...
- Map of Science and Temporal Graphs significantly faster
- Under 3 seconds for a 1,218,694 Quad dataset (previously 1 minute 20 seconds)
- Approx 2 minutes for a 24,647,681 Quad dataset. Contains 145,000 people, 155,000 publications and 14,000 journals
- Person level Map of Science return in under 2 seconds, using direct queries of the triple store
- Person level Map of Science will use the system-level cache once queries take longer than 2 seconds, if the system-cache has been populated
- Background refresh of Map of Science / Temporal Graph data - once populated, all requests are served from the cache whilst refreshes occur
- CoAuthor and CoInvestigator visualisations use short-lived caches to prevent multiple executions of the same query in rendering a single visualisation
- Minor tweak to CoAuthor query to improve performance
- Sparklines use some of the under the hood improvements
- New Added AltMetric embed code to display badges on the article pages - enable via the runtime.properties (see example.runtime.properties for details)
Memory Usage
- New data structures for Map of Science / Temporal Graphs use lightweight Java objects instead of Jena models (should use much less memory)
- Search Indexer does not queue statements to index if paused and a full rebuild has been requested (much lower memory usage during reasoning)
...