A Swarm of agents for language editing

Abstract

Natural Language Editing has many complexities and ambiguities involved with the process. Natural Language editing in an automated environment is a complex task and due to this complexity, current automated tools face the limitation of perfectly editing a given write up considering both syntactic and semantic accuracy. Therefore, there is a need of evolutionary, adaptive, autonomous and dynamic approach for natural language editing. As the solution, ‘A Swarm of Agents for Language Editing’ is a multi agent based approach for evolutionary English editing. This system comprises a swarm of agents with spell checking agents, noun agents, verb agents, determiner agents, style agents and manager agent where each agent activates upon a request, executes and kills them at the end of the task. Unlike traditional editing software, this system is based on negotiations among different agents rather than instruction driven. In this system, English editing task involves interaction among different agents who take care of different aspects of a sentence structure such as noun, verb and determiners. These agents together with the manager agent negotiate with each others to generate an agreeable output. A common message space is used to enable the communication and thus negotiations among different agents in the swarm to come up with a commonly agreeable edited text.