inca’s conversational AI makes the discussion flow.
While other chatbots answer, inca asks. With its deep Natural Language Processing (NLP) capabilities, inca mimics in-person one-on-one interviews and the evolving nature of qualitative research. Our architecture leverages powerful multilingual, multipurpose, and adaptive machine learning techniques to create meaningful and engaging conversational experiences.
Here are some of our custom-trained NLP models that make inca a virtual moderator.
Topic Detection
Our algorithm automatically categorizes responses into predefined categories that are meaningful to your research. Now available for ad and concept test. More pre-trained topics coming soon.
Non-Informative Answer Detection - Elaboration
What are you supposed to do when 10 open-ended questions return 10 ‘meh’s’? inca makes sure you never have to deal with that, by automatically detecting non-informative answers and asking for elaboration, like a real conversation.
Emotion Detection and Mapping
Emotions go beyond just positive and negative; they are rich, subtle, and powerful. We understand that how we feel impacts what we remember and use algorithms to detect emotional cues.
Concept Matching
A human can understand that "song" and "music" are the same concept, and now, so can computers. inca has a unique concept matching algorithm that matches verbatim with concepts. Think of it as a more powerful keyword search, but using concepts.
Opinion Chunking Model
“I like the product, but it’s too expensive” are two separate opinions, and should be treated as such. Our algorithm splits different opinions into chunks to help analysts analyze what's being said more easily.
Keyword/Phrase Extraction
Our model goes through all the responses and extracts key information. Unlike other existing models that only pick up nouns, our model picks up multi-word expressions, so things like “the new trendy survey platform“ and “happily programming“ would be picked up.
Text Sentiment Analysis
Trained using state-of-the-art machine learning on 7 billion sentences, inca’s sentiment analysis model accurately picks up on tonal nuances and identify positive, negative, or neutral responses.
Gibberish Detection Model
Wouldn't it be great to not see "asdfjb" in your responses? As a part of our quality enhancement strategy, inca uses an algorithm to nudge gibberish typers to give meaningful answers.
and more...
Our technology doesn't stop here. We are always developing something new and innovative. Why not connect with us and see what we're up to?