Combine NLP+ML+DevOps for Data Filtering Pipelines

[josh]

Examples:

Building databases/web pages listing all local calendar events of interest sorted by type, indoor/outdoor, fees, age range, level of interest, etc. Involves searching many dynamic sources & filtering the interpretations.

Sets of news/data relevant to predicting prices of some commodity

Dynamic advertising appeal of various celebrity icons

[josh]

A given pipeline might be in a stage of development where production use insists that everything pass through human experts. That doesn’t mean there is no payoff until AI beats the experts in a demo. The current payoff might be all the filtering stages that make sure the human experts, with limited time & focus get to look at the most promising cases/data points. And then further analytics might help to double check or catch their occasional mistakes. There are so many ways that these sort of processes can gradually improve over time. Automation of compatible formats can help a lot. Putting smart NLP with sophisticated format cleaning on the front end can help a lot

[Author]

Posts