Input - getBNN

Description

This is an interface to the Mitre Broadcast News Navigator (BNN). This PIM goes out to the MITRE Broadcast News Navigator site and looks for stories taken from the selected broadcast source and stored at BNN site. The stories are taken directly from the HTML generated by the BNN. The date, source, summary, text, and id of each story found are put into a BasicProduct. All of the BasicProduct Objects created are then contained within a VectorProduct, which is the PIM's product. Each product is augmented with meta-data from the BNN.

  

Configuration Variables

nDaysToSearchBack
The number of days back starting from the current date to search for stories. The default is one.

sourceSpec  

The news source from which to search for stories. The default is CNN_Prime_News.

searchWords

A list of words separated by space. The words result in an AND search.

findTopic

A topic name to search on. Must be a BNN generated Topic, a list of which can be found on the BNN web site.

findPerson

A person name to search on. Must be a BNN generated Person, a list of which can be found on the BNN web site.

findOrg

An organization name to search on. Must be a BNN generated Organization, a list of which can be found on the BNN web site.

findLocation

A location name to search on. Must be a BNN generated Location, a list of which can be found on the BNN web site.

bnnHostQuery

The query string to access BNN. A default is supplied.

Product

A VectorProduct containing all of the stories found, with each story stored as two BasicProduct Objects. The first listing of a story contains the story summary as the description. The second listing contains the full text of the story as the description. The Id of each story is in the form:

Each product is augmented with meta-data from the BNN. The meta-data includes organization, people, and location. The meta-data attributes are:

 

How it works

The PIM first generates the query string with which to get the stories using the current date, the number of days to search back, and the broadcast source. The query string is added to the base URL, http://videoserver3.mitre.org/tr/owa/BNN_REVG.BnnFindStories?, to form the full URL. The PIM then tries to connect to the URL. If there is a problem connecting (i.e. the connection timed out or could not be made), the PIM will make up to two more, for a maximum of three, attempts at connecting. If still unsuccessful at connecting the PIM will return a null product. If the connection was opened, the HTML page is parsed, generating a BNNStory Object for each story found. During parsing, each BNNStory found is added to a Vector holding all of the Objects. If there was a parse error found in the page, only the BNNStory Objects found before the parse error will be in the Vector. After parsing, the BNN PIM builds it's product from the Vector. For a more detailed description of the product created by the PIM, look at the Product section.

Revised: 24 Apr 1998