Currently, Aigents extracts pattern-based "news items" on per-topic (is attribute) and per-url (sources attribute) basis for a specific day (times attribute), which are represented as short excepts.
Task: Create aggregated content generation based on the "news items" found above
Level 1: Simple aggregation: can be defined as a user-specific property on how to cook the news for every specific user - using "news aggregation" property with 5 values (none, summary, overview, digest, history)
no aggregation - "none"
aggregation per day+url+topic - "summary"
aggregation per day+url - "overview"
aggregation per day+topic - "digest" (similarly to the format of the digests currently sent by Aigents email notifications)
aggregation per topic (across days) - history
Level 2: Complex formation: In addition to the above, combinations of the topics corresponding to each other and clusters of related topics can be used together with LinkGrammar-based formal grammar (and possibly some underlying ontology) to generate literary content generation describing novel (salient and "surprising") combinations of topics - based on progress with #22 .
Currently, Aigents extracts pattern-based "news items" on per-topic (is attribute) and per-url (sources attribute) basis for a specific day (times attribute), which are represented as short excepts.
Task: Create aggregated content generation based on the "news items" found above
Level 1: Simple aggregation: can be defined as a user-specific property on how to cook the news for every specific user - using "news aggregation" property with 5 values (none, summary, overview, digest, history)
Level 2: Complex formation: In addition to the above, combinations of the topics corresponding to each other and clusters of related topics can be used together with LinkGrammar-based formal grammar (and possibly some underlying ontology) to generate literary content generation describing novel (salient and "surprising") combinations of topics - based on progress with #22 .