NLP and Consumer Insights
Social media provides a medium for consumers to rant, rave, and recommend products, brands, and services. This digital voice of the customer provides brands, and the agencies serving them, insights into the when, where, how, and why products are used, who is buying them, who is using them, and information about those individuals including their associated beliefs, needs, wants, and preferences. These insights facilitate understanding of consumer behavior, which can allow brands to better build, design, and market their products and services to meet consumers’ needs and desires. Given that the majority of the consumer conversation is textual in nature (e.g. product reviews, tweets, Facebook posts, etc.), Natural Language Processing along with Machine Learning and, in more general, Artificial Intelligence are now critical components in gaining a 360 degree view of consumers.
So what is Natural Language Processing (NLP) and how does it help in discovering consumer insights? NLP is a field of A.I. that deals with decoding the meaning of language. If you have ever used sentiment analysis, keyword extraction, summarization tools, or the personal assistant on your smartphone, you have used an NLP application. There are a number of NLP methodologies and applications that are often used and extremely helpful in decoding and transforming the digital voice of the customer into actionable consumer insights.
- Aspect-based Sentiment Analysis: While traditional sentiment analysis can tell you if an entire comment or post is positive or negative, aspect-based sentiment analysis goes deeper by telling you what in the comment is positive and negative. More than that it determines what aspects, or attributes, of the product or service are important and the consumer’s opinion about those aspects. For example, in the sentence:
Scent would be identified as the aspect of the product being reviewed and be associated with a positive sentiment due to it being “light and not overwhelming.” This makes sentiment analysis actionable as we now know what the customer likes about the product and in turn an aspect that is important in the buying decision process.
Named Entity Recognition and Resolution: Identifies the real-world objects, e.g. people, locations, and organizations, mentioned in text and links those objects to canonical forms (i.e. knowledge bases like Wikipedia). For marketing oriented insights these systems can be trained using custom data to also recognize and link product and brand names. This is extremely useful for monitoring social media for brand or product mentions.
Keyword Extraction: Identifies significant words and phrases in a given document or collection of documents (e.g. user reviews, tweets, etc.). Methodologies exist for matching a known set of keywords (think ontology or taxonomy) or to discover the keywords from the data.
Topic Modelling: Discovers the topics or themes of conversation from a collection of documents. For example, our consumers talking about price, service, quality, etc? Topic modelling is useful in obtaining a high-level view of what consumers are discussing.
These methodologies represent some of the basic building blocks in most analytic systems used for generating consumer insights. In my next post, I will take a look at how NLP can be used to make the combination of Psychology and Linguistics (Psycholinguistics) a computational reality and provide information about the consumers behind the text. You can take a look at how we at Oculus360 used Psycholinguistics to provide insights on how eSports can drive Millennials and Gen Z to your brand.