Managing Unstructured Data in Just Nine Steps
February 25, 2015
I’m sure Datamation will help some companies with its post, “Big Data: 9 Steps to Extract Insight from Unstructured Data.” However, we think these steps may make other companies look for easier options. Writer Salil Godika explains why he feels these steps are worth the effort:
“Organizations have to study both structured and unstructured data to arrive at meaningful business decisions…. Not only do they have to analyze information provided by consumers and other organizations, information collected from devices must be scrutinized. This must be done not only to ensure that the organization is on top of any network security threats, but to also ensure the proper functioning of embedded devices.
“While sifting through vast amounts of information can look like a lot of work, there are rewards. By reading large, disparate sets of unstructured data, one can identify connections from unrelated data sources and find patterns. What makes this method of analysis extremely effective is that it enables the discovery of trends; traditional methods only work with what is already quantifiable, while looking through unstructured data can cause revelations.”
The nine steps presented in the article begin at the beginning (“make sense of the disparate data sources”) and ends at the logical destination (“obtain insight from the analysis and visualize it”.) See the article for the steps in between and their descriptions. A few highlights include designating the technology stack for data processing and storage, creating a “term frequency matrix” to understand word patterns and flow, and performing an ontology evaluation.
Writer Salil Godika concludes with a reminder that new types of information call for new approaches, including revised skillsets for data scientists. The ability to easily blend and analyze information from disparate sources in a variety of formats remains the ultimate data-analysis goal.
Cynthia Murrell, February 25, 2015
Sponsored by ArnoldIT.com, developer of Augmentext
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
