A mention is the base unit of analysis within Sprinklr Listening. One mention is equivalent to a single post or comment from a data source.

Sprinklr’s Product Insights models identify insightful phrases or sentences within each mention, which are then mapped to relevant categories (L1/ L2/ L3) and sentiment.

For example, the AI model identifies 3 insights in the mention “I love the performance, but it seems a bit pricey. However, will definitely recommend it.”

This process can be explained in detail using the example of the below dummy message:

Step 1: Message Tokenization

The message is first split into its constituent sentences

1. If you’re looking for a long lasting product, go for Airwick Freshmatic as its fragrance lasts longer even though it’s a bit expensive.

2. Febreze Air Effects is usually never available and also costs more than the Freshmatic.

Step 2: Entity Identification

Next, the message is processed to identify entities (brands / products), cohesive phrases, and parts of speech.

For example, in the first sentence, "If you’re looking for a long lasting product, go for Airwick Freshmatic as its fragrance lasts longer even though it’s a bit expensive", the Brand is Airwick, the Product is Freshmatic, the insightful phrases are "go for it" and “fragrance lasts longer”.