Overview
What it is: Text Mining is a process wherein machine learning technology is used to read patterns in unstructured linguistic data like blogs, social posts, and help desk comments. Examples of this can include categorization, entity extraction, and sentiment analysis.
What it does: Text Mining allows companies to gather important information about their customers from data sources that would otherwise be impenetrably dense. Once trained on sample text mining algorithms break down unstructured communication by emotional tone and subject matter. This makes it possible to get a genuine sense of, for example, what customers feel about a new software update, or whether dealing with your customer support department is a pleasant experience, just by looking at everyday communications
Why it matters: A huge percentage of data is unstructured, and many predict up to 80 percent of data will be unstructured by 2025. This means that companies that don’t incorporate text mining tools (or other ways of incorporating unstructured data) will lose out on a primary source of valuable information.
What to do about it: Use text mining to take advantage of data sources that are currently untapped, or to speed up research that’s already taking place. Incorporate the resulting insights into targeted messaging, customer service, and wherever else they apply.
What Text Mining Services Are Available?
Although it’s possible to make fine distinctions between the feature sets of different applications that bill themselves as either “text mining” or “text analysis,” the overall concept is ultimately the same. There are many offerings on the market, and more will emerge. Amazon, Microsoft, and Google all offer text mining tools as part of their public cloud services. Dozens of stand-alone text mining applications are also available, and while most are commercial/proprietary, some are open source. Accordingly, text mining is something that an enterprise of any size can invest in, to some degree.
Business Advantages of Text Mining
- Text Mining of existing content can deliver results much faster than traditional surveys and other similar information-gathering practices.
- Text mining programs can be run all the time, meaning that they can provide a constant read on the emotions of your customer base, your employees, or anyone else whose unstructured text you have access to.
- Since reports are based on native communications rather than solicited survey data, they’re arguably more accurate.
Applications
- Risk management
- Knowledge management
- Cybercrime prevention
- Customer care service
- Fraud detection through claims investigation
- Contextual advertising
- Business intelligence
- Content enrichment
- Spam filtering
- Social media data analysis
Case Study:
An example of the power of text mining is its use in the US Social Security Administration. The SSA uses a custom-built text mining application developed by Elder Research to identify Social Security claims filed by high-risk applicants, so they can be automatically approved, freeing up time for more demanding cases that require human intervention.