数据挖掘工具Kimono，将会改变App开发

互联网上的网页数量超过了20亿，它就像是一个庞大的原始信息库。相比而言，互联网目前大约只有1万个网页应用程序接口(API)——应用程序接口是一种虚拟渠道，它可以让软件开发人员访问、处理、以及重新打包相关数据（让小编用通俗的语言解释一下：API就是汽车的钥匙、仪表盘什么的，直接拿过来可以用，但是要修的话你就得研究下它是怎么构造的了）。

换句话说，如果你想要利用网页上的一些东西，那么就需要自己去挖掘。即使对于一些知道如何实现数据挖掘的人来说，这份工作也是比较单调乏味的。不过，Ryan Rowe和Pratap Ranade想要改变这一状况。

在过去的五个月时间里，Rowe和Ranade开发出了一款网页应用——Kimono，它允许用户从任意一个网站上获取数据，并把它即时转化成一个应用程序接口（API）。用户只需使用书签工具把想要挖掘的网站部分给标注出来，接下来，Kimono帮你搞定剩下的一切。对于那些需要程序入口的开发人员来说，可以将Kimono获取的代码加工一下，整合进自己的应用程序里；而对于那些不会写代码的用户，Kimono会自动修订抓取数据，将其整合成为动态图表，列表，或是简单的网页应用程序。从本质上来说，Kimono是一个即点即选工具包，它可以把网页进行拆分，然后让人们各取所需，用它构建新的东西。

Kimono已经从一些知名风险投资人(比如Ron Conway)那里筹集到了资金，而且他们的创始人也已经收到了几份收购offer。目前Kimono的用户数量大约有1.5万，他们目前还处于Beta测试阶段。但是对于Rowe和Ranade来说，事情才刚刚开始。

消除瓶颈

Kimono这个想法来自于Rowe，当时他在一家设计公司Frog担任开发工程师，工作中，他总是遇到许多同样的问题，非常令人沮丧。公司的设计师可能产生了一个想法， 但是其中会涉及到一些网页方面的东西，但是他们必须要先去找开发人员，才能确定自己的想法是否能够实现。

“你肯定遇到过这种状况，设计师和分析师的确想要用数据做出一些东西，但是他们却不知从何获取,” Ranade说道, “我们解决了这个瓶颈。”

对于那些非专业人士，他们对互联网并不是很了解，特别是涉及到一些流媒体，源代码，或是API这些专业术语时，所以他们不太能够了解Kimono的潜力。但是一些早期使用者已经开始在各种项目中使用Kimono工具了。当Rowe和Ranade注意到索契冬奥会没有相关的官方API时，他们就用Kimono工具自己创建了一个。开发人员和设计师可以点击此链接下载这个工具，之后，他们就可以开发奥运奖牌榜应用程序，赛程还有一些动态地图，等等。

让数据挖掘变得更加民主大众化

有些Kimono的使用案例听起来会让人觉得很难理解，在某种意义上来说，它的确是这样的。但是实际上Kimono把数据挖掘这项技术带到了更多人面前。Kimono工具让艺术家，历史学家，社会学家可以从各种数据源里进行挑选、并整合内容，最终用一种新颖的方式表达出来。

用一种不同的方式纵观世界

就短期而言，Rowe和Ranade计划通过向用户收费的方式赚钱，Kimono会根据用户使用API的数量，以及更新的频率向用户收取不同的费用，目前该服务仍然处于Beta测试阶段，也就是说，现在任何人都可以使用不限量的应用程序接口。不过，他们已经听说许多企业客户都对Kimono感兴趣。

但是，这两个人早已经开始考虑更大的发展蓝图了。对他们来说，Kimono如今巨大的潜力完全可以应用在移动手机，以及下一代可穿戴设备，甚至是物联网这些领域里面。

虽然现在还有不少人对Kimono持有怀疑态度，但是如果可以先暂时不考虑那些质疑的话，Kimono还是有非常大的潜力的。Rowe为Kimono的未来发展进行了设想，他认为，Kimono工具完全可以作为新一级交互体验的结缔组织，就像是物联网的神经系统一样，并以此为基础，生根发芽。你可以想象一下，Kimono不单单用在网站这类领域里，而是去挖掘其他类型的数据流，比如说，让实物对象对声音有所反应，或是开发一些应用程序对现场视频进行回应，等等。

在未来，Kimono的应用范围绝不仅仅是在网页抓取和数据挖掘领域。“现在，Kimono工具可以把一个网站转换成一个应用程序接口，仅从这点就可以看出它的能力是非常强大的，”Rowe说道，“但是以后，如果能把任何东西都转化成一个应用程序接口，那么绝对能创造历史。”

英语原文：

The number of web pages on the internet is somewhere north of two billion, perhaps as many as double that. It’s a huge amount of raw information. By comparison, there are only roughly 10,000 web APIs–the virtual pipelines that let developers access, process, and repackage that data. In other words, to do anything new with the vast majority of the stuff on the web, you need to scrape it yourself. Even for the people who know how to do that, it’s tedious. Ryan Rowe and Pratap Ranade want to change that.

For the last five months, Rowe and Ranade have been building out Kimono, a web app that lets you slurp data from any website and turn it instantly into an API. Using a bookmarklet, you highlight the parts of a site you want to scrape and Kimono does the rest. Those with programming chops can take the code Kimono spits out bake it into their own apps; for the code illiterate, Kimono will automatically rework scraped data into a dynamic chart, list, or a simple web app. In essence, it’s a point and click toolkit for taking apart the web, with the aim of letting people build new things with it.

Excitement’s already bubbling around the potential. Kimono’s already raised money from big-name VCs like Ron Conway and its founders have had to turn down at least one offer for an early buy-out. The site’s already managing some 15,000 users–and it’s still in beta. But for Rowe and Ranade, things are just getting started.

Eliminating the Bottleneck

The idea for Kimono was born out of Rowe’s time as a developer at the design consultancy Frog, where he continually ran into the same frustrating problem. A designer would have an idea that revolved around web stuff of one sort or another, but they’d have to find a developer before they could even get a sense of how the idea might actually work. “Getting the data just to prove if these apps would be interesting or not took a huge amount of time, which sucked,” Rowe says.

“You have these situations where designers and analysts really want to do stuff with data but have no means to get it,” adds Ranade, whose most recent gig was at consulting firm McKinsey & Company. “We realized that there doesn’t need to be that bottleneck.”

To laypeople who don’t already think of the web in terms of streams, sources, or APIs, it can be hard to grasp Kimono’s potential. But early adopters are already using it for a striking variety of projects. When they noticed there was no official API for the recent Sochi Olympics, Rowe and Ranade used Kimono to create one themselves. Devs and designers took it from there, building elegant medal-tracking apps, dynamic maps that visualize when and where Olympians were born, and more.

Around the time the Kimono beta went live last month, Golan Levin, a pioneer of computational art and design, was introducing his students at Carnegie Mellon to the unglamorous first steps of any data viz project: acquiring, parsing, and cleaning data. He thought it’d be valuable to acquaint them with the process. While new tools like Temboo are making it easier than ever to work with official APIs for big-name sites, there traditionally haven’t been straightforward ways to get structured data off the majority of pages on the web. “Kimono came along and really changed that,” Levin says.

Levin himself is using Kimono to track real estate purchases in his home town of Pittsburgh. He also cited an upcoming meeting of civic-minded coders called the Pittsburgh Data Brigade, where he expected Kimono to see some use. “Pittsburgh’s information systems are so old and creaky that getting data out is really hard,” he explains. It’s a problem many municipalities face; they’re eager to open up their data but lack the means to actually open it up. Kimono could help bridge that gap.

Democratizing Data Scraping

These use cases might sound esoteric, and in some senses, they are. But part of the ambition with Kimono is bringing data scraping to a wider audience. It’s about letting artists, historians, sociologists and more cull and combine content from various sources and present it in novel ways.

As an example, Ranade brings up Malcolm Gladwell’s theory about elite hockey players and how their success might be explained by where their birthdays fall in relation to Canada’s little league cutoff dates. A successful author like Gladwell can presumably tap a research assistant to trawl Wikipedia and collect the relevant data. A grad student probably cannot. With Kimono, however, she could amass a list of Wikipedia URLs, point Kimono to the “date of birth” and “place of birth” fields, and let it corral the data for her.

This sort of birthday/little league cutoff connection isn’t going to be made by a random developer, Ranade posts, but rather by a person who has “domain knowledge” in that field. “They might not be a programmer,” he says. “But if we gave a little bit of programming capability to that person, how could they look at the world in a different way?”

Looking at the World in a Different Way

In the short term, Rowe and Ranade plan to make money by charging users depending on how many APIs they use and how frequently they update (right now the service is in beta, and anyone can make however many APIs they want). They’ve already heard interest from a number of corporate clients, who see Kimono as a means to free the flow of data between departments and project teams without relying on an internal IT team to act as translator in-between.

But the duo is already thinking even bigger picture. To them, Kimono’s greatest potential comes out as we move from today’s mobile phones and their attendant apps to the next generation of wearable devices and the internet of things.

“Smartphones are only a transitional point,” Rowe says. “From there we go to smartwatches and Google Glass and other ways of interacting with data around you that don’t involve a screen. And to get from there to there to there you need to package up web data and make it consumable in all these different contexts. We’re trying to position Kimono to be the framework for that conversion.”

“When the killer apps finally start coming out for things like smartwatches and glasses, they’re not going to be made by the companies that have the most interesting data,” he continues. “They’re going to come from the developers and designers who are thinking about it a little bit differently.”

If we suspend our doubts for a moment and peer into this crystal ball, Kimono starts to look like something very big indeed. In the scenario Rowe lays out, it takes root as a sort of connective tissue for an entirely new class of interactions and experiences–something like a nervous system for the internet of things. You could imagine pointing Kimono at not just websites but other sorts of streams, making objects react to sound, say, or building applications that respond to live video feeds.

At that point, you’re well beyond the esoteric realm of web scraping. “The ability to turn a website into an API is a very powerful thing,” Rowe says. “Being able to turn anything into an API is epically powerful.”

via：快鲤鱼