At some point, we determined that fetching, holding, and/or manipulating an entire message with the knowledge graph was problematic, so we set up a whole bunch of machinery to fetch node/edge details from Neo4j on demand. I'd like to revisit this decision in light of #523.
At some point, we determined that fetching, holding, and/or manipulating an entire message with the knowledge graph was problematic, so we set up a whole bunch of machinery to fetch node/edge details from Neo4j on demand. I'd like to revisit this decision in light of #523.