WolframKnowledge
Combining Data, Context and Algorithms to Create Knowledge
The continually growing Wolfram Knowledgebase is the world's most advanced repository of reliable, contextualized and factual data, tightly integrated with tens of thousands of algorithms.
Isolated facts, without the ability to understand, compare and transform them, are like answers to trivia questions—interesting, but not powerful. Wolfram|Knowledge is instead an integrated, coherently connected web of data and algorithms that turns information into understanding, facts into knowledge and data into meaningful conclusions.
Data
Wolfram|Knowledge contains trillions of individual data points drawn from a vast range of public and proprietary data sources, authoritative scientific reference databases and independent research performed by in-house developers and data curators. The knowledgebase spans hundreds of top-level domains, and is structured to maximize the ability to perform complex computations and analyses within and across domains.
The Wolfram|Knowledge team continuously gathers and updates the data with a combination of custom automation tools and human curation. AI systems scrape and analyze texts and data tables, but humans check and crosscheck using sophisticated statistical analyses to look for anomalies and suspicious outliers that are always present even in solid data sources, let alone the internet at large.
Context
All data in Wolfram|Knowledge is semantically contextualized. For example, all numerical quantities are tagged with their units in a consistent form across the entire knowledgebase, so information about food Calories can be reliably compared with exercise Calories, and with reaction energies expressed in lowercase calories (which is a completely different unit). This eliminates the all-too-common mistakes that happen when interpreting numerical tables from disparate sources.
Similarly, despite the complexity of historical calendar systems, significant historical dates are compatible with the dates and times used in astronomical data, so the birthdate of an ancient Chinese scholar can be compared to the calculated date of an eclipse. Decades of work have gone into developing the underlying semantic engine—of surprising depth and complexity—necessary to make this kind of cross-database consistency possible, not just for units and time, but for all categories of data.
Algorithms
Data is nothing if you can't operate on it. Wolfram|Knowledge incorporates the world's largest collection of algorithms and transformation rules, all fully integrated with our semantic data. This allows the system to combine multiple data points from multiple sources to reach reliable conclusions.
For example, how many tanker trucks of milk per day are needed to supply 10% of the calorie requirements for all schoolchildren in Los Angeles? Given the dimensions of the tank, Wolfram|Knowledge can calculate the volume, look up the density of milk, calculate the weight, look up the calorie content of milk, look up the school enrollment in Los Angeles (city, county or school district) and divide by the calorie requirement per school-age child. All the data and each formula used in this process came from a different source, but all work together seamlessly in the Wolfram|Knowledge system to arrive at the answer: 6.5 trucks per day. Because the computation is fully automated, the system can effortlessly provide a graph showing that, based on declining school population, the requirement has declined from 6.8 trucks/day in 2009.
Accessing Wolfram|Knowledge
Wolfram|Knowledge can be explored through the public website wolframalpha.com. Type in whatever you're curious about, or start from the examples by topic.
API access is also available, supporting both natural language queries and specific data or computation requests that bypass the natural language parsing stage for greater speed.
For wholesale customers, the data portion of Wolfram|Knowledge, in whole or in targeted subsets, can be exported and licensed in standard or custom formats (JSON, CSV, custom XML, extended triples, etc.). API access is similarly available for licensing through Wolfram servers, AWS or Azure, or customer-hosted solutions.
For research and prototyping purposes, Wolfram|Knowledge is available as an integrated feature of Wolfram Language. Trial cloud APIs can be created and instantly deployed in the Wolfram Cloud through simple function calls.
Wolfram|Knowledge is the result of over 30 years of work developing languages, algorithms and data collection techniques by world-class teams at Wolfram Research, Inc. Groups of subject-matter experts work together with data specialists and algorithm developers to create the most efficient curation tools possible, allowing them to locate, ingest and quality check vast amounts of data rapidly and efficiently.