Determining the Effectiveness of Multi-user, Hybrid, Human-Computer Assessments for High Recall Information Retrieval Systems