Safron API Documentation
Safron is a dynamic API that combines keyword analytics and targeted search capabilities for tech content. Safron is designed to help users discover and analyze trending tech topics by extracting keywords and linking them to their original content sources.
Safron is experimental and thus free, but we limit access for public use.
How does it work?
With Safron you can understand what people are sharing and talking about, and with what sentiment, by seeing the amount of keywords that are trending, within various categories. Instead of looking at only keyword statistics, a key feature is that you can track down the exact mentions and links where we found the keyword to read about what was said.
This allows you to do keyword tracking, build newsbots or do product and social research.
Don't just take our word for it though » try it out
Time Periods
Safron uses time periods to calculate trending statistics. It is by no means dependent on it but it is a user friendly way to build logic so it is easy to work with. When you query the keyword and graphing endpoints, you set the period to daily, weekly, monthly, quarterly or yearly. Data is updated around 6 AM to 6.30 AM UTC.
Safron does not offer real-time data, as it primarily works as a search engine. It is a way to scout information and engagement after something has been discussed. Expect to look at data the day after, to understand what has been said as a whole.
Our Models
We use advanced natural language processing to process and extract data. Under the hood Safron is at the moment using three smaller transformer models but we are working on a fourth. It is a way for the system to extract keyword data, categorize the keywords and then analyze sentiment. We have opened sourced a first version of the keyword extractor that you can find here.
API Limits
We want to keep a public endpoint for free use but to keep it from being abused we have certain limits on use.
Throttling and burst limits are set at 100 queries per second for the entire API. The public endpoints may also be lagging from time to time as it is storing data in a shared cluster. The public endpoints has access to daily, weekly and monthly reports and some access to quarter reports, although it may be less performant. The public endpoints do not have access to half-year and yearly reports as the shared clusters are unable to processes this amount of data at once. Both endpoints also has rate limits per IP. Make sure you disconnect any VPN before making a call as it will block it otherwise.
If you want to use a private endpoint that uses a more robust data store please get in touch to get an API key.
At the moment we are not forcing users to use authentication to query the API. This may change in the future if the endpoints are abused. If you are using the API, do sign up for notifications so you will get information in time if we do a switch.
Resources
Keywords
Learn how to work with the keywords endpoint to get top and trending keywords for a period.
Graph
Learn how to query the graph endpoint to get statistics period to period to graph its progression.
Connected Keywords
Learn how to use the connected endpoint to get the top and trending keywords connected to other keywords.
Sources
Learn how to get back sources for the keyword statistics along with doing isolated research in the database.