NLP insights

Experiment from #buildTheNews hackathon .

Experiment with Alchemy API

While working on quickQuote I made a working proof of concept prototype using alchemi API see screenshot in attachment.

at the time it used spokendata to generate the transcription. Spoken data returned srt files, so interpolated the line timecodes to generate word accurate transcription representation.

the alchemy API for the NLP, to identify, entities, concepts , keywords see "Retirement of AlchemyAPI service" replaced by IBM discovery, and natural language understanding.

SMMRY API to generate the summary from the transcription.

Every concept, keyword, entity, and line of summary, was clickable and would take you back to that point in the video/transcription.

The transcription would also stay in sync/scroll with the video.

Last updated