Kachako:A Hybrid-Cloud Unstructured Information Platform for Full Automation of Service Composition,Scalable Deployment and Evaluation Natural Language Processing as an Example

摘要

Automation is the key concept when designing a service platform, because automation could reduce human's work. Focusing on unstructured information such as text, image and audio, we implemented our service platform "Kachako" in a hybrid-cloud way where services themselves are transferred on demand. We suggest making each service specified by its input and output types, and executable of the service portable, compatible and interoperable. Assuming such services, Kachako thoroughly automates everything that users need. Kachako provides graphical user interfaces allowing end users to complete their tasks within Kachako without programming. Kachako is designed in a modular way by complying with well-known frameworks such as UIMA, Hadoop and Maven, allowing partial reuse or customization. We showed that Kachako is practically useful by integrating our natural language processing (NLP) services. Kachako is the world first full automation system for NLP freely available.