Information Technology Reference
In-Depth Information
APIs, etc. on the web. The exponential growth in the amount of field data means
that innovative measures are needed for data management, analysis, and accessi-
bility. Therefore, almost users' needs are changing. Online databases have become
crucial to access and publish various data depending on the user's purpose, task, or
interest.
First, for leading the solutions, we have to marshal what Big Data is. We con-
sider that the Big Data includes two prominent types of direction for ICT research
purposes.
The one is the scale and the speed issue of data processing. A lot of researchers
have done this theme such as HPC, parallel distributed processing, and etc. Another
is a schemaless data processing issue. It is important to real-timely discover the
answers or clues for a user. A system has to create an appropriate schema from the
data themselves given by a user's query. Until now, data organized on the database
schema. Currently, there are only fragmental various data on the web. This is a
big paradigm shift. That is, it is necessary to create the schema and data structures
corresponding to user's required processing after a user inputs some queries. We
have to shift the system from closed assumption to opened assumption. In the case
of the closed assumption, the schema was designed in advance in consideration of
orthogonal or independence. In the case of the opened assumption, we cannot care
orthogonal and independence when the system dynamically creates schema.
On this background, it is important to research new data analytics method. Es-
pecially, it is necessary to discover the relation between each thing, event and phe-
nomenon. By this discovery, we can do risk aversion, can pursue a cause, or can
predict the phenomenon which may happen in the near future. On the other view-
point, we consider that the essence of Big data is not only massive data processing,
but also optimization of the real world by the knowledge acquired from aggregated
data. The current tendency of the research on Big data is how to aggregate massive
data and how to process these data quickly. The future tendency of researches will
become discovery methods of the optimized solution from the Big data. It is impor-
tant to realize risk aversion, the cause unfolding, and the phenomenon prediction in
the near future. In the first step, it is necessary to extract relationships between each
heterogeneous thing, events and phenomena.
We propose a new concept, anteroposterior correlation. The anteroposterior cor-
relation means the correlation based on the time anteroposterior relation. The antero-
posterior correlations are represented in relative comparison with each conditional
probability distribution. The one of the most important points is relatively compara-
ble. In the other words, this system discovers the relation with higher correlation by
comparing each value of a conditional probability as each weight. In this paper, we
represent a new discover method of anteroposterior correlation between each het-
erogeneous thing, events and phenomena. We discover a correlation in consideration
of the continuity of time. By our method, we effectively discover relationship be-
tween heterogeneous things, events and phenomena. The one of the features of our
method is a measurement correlation by using conditional probability. We calculate
the anteroposterior correlation relative by representing all in conditional probability.
Search WWH ::
Custom Search