My attempts at simplifying down a fairly complex procedure that I started working on in Week 11 is my Hierarchy system. Basically the point of this is to give a better chance of viewing to those RE:MAP stories which share common keywords with the negative articles from traditional media accessible from that location over the past 2 days. I imagined this to be an automated system so I have mapped out the commands a computer (maybe somewhat more sophisticated with language than current software) would take to complete the hierarchy structure.
To cut it brief (however you can rea into the diagram as required), a score is given to each story based on a percentage of media coverage for said keyword. All the keywords in one story are than added up and tallied against the other stories to give a number which can be compared across all stories. This number than determines the order in which the stories show up.
The keywords are derived from a large database separated into 3 tiers...
Each tier represents a different level of perspective. If we go along the top row for example Afghanistan is a subset of the Middle East, and Kabul is a subset of Afghanistan. This linking is important as if a top trending news articles continuously refers to "Afghanistan", than a story which mentions Kabul, Kandahar or Herat for example, will be linked to this keyword, but given a coefficient when scoring to represent that indirectness.
The other thing to note are there are 3 main topics the database is made up of. The first is the easiest to categorise and is Geographic words. These include town, city, state, country, and continent names. The second topic is religious words, categorised into the main religious metagroups, the smaller branches of these religions and finally the iconography or terms associated with that religious subgroup. The third topic are cultural terms. This refers to indigenous cultures, immigrants, refugees and other non-religiously defined cultural groups.
No comments:
Post a Comment