Diffbot is an AI startup that provides knowledge as a service to power intelligent applications. As with most such services, its reach (by default) exceeds its grasp in terms of performance. We use it to clean up articles that are added to our system, as do a number of others. We found it slow -- indeed when I took it out, processing time improved by almost 50%. In this piece, I will share what the magic parameters I found are to accomplish this. Hopefully, you will also find it useful.
I stumbled upon a support article entitled Improving API Response Times, from which these parameters are sourced. Setting the discussion parameter to false will remove parsing of any comments or reviews. The second (and key) use case is the paging parameter. Setting this to false downloads only the first page of a multipage article.
A site like Fiskkit can not use the first page of a multipage article, but your use case may not require it. Enjoy, and if you'd like to hire me, I'm always available for contract work.
No comments:
Post a Comment