Skip to main content
Searching in Standard searches all Standard records on the hard drive. Since third-party services or the system itself require your permission or action to commit changes to records, all records in the Standard System belong to you and should be universally searchable.

​
Articles

Articles (article) are a standard table of records in the Standard system that contains all Wikipedia articles, all verifiable Common Crawl web pages, and points of interest. When searches are executed, Standard will always try searching the article standard if context allows. Searching “when is my next meeting” would not trigger an article standard search because articles will not contain anything that would concern your next meeting. If you search “how to make chicken noodle soup,” then Standard will search for a recipe to make soup from the article standard. Generic search is when your query and the answer to it are straightforward. 2 + 2 will always be four. Ford Motor Company will always be founded on June 16, 1903. These are answers that can quickly be deduced from the onboard data set using simple machine learning and natural language processing (no massive ML necessary). Detailed search is when your query can have multiple answers (results). This may be if you are looking for images of things or suggestions for places to eat. These searches may require a little more time to gather additional information, but the information is already on your device so Standard does not have to scour the internet.

​
Content Search

Content search can be generic or detailed searches but requires Standard to look into additional standards. When queries are submitted to Standard, an initial context sensing stage is executed to determine the type of query and if engagement with the language model is required to fulfill the request. Standard can use garden-variety NLP algorithms to answer “When is my next meeting.” If you ask “When is my next meeting and is there anything I need to prepare for it?”, then Standard will trigger a language model request (Standard Intelligence) and generate an answer. Content searches are searches outside the article standard that do not require the LM to produce a result. This search type could take longer than generic and detailed searching, but will be faster than engaging with an LM.

​
Intelligent Searches

Typically, if your query prompts a follow-up action or has more than one question, Standard will engage with the on-board language model to fulfill your request. This type of searching will not always be triggered. Keep it simple and stupid.