。数据每天晚上发送到医院的中央数据库。数据中心在接收到完整的数据后,自动生成一份报告给主治医生。医生通过报告,再接合问诊,形成最后的诊断结论。

例子二:交通应用。
场景:张三和其他数百万的用户一样,使用GOOGLE的车载导航仪。导航仪的芯片能够实时收集车辆的位置、速度等信息,并实时发送到数据中心。数据中心在综合分析这些数据之后,得到实时、准确的交通流量报告,并下载覆盖到导航仪的交通流量图层。因此张三在开车的同时,可以实时接收到前方的路况信息,从而调整自己的行车路线。

例子三:监控应用。
场景:张三的孩子很小,他希望在上班时也能够看到他是否安全,因此他在家里安装了几个智能摄像头,视频可以通过网络实时查看,同时每分钟截一张图存放到相册中。如果几个摄像头同时捕捉不到小孩的画面,将触发警告,通过短信和邮件通知张三。

从以上的例子,我们可以看出MGC的一些特征:1、数据的采集和处理可以是分离的,采集一般在本地,而处理可以放在云端。2、数据在分析之前并没有明确的含义,因此数据类似设备日志,包含大量的冗余。

下面我们再分析一下MGC的关键因素。
因素一:数据采集。
数据的采集包含以下方式:
1、手机APP,APP通过手机内置的GPS、重力、气压等感应器采集数据。
2、手机外设,外部设备利用手机的USB、音频接口等与手机连接,并配合算法实现数据的转换。
3、内置ANDROID的专有设备。

因素二:数据传输。
对于小的公司,建议采用以下两种方式的数据传输。
1、将Twitter作为一种协议,通过互联网传输数据。
2、利用AMAZON EC2,通过开源MICROBLOG软件,搭建实时传输的私有云。
通过以上两种方式,我们可以将数据实时地通过互联网进行传输。

因素三:数据处理。
数据处理包含了核心的业务逻辑。
以人体作比喻的话,那么人体表面的感应器和神经元实现了数据采集,神经系统实现了数据传输,而大脑实现了数据处理。
而一个未来的典型MGC应用,可能就是以ANDROID为神经元,以TWITTER为神经系统,而大脑放在GOOGLE或AMAZON的机房里。