Subscribe to DSC Newsletter

Making the Business Case for Text Analytics

The key to making a business case for any Analytics initiative, not just text analytics, is to  identify specific business problems and pain points and use analytics to address them, instead of merely seeking insights.

The top business case for Text Analytics

Companies find themselves in a world where an increasing number of their customers are using social media, and the one thing, people LOVE doing on social media is talk (tweet/post/blog/whatever...) They talk about their experiences in dealing with the company and its services or products, about its competitors and about how they really feel. (Consumer feedback online: How you can spot it, sort it, and react quickly, using text analytics.)

So, as a company, you have all this customer feedback out there, in the form of text, just waiting to be gathered. The risk company management faces, for not capturing this customer feedback, is just too great to ignore. They face the risk of looking bad (think PR nightmare) and losing their competitive advantage, if they do nothing about it, which brings me to the single most important use case driving text analytics in the enterprise today, which is,  the compelling need for Social Media engagement and analytics. While some may argue that, this is too narrow a focus for the application of Text Analytics and while other use cases for text analytics may have greater ROI potential, analyzing unstructured text for social media, is often the first and most appropriate use case for companies to begin with and demonstrate ROI, before moving to other use cases.

Meta S. Brown, has an interesting viewpoint on building a business case for text analytics and the pros and cons of taking a cost saving vs. revenue increasing benefit approach, which can be found here.

Key Trends in Text Analytics

There are some other major trends that stand out in Text Analytics, from this 2013 report by Hurwitz and Associates  which are still relevant today and are useful to know when making the business case. I have listed below what I felt to be most relevant ones:

Text Analytics is moving beyond sentiment analysis

Sentiment analysis is evolving, with vendors, offering sophisticated sentiment analysis on multiple scales, rather than simply classifying a document, or a phrase, as positive, negative, or neutral. Text analytics is being used to identify "emerging issues" or the "birth of a trend". This helps companies to be more proactive in dealing with issues before they become real "issues" that could have been "nipped in the bud".

Marrying structured and unstructured data is becoming more popular in text analysis

End users have begun to realize the value of analyzing unstructured text data in conjunction with structured data (e.g. sales and demographics data) which can be used to provide a lift to predictive models.

The cloud is becoming increasingly important as a delivery model for text analytics

Companies that are lacking in in-house skills to aggregate and analyze unstructured information are turning to Software as a Service (SaaS) solutions for help.

Views: 4231

Comment

You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Dan Sullivan on February 9, 2015 at 9:17am

Scott, it's hard to describe all elements of a text mining architecture but basic building blocks could include: R and the twitteR and tm packages.  Python and NLTK are also possibilities but I've had some performance issues with NLTK on large data sets. See http://text-processing.com/demo/sentiment/ for some pointers and tips.  For large volumes of text, I like to work with Spark. It has some basic feature extraction functions in the MLib package. See https://spark.apache.org/docs/1.1.0/mllib-feature-extraction.html for more. Also, you might want to look into word2vec https://code.google.com/p/word2vec/.  I'm using it for text mining biomedical literature but I'm not sure how well it would work with Twitter.

Good luck!

Comment by Sione Palu on February 4, 2015 at 10:19am

Here's a company from Auckland , New Zealand that develop cutting edge text analytics for the enterprise, like one of the products is taking a corpus of unstructured text and turning that into an ontology. Its automated. This company linked closely with local New Zealand universities for co-funding of R&D.

http://pingar.com/

Comment by Mark Sharma on February 4, 2015 at 9:45am

Thanks for the feedback Scott. There is a definite need for a next level down practitioner view of text analytics deployments.In a nutshell, what I see with most of the clients that I work with (typically larger Fortune 50 companies) is vendor supported Social Media/Text Analytics deployments (e.g. Salesforce CRM + Radian6)  within a single department. There are Big Data (usually Hadoop) infrastructure build outs that are happening in parallel but with little or no engagement from the inhouse DW, BI and Analytics units (typically SAS). I will post a blog entry that talks more about this. Thanks again for suggesting the topic.

Comment by Scott Leandro on February 4, 2015 at 7:47am

Manoj - Good overview on the importance of text analytics.  It would be extremely useful if someone would discuss the next level down for practitioners, namely the types of technical architectures that would support it.  Any thoughts?

Follow Us

Videos

  • Add Videos
  • View All

Resources

© 2017   Data Science Central   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service