Article Review: Knowledge Discovery through Text Analytics

The article “IT innovation adoption by enterprises: Knowledge Discovery through text analytics” was select as an article in the field of study. The path to get to this article is articulated below:

Science Direct (Elsevier) > Search Term: “Text Analytics”

This article was chosen due to my interest in text analytics of huge data sets, to help derive knowledge from this unstructured data set. This was a primary topic/interest for the dissertation, along with identifying another unconventional way to do literature reviews.


This study investigated two premises: (1) Is it possible to use text data mining techniques to conduct a more thorough and efficient literature review on any subject matter? (2) What are the drivers of IT Innovations? After having identified 472 quality articles spanning multiple fields in business administration and 30 years of knowledge.  The authors used a tool called Northernlight (, where they were able to answer both premises. 

The authors, state that current methods of most literature reviews are time-consuming, usually focus in the last five years and involve tons of attention. Most articles are scanned by title and abstracts before the researcher considers them to be read in their entirety. This method, as argued, is not useful. Thus, effective techniques consist of the use of “meaning extract” of a large set of documents (usually considered unstructured data sets) across various domains should help the researcher to obtain and discover knowledge efficiently. Therefore, the first premise deals with the utilization of text data mining techniques. These techniques shouldn’t just merely revolve around a core system of counting the identified keyphrases (or “themes”), but on automating “meaning extraction.” “Meaning extraction” measures the strength between keywords or phrases that are related to others. The end-user/researcher can apply rules to help enhance meaning extraction between sets of keywords. The authors conclude that these techniques are an excellent way to do a first-pass analysis. The first pass analysis can help generate more questions, which can lead to more future insights.

They prove the first premise by applying the Northernlight system towards IT innovation. The authors then used 472 data sets, in which IT innovation is mentioned in multiple disciplines across the field of business administration. By setting rules to identify keyword proximity to other keywords (or their equivalents) they were able to garner some insights into IT Innovation. Proximity could be measured as far as a sentence (~40 words) or a paragraph (~150 words). Thus, they determined that cost and complexity are the two most frequent IT innovation determinants (as well as complexity, compatibility, and relative advantage) based on an IT department’s perspective. However, on the enterprise level, perceived benefits, perceived usefulness, and ease of use, were determinants of IT innovation. Finally, organization size and top management support positively correlated with IT innovation with cost being negative towards IT innovation.


The obstacle that came in here was that some articles with really creative titles and were recently published came at a price. So the article that was chosen was still a good read, but one does wonder how good those papers are that have been priced/paywall. Is having a paywall from publicly funded sources be hidden behind a paywall. We paid for it through our taxes, why should we have to pay for it once the results are out.