Big data analysis will uncover MH370 's missing answer

Ma Airlines MH370 passenger plane missing mystery if by large data analysis uncovered, I do not feel an accident. The person is doing, the day is looking, the deception is impossible, just need to discover the truth details and clues.

Today, the US satellite operator DigitalGlobe (Digital global) launched the Tomnod crowdsourcing platform (crowdsourcing), released the suspected area of satellite images for volunteers to "carpet" search. Similar to human flesh search, which is also a large data application, I think perhaps the truth is hidden behind the satellite image data.

Human flesh search is certainly a forced way. In fact, the public security cases are also dependent on this approach, through the manual to troubleshoot the local surveillance video. Boston Marathon bombing is a successful case, through video screening, found the suspect's clues, and finally successfully solved the crime. What the Boston police used, big data, or human flesh search, is unknown. But there is a point, if can rely on the system, then the efficiency of the case will be increased, relying on human flesh, unless it is a war, human and material resources will be limited, the efficiency is difficult to protect.

It is said that it is not difficult to use the machine to do the data, the core is modeling, the other things can be handed over to the machine. The difficulty of large data application is also modeling. Taking Google as an example, according to the characteristics of the search, through modeling, to build a Google search business model, has been successful. It is very important for industry enterprises to make large data application and successful modeling.

For MH370, if you have a successful modeling, you will definitely create more powerful clues for event tracking. The data are not necessarily limited to the provision of satellite data in DigitalGlobe. Everything has a connection, everything will have a trace, when these scattered around the data gathered together, some things of the essence, the truth will emerge. DigitalGlobe used the "carpet-type" search, I think is also trapped in modeling, there is no reasonable analysis of the data, but also had to use tactics.

Public security cases are often used tactics, the video surveillance data for each card is manually checked. If through modelling, can automatically draw the suspect vehicle, the suspect's action track, this is the gospel for the police handling case. For video information, may involve the image recognition technology, in fact, technology is not difficult, modeling is difficult. After the modeling is completed, how to analyze and retrieve the data through Map/reduce, and finally form the result, which is also a problem to be considered.

For Internet enterprises, due to the accumulation of a large number of software personnel, their own sufficient software development capabilities, can be based on their own business needs, modeling, map/reduce their own business model. But for MH370, for public security, this is not a simple matter. In my opinion, MH370 more load, the vehicle has a license plate, and Mh370 even a clear logo is not, so in a haystack.

Because of the difficult, big data to MH370 is timely. Through the MH370 event, hopefully there will be a model of large data analysis, and if so, it is the Gospel of mankind.　　Believe in big data, believe in the power of big data. Through the practical application of large data, the graphic search technology can bring real value to the Enterprise--neo Marvell Company software engineer Max De Marzi.

The connections between different things can reveal the insights and patterns behind them by finding them. The work of Max De Marzi is to build a concept validation (POC) model for customers using NEO4J NoSQL database and cypher query language. In other words, he can show us how to use this new technology to solve specific business challenges.

A recent POC project by Max De Marzi was inspired by the Facebook Social search Tool graph search, which is equivalent to a search engine that provides users with specific results through a "natural language" query. At this year's Big Data Techcon conference, Max De Marzi will show us how to use large data to create the enterprise's own graph Search.

Let the big data be effective

The best way to get value from big data is for consumers, or more directly, retail companies. Max De Marzi said that most of the retail business sites he visited would simply ask consumers about their general points of interest. In his view this is not enough, and if retailers can dig deeper into the consumer's family, age, and other things of interest, they will be able to more accurately push the consumer to the goods they are interested in.

"It's about how you use the data, and you can use social media data, like microblogging data." You don't have to go to the consumer to collect this data yourself. There is already a lot of information available in social media, and all you need is a dynamic approach to integration and use. Max De Marzi said.

Although retail companies are best able to derive value from graph search, it is not that the business sector is not. Max De Marzi also cited examples of results, such as the fact that banks can detect connections through large data to avoid fraudulent behavior. If a credit card is used by 20 people, it is likely to be fraudulent.

Some of the more complex situations, such as gang fraud, are hidden behind misleading data. The traditional method of fraud detection is to address this information as a separate data point, ignoring the relationship between the data, such as the above 20 people with a single card. And under the guidance of large data analysis, banks can quickly find the existence of the strange.

Create your own graph search

"Whether it's a traditional DBA or a technician who is familiar with NoSQL, there will be a learning process." Max De Marzi said. First, developers need to learn how to understand natural languages in order to create new cypher queries. He thinks the difficulty in the transition is not at the technical level, but in the transformation of thinking. Developers need to change the traditional way of thinking and more accept the concepts of dynamic data points and interconnected objects.

"This shift in thinking may take only a few hours, or it may take days." But once the developer completes the process, everything becomes simple and easy to use. Max De Marzi said.

Fortunately, this is an era of information explosion, at present, both online and bookstore have a lot of relevant content for developers to get started.