Talking about Alibaba Big Data: Data ecosystem

V) Data ecosystem

The DT era will catalyze the big data ecosystem. The data ecosystem of the DT era, I define it from the following two aspects:

1) Data exchange / trading market

The cornerstone of smart business is data. As the top priority of smart business, data is the most important.

Data as a means of production, the blood of the era of big data, like the gasoline of a car, without gasoline, and then the beautiful high-end cars can not operate. The source of data is often multi-faceted. In the future, the data used by a company is often not only its own data, but also the data exchanged, integrated, and purchased by multiple channels. For the big data business form of “wool out on pigs”, the data must be mobile, and the data can only be used for greater value if it is integrated.

But the data needs to be exchanged and traded. As I said above, we must finally solve a series of problems such as laws and regulations, data standards and so on.

2) Algorithm economy / ecosystem

According to Gartner's analysis, the algorithm will form a global trading market, just like the app of the year, which will spawn a new generation of professional technology start-ups and revolutionize the way machines and machines interact.

At the same time, more data will generate better models and user experiences, which in turn will attract more users and more data, which will result in a continuous reduction in the cost of storing and computing data.

Gartner has published a report assessing the market impact of the algorithm economy.

Gartner believes that, inevitably, the algorithm economy will create a whole new market. People can buy and sell various algorithms, bring together a large amount of extra income for the current company, and spawn a new generation of professional technology start-ups.

Imagine a market where billions of algorithms can be bought and sold. Each algorithm represents a piece of software code that solves one or more technical problems or creates a new opportunity from the exponential growth of the Internet of Things. .

Algorithms are the cornerstone of creating intelligent applications and are the core value of big data.

In other words, multiple machine learning algorithms can be combined to become more powerful algorithms to better analyze data and fully exploit the value in the data.

In the algorithm economy, cutting-edge technology projects, whether advanced intelligent assistants or drones that can automatically calculate inventory, will eventually be implemented into real code for people to trade and use.

The generalized algorithm exists in the entire closed loop of big data, and there are algorithms support from every level of big data platform, ETL (data acquisition, data cleaning, data desensitization, etc.), data processing, and data products. Algorithms can be traded directly, or packaged into products, tools, services, or even platforms to trade, ultimately forming an important part of the big data ecosystem.

Some even think that good algorithms can get rid of the excessive dependence of many companies on big data. Although data may be the most expensive production data in the DT era, big data is not necessary if the algorithm is powerful enough. For example, migration learning can make computers get rid of the heavy dependence on big data, so that artificial intelligence is no longer just a "rich game."

Just as the App economy has transformed the way humans interact with machines, we will see that the algorithm economy will drive the next leap in machine-to-machine interaction evolution.

People will evaluate its performance through the algorithm used by the product. The competitiveness of enterprises is not only about big data, but also algorithms that can transform data into practical applications. Therefore, CEOs should focus on the company's proprietary algorithms, not just big data.

Emerging machine intelligence platforms can host pre-trained machine learning models with a “model as a service” approach, making it easier for companies to turn on machine learning and quickly transform their applications from prototypes to products. As companies adopt the microservice development paradigm, the ability to access and use different machine learning models and services to deliver specific functionality will become increasingly valuable.

All of this, ultimately, can not be separated from cloud computing, the data platform is naturally based on cloud computing. For data exchange, algorithm trading requires a store, and the cloud is currently the best store. Whether it's data interoperability or cloud-based pre-training and hosted machine learning models, each company's data products will be able to leverage algorithm intelligence on a large scale.