| text
| -
Zuck:
With regards to the following statements:
"Before I could build any AI, I first needed to write code to connect these systems, which all speak different languages and protocols."
And:
"But even if I spent 1,000 more hours, I probably wouldn't be able to build a system that could learn completely new skills on its own -- unless I made some fundamental breakthrough in the state of AI along the way. "
I have the following comments:
Courtesy of RDF Language and terms from the Schema.org vocabulary [1], you can create abstraction that exposes Actions and their respective invocation signatures via structured entity descriptions.
In a nutshell, you can build Smart Agents that learn new skills by simply discovering and reading Action descriptions i.e., this whole affair becomes much more declarative in nature and the key development activity boils down describing ourselves, our smart agents, and various actions they are authorized to perform on our behalves.
Descriptions take the form of RDF Language sentences that are comprehensible to both humans and machines (software).
BTW -- what I describe is nearly completed
for Facebook Messenger (one of many planned invocation interfaces for interacting with our Bot that's 100% driven by structured data descriptions), so expect something for the public to play with in the new year :)
Entity You:
1. Identifier that identifies entity "You"
2. Profile Document that describes entity "You" (which would cover a wide variety of individual preferences including those associated with your appliances)
3. Specific relations in the profile document above that describe the nature of the relationship between you and various agents (bots etc..).
Your Smart Agent:
1. Identifier that identifies entity "Your Smart Agent"
2. Profile Document that describes "Your Smart Agent" (which would cover a wide variety of Smart Agent preferences including those associated with your appliances)
3. Specific relations in the Smart Agent profile document above that describe the nature of the relationship between Smart Agent and its Authorized Masters e.g., You).
Smart Agent Actions:
1. Identifier that identifies a specific Action
2. Action Profile Document that describes a specific Action (written by hand or generated in automated manner from API description docs using notations such as WSDL, Swagger, basic RESTful HTTP patterns, etc..)
3. Specific RDF Language statements in the Action Profile Document above that describe the invocation signature of a given action and nature of responses.
Note: all identifiers take the form of hyperlinks (i.e., HTTP URIs) that unambiguously identify their referents (You, Your Profile Document, Your Agent, Your Agent's Profile Document, Actions, Action Profile Docs, etc..).
Links:
[1] http://schema.org/Action -- About an Action
[2] https://medium.com/.../describing-web-services-using-a... -- Action Description embedded in a blog post (I could replicate this in a Facebook note via copy and paste)
[3] https://osdb.openlinksw.com/osdb -- Rough cut edition of docs (***This is very much a work in progress with the usual documentation placeholders, you need me to grant access via our Messenger App to experience what's already working e.g., Action invocation via Messenger's conversation interface **).
|
![[cxml]](/fct/images/cxml_doc.png)
![[csv]](/fct/images/csv_doc.png)
RDF
![[text]](/fct/images/ntriples_doc.png)
![[turtle]](/fct/images/n3turtle_doc.png)
![[ld+json]](/fct/images/jsonld_doc.png)
![[rdf+json]](/fct/images/json_doc.png)
![[rdf+xml]](/fct/images/xml_doc.png)
ODATA
![[atom+xml]](/fct/images/atom_doc.png)
![[html]](/fct/images/html_doc.png)
About
![[RDF Data]](/fct/images/sw-rdf-blue.png)
