Debunking Natural Language Processing: Detecting Events

By: Ellen Loeshelle

January 22, 2019

Tags:
Clarabridge Analytics
Text Analytics
Artificial Intelligence
Machine Learning
Natural Language Processing
Natural Language Understanding
Speech Analytics

For many people, September 19 is a remarrrrrkable occasion for which pirate memes and puns abound. National Talk Like a Pirate Day, invented in 1995 as an inside joke between two friends after a contentious racquetball game, has become an internationally acclaimed parodic sensation. (September 19 also happens to be my birthday.) Events bookmark our perception of time and give us a shared cultural experience. Specific events also help keep certain businesses afloat. Many retailers make more money between Black Friday and Christmas than they do the entire rest of the year.

In the previous blog posts, we discussed some of the ways in which Clarabridge uses Named Entity Recognition to find meaning within customer feedback. Many customer experiences are dictated not just by specific products, brands, companies, etc. but by the events for which those items were purchased. Events, of course, make up a broad category that includes national observances, religious holidays, cultural events, seasons and life milestones, to name a few. Some events take place on specific days, some span weeks or months, and each comes with its own traditions and customs

For those in the customer experience space, understanding consumer behaviors in relation to specific events can be extremely valuable information as it can power development, marketing and commercial decisions. In addition to detecting the named entities described in previous sections, Clarabridge automatically tags events of all kinds. Our dictionary of events is geographically diverse and includes entries for all of the event sub-types mentioned in the previous paragraph, including Diwali, Halloween, Chinese New Year, Easter, Shabbat, The Olympics, The Oscars, Black History Month, back-to-school, weddings, baby showers, graduations, funerals and more. It even includes emojis! For events that have a dedicated emoji such as or , Clarabridge detects this emoji as an event and maps it to its standard text form for a more complete understanding of customer feedback.

How are these event attributes best leveraged? Here are popular use cases for this feature:

Seasonal Trends
Track mentions of events in conjunction with discussions of sales and promotions (Independence Day Sale, Black Friday, Cyber Monday) to determine which ones are generating buzz. Filter down these mentions to those discussing availability to determine if demand is outpacing supply at critical times of the year.

Gift-Giving Opportunities
Determine for which occasions products are being purchased by individuals for themselves or as gifts for others. Many products and brands are favored for certain milestones. Mentions of weddings, engagements, baby showers, graduations and so on may help highlight how best to market and price items to target specific buyers.

Product Uses
Discover how products are being used by analyzing them in conjunction with mentions of cultural events or occasions. Understanding that a refrigerator is used for Shabbat or an oven for a Christmas dinner can shape the features that the product should offer and the marketing strategies that are used to get it to the most ideal consumers.

These examples have led to valuable insights for our customers, and we’re excited to see how they continue to discover more uses for this feature.

detecting events