Explore Data

HumanFirst allows you to centralize and easily explore your unstructured data using powerful search and navigation features made for conversational data.

Features#

Full-text search#

Full-text search is often a starting point for exploration of data, if you have particular keyword(s) you're looking for. Semantic similarity can then be used to further explore results.

Semantic similarity search#

This feature finds semantically similar utterances to the ones you've selected, from your entire data corpus, in real-time. It's great for quickly surfacing related questions based on semantic meaning (and not keywords).

Single and multi-turn dialogs#

Studio allows you to explore your data as a flat lists of utterances, as well as navigate full multi-turn dialog exchanges.

Rank by entropy (advanced)#

You can sort of unlabeled utterances by decreasing entropy; this helps discover utterances that have a low confidence of belonging to any existing intent in a workspace, hence helping with the discovery of utterances that merit a new intent to be created.

Clustering#

Clustering helps you find groups of utterances that are very likely to belong together. You can combine clustering with the other search, filter & sort features to obtain exactly what you need. Clustering can also be initiated from an existing intent, this would yield unlabeled data clusters matching that intent. More on clustering

Video Demo#