We have proposed a next-level interface and query language for PMG, that could help their customers gain more control over the search profiles they define that filter the news stream.
We have built a prototype that runs on production data to show how using a more high-level query language than the legacy boolean query format, can help the system be more more pro-active and give semantic recommendations, and help give the end user more precise feedback on the effectiveness of their queries.
By changing the ElasticSearch analyzers and query structures, we are able to achieve higher precision with less performance impact.
We are now assisting PMG in analyzing what it would take to bring a system like this to their end users.