Filtering and analyzing unstructured business data has enormous potential to provide a competitive advantage to organizations. The key to realizing those benefits is actually being able to work with the data to reveal those insights whether they are from private or consumer data. Because while there are probably true valuable nuggets of actionable insights, there are also references to the company picnics littered throughout your data.
Filtering and analyzing unstructured business data has enormous potential to provide a competitive advantage to organizations. The key to realizing those benefits is actually being able to work with the data to reveal those insights whether they are from private or consumer data. Because while there are probably true valuable nuggets of actionable insights, there are also references to the company picnics littered throughout your data.
Imagine a collection of customer service chats of customers expressing a desire to switche services, based on a variety of attributes like level of customer service, broadband speed or modem/router quality, competitor deals, etc. Using a powerful language modeling technology, you can more accurately organize text based on how consumers are talking about a category, brand or product. The resulting information can give your organization a more concrete idea of your customer’s value perspective; what they think is important.
The image below displays the volume (# of conversations) associated with the following dimensions. In other words, how often customers are expressing an opinion or intention around:
- Advertising
- Affinity
- Customer Service
- Intent to Switch
- Problem
Click image to enlarge
Now that you have content isolated for each of dimension, you can begin to drill down by data to extract the actual text for that time frame.
Click image to enlarge
Language Modeling Shows Context at Its Best
You cannot get to this point in your analysis if you are relying on brittle or cumbersome technology that requires a lot of manual tuning or that doesn’t shift as the context or language changes. Using semantic filters provides a more advanced form of language modeling that deciphers the context of the language used – the meaning, not just what terms are present – and matches semantically similar content. This means that more robust and accurate categorization of topics is possible.
Why does content categorization matter?
Organizing on-topic content into categories that match key performance indicators can help optimize your analysis to track important business metrics. Adding sentiment and dimension analysis elevates your analysis to a whole different level. But its the initial effort to define what your organization is wanting to track and analyze and then mapping the resulting categorization to business metrics that will make the best use of your organization’s data resources.
Click image to enlarge
