Machine Learning
Sentiment Analysis and Machine Learning
May 11, 2017 Erik Peterson

Sentiment analysis has traditionally been used to quantify quantitative, semantic information for assessment and benchmarking. By reducing verbose, semi-structured or free-form natural language information to binary data - whether positive or negative - customer feedback can be objectively and accurately measured and compared.

Sentiment analysis has traditionally been employed by enterprises to benchmark “how they’re doing” in a given market or with a given audience. It frequently takes the form of media analyses, counting positive and negative coverage over one or several time frames. Because of its labor-intensiveness, such information has usually been often beyond the reach of small to medium sized organizations, who instead default to other secondary metrics as a gauge of their marketing and public relations performance.

But thanks to social media streams and rapid evolution in data mining and machine learning, sophisticated natural language analytical tools are within reach of even modest businesses. Unstructured semantic information in multiple languages can be parsed by computers using a variety of tools, including open source software such as Python’s Natural Language Toolkit, Stanford’s Core NLP Suite or Apache Open NLP. Instead of paying providers to employ humans to read and evaluate written content and customer feedback, businesses use such solutions to process feedback contained website comments, emails and social media streams.

Better yet, sentiment analysis can be applied beyond the traditional top-level binary yay-or-nay evaluation of a business or brand to specific products or features within that business or brand. Provided that there’s sufficient data/feedback for analysis, machine learning algorithms can be deployed in tailored, specific fashion. Further still, sentiment analysis can be used to combine disparate types of information. For example, semantic information can be correlated with geolocation data, to rapidly visualize patterns and trends. Not just businesses, but disaster preparedness and first responders can benefit from these methods and technologies.

If previous generations of technology were limited by the tools available, machine learning, IoT, and sentiment analysis are limited only by the ingenuity and imagination of its users. The proliferation of digital information and processing ability has delivered unprecedented ability for businesses to understand their audiences and environment.

Erik Peterson
DevOps & Cloud Infrastructure