Comparison of Two Proposed Formats for Semantic Microblogging

Proposed Protocols for Semantic Twittering

Recently twitterdata.org introduced a format for sending semantic triplets through Twitter.

An alternative format is, of course, our RoboCrunch Action Protocol.

In #twitterdata a "predicate" must be a single word beginning with a $sign.  A "predicate" is followed by a (possibly multi-word) "object" until another $sign is encountered.  Several "predicates" can be in the same Tweet.  A "subject" an be (optionally) provided, and one can even switch to another "subject" in the middle of a Tweet.

Twitterdata.org is using the words name and value for "predicate" and "object", but in this posting we are using the familiar Subject - Predicate - Object terminology.

Action Protocol is using a slightly different approach.  Like #twitterdata, it allows for including multiple semantic triplets in one Tweet.  Each Semantic Tweet starts with the name (or a link to a description) of an Entity. Then, the list (enclosed in square brackets) of Attribute-Value pairs follows, providing as many sematic triplets as necessary for a given entity. A colon separates between an Attribute and a Value.  Transition to a new Attribute is signalled by a semicolon.

Advantages of Robot RoboCrunch Action Protocol

1) The Action Protocol uses a format more similar to a V3 Protocol familiar to students of the Semantic Web ("don't fix what ain't broken"".  

2) The special characters ":", ";", "[", "]" are used with a meaning familiar both to a non-technical reader and to developers. 

3) The Action Protocol also provides a query language with a syntax identical to the Tweeting format.  When the square bracketed list of Attribute-Value pairs doesn't follow a name of an entity, it is understood as a Query.  The implementation is supposed to return a list of entities that fit the given Values of the specified Attributes. Therefore, the Action Protocol can be used not only for teaching the Twitter community about new facts, but also for retrieving already existing information.

4) In the Action Protocol, all three components, (Entity, Attribute, and Value) can be assigned multi-word expressions.  We are doing away with the archaic tradition of limiting identifiers to a single word.  This significantly improves the readability of the scripts.

5) There are times that one may want to provide multiple Values for an Attribute. In Action Protocol the alternative values are separated by a vertical line ("|"). Commas, hyphens, exclamation points, and various other punctuation symbols can be used in the names of Entities and Values avoiding unnecessary acrobatics needed to encode values.  For example, Los Angeles, CA, doesn't need to become something like LosAngeles_CA, etc.

 

 

Attachment

2_g1.jpg

12/v8/gr/zm/9-/38/g/b0bsl79jx59rwtw7lr978psshbdk7tsxlb1w4wlpmfhprmt/2_g1.jpg