Cognitively Mediated Research Discovery - A context-aware rich visualized knowledge graph co-created by humans and machines using a common language

Abstract We believe the language of choice, and the means by which humans and machines interact, and contribute knowledge, will play a key role in the success of future cognitive systems. In earlier research, we have investigated the potential role of a hybrid human-machine language for knowledge representation to serve as the basis for a highly agile and collaborative interface between human and machine agents. Inspired by the promise of rich semantics but motivated by the desire for simplicity, practicality, and applicability - especially the ability to engage with subject matter experts (SMEs) rather than technical specialists we developed a Controlled Natural Language (CNL) named Controlled English (CE). We have then applied CE in a wide variety of problem domains from decision-making to improving situation awareness among communities of interests, especially in military coalition environments wherein information is dynamic and complex. Our recent work has been focused on integrating natural language conversational capability to the core CE, thus enabling the human users to request information and, more importantly, contribute knowledge in a fully naturalistic manner. We have published experimental results, based on a series of crowd-sourcing collective intelligence games, which show that untrained human users can converse with such a system, learning to contribute meaningful information within a short timeframe. We will present details of this language, the principles against which it was created and highlight key differences to similar languages available.
Authors
  • Geeth de Mel (IBM UK)
  • Dave Braines (IBM UK)
  • Anna Thomas (IBM UK)
  • Tien Pham (ARL)
  • Will Dron (BBN)
Date Sep-2017
Venue Hybrid Human-Machine Computing (HHMC): From Human Computation to Social Computing and Beyond