Designing a Human-Machine Hybrid Computing System for Unstructured Data Analytics
This work addresses the need for more accurate and reliable analytics of unstructured data, which is a domain-specific problem, but it appears incremental as it builds on existing hybrid computing concepts.
The paper tackles the problem of low accuracy in machine algorithms for unstructured data analytics by proposing a hybrid human-machine computing platform with integrated service level objectives (SLO) management, achieving highly encouraging initial experimental results.
Current machine algorithms for analysis of unstructured data typically show low accuracies due to the need for human-like intelligence. Conversely, though humans are much better than machine algorithms on analyzing unstructured data, they are unpredictable, slower and can be erroneous or even malicious as computing agents. Therefore, a hybrid platform that can intelligently orchestrate machine and human computing resources would potentially be capable of providing significantly better benefits compared to either type of computing agent in isolation. In this paper, we propose a new hybrid human-machine computing platform with integrated service level objectives (SLO) management for complex tasks that can be decomposed into a dependency graph where nodes represent subtasks. Initial experimental results are highly encouraging. To the best of our knowledge, ours is the first work that attempts to design such a hybrid human-machine computing platform with support for addressing the three SLO parameters of accuracy, budget and completion time.