Intelligent Document Routing as a First Step towards Workflow Automation: A Case Study Implemented in SQL
In large and complex organizations, the development of workflow automation projects is hard. In some cases, a first important step in that direction is the automation of the routing of incoming documents. In this paper, we describe a project to develop a system for the first routing of incoming letters to the right department within a large, public portuguese institution. We followed a data mining approach, where data representing previous routings were analyzed to obtain a model that can be used to route future documents. The approach followed was strongly influenced by some of the limitations imposed by the customer: the budget available was small and the solution should be developed in SQL to facilitate integration with the existing system. The system developed was able to obtain satisfactory results. However, as in any Data Mining project, most of the effort was dedicated to activities other than modelling (e.g., data preparation), which means that there is still plenty of room for improvement.
- [Show abstract] [Hide abstract] ABSTRACT: Workflow management deals with a fair amount of free text -- scanned paper documents, web input forms, e-mail, input from mobile devices as well as texts from government agencies to citizens. The free-text input is intended for a human reader. Still, in eGovernment settings we would like to process such input before, while and after handling officers read it. This paper presents some examples on how language technology can increase efficiency and quality of workflow management.0Comments 2Citations