Post-Text Mining and Text and Numerical/ Structured Information Integration

At K-Praxis we work with what we usually dub as "Post-Text Mining" technologies and methodologies that allow us to process information from various textual sources, including the Internet, and to arrive at insights and analyses. Some of these insights and analyses are reflected in our articles here. As we delve deeper into the world of text and unstructured information or unstructured data - even though in only one language (English) - we realize time and again that textual information is often integrated with what the world calls structured information or numerical data.

Whenever we think of context today we usually think of either searching of information on the Internet, which is mostly textual information. For search engines and for unstructured information management companies (if they are savvy enough to understand the value of context vis-a-vis brute force of computing) context is the Holy Grail that becomes more and more complicated and unachievable the closer one gets to it.

But a text itself never is "pure" in the sense that there is a great deal of information in the text that is not textual, and anybody who is processing textual data will need to understand the numerical/structured side as well.

Business Intelligence Solutions and Context

Extending the earlier argument, one could easily imagine that although computing and technology have taken great strides in numerical and structured data analysis we think that unless and until business intelligence systems are able to work with these two types of information seamlessly, they would never be able to make sense, never would be able to understand and gauge the context or provide predictive analysis. Providing predictive analysis either purely from numbers or from text will be living in oblivion. Any body who has worked with an analytical data mining system, or a BI solution, would tell how much of human intervention and preparation takes place before the machines begin show you some results.

We pose a few questions here:

1. What type of context understanding and context creation is available in BI systems?

2. Can BI systems work with numerical and preformatted data without taking any reference from textual data?

3. Can text-processing systems work without understanding structured and numerical data, or is it safe to assume that because most unstructured data management systems borrow heavily from data mining they inherently understand numbers/ preformatted data side of the game?

These questions are a beginning as we develop more context based text analysis systems and we already are seeing the benefits of fusing numbers and text.

Please write to us with comments questions and suggestions our email is: info@k-praxis.com