From news to facts: An Hadoop-based social graphs analysis

摘要

This paper describes a system combining a distributed setup based on Hadoop, MapReduce, Impala and a general semantic model focusing on common entities (people, organizations, places) and their connections as co-occurrences and facts offering the analysts the opportunity to do mining in social networks. Emphasis is given to recall rather than precision, suggesting the analyst many possible relations and connections to be explored. Early studies have shown interesting results, and further explorations are planned to extend the reach and abilities of the system.