Researcher
Research institution
Champion
Focus team
Project status
Year ended
2013
Project ID
201004
Abstract

Dr. Sandeep Purao of Penn State University developed a computer-based tool to parse procedures into parts of speech and organize multiple steps as a chunk. Since it was found that over 90% of the steps in procedures were not unique (appearing in more than one procedure), the chunking of the steps would reduce the volume of procedure material required to be kept current by 90%, with greater consistency in the steps themselves.

Objective

The creation of a tool that performs a semantic analysis of procedures to extract structured knowledge and generate knowledge chucks, outputted using XML (SPA 1.0) with three goals: (1) make Explicit knowledge more accessible and manageable, (2) understand how operators use tacit knowledge; and (3) understand tacit-explicit conversions.

Driving questions

Task 1:

  • How to better maintain the operating procedures?
  • How to make the procedures available to the operators at the point of action?
  • How to overcome the problem of increasing scale?

Task 2:

  • What are the key components of expertise for operator behaviors?
  • What are the differences between expert and novice operator behaviors?
  • Is it possible to specify these behaviors and the differences crisply to allow their use in hastening the move from novice to expert?

Task3:

  • How and why do we to move from tacit knowledge to procedures?
Background

Recent events industry have further accelerated the increase in procedure importance, resulting in more procedures with greater length. The volume of material tends to inhibit revisions and updating. This is compounded with procedures being written on a per unit basis, with some operators (notably console) having responsibility for more than one unit during key upsets.

However, little of the information in each procedure is unique. Up to 80% is just duplication of the basic tasks for different events and in different sequences. Modularization of the 20% that is unique would enhance updating and allow the modules to be combined for maximum benefit of each operator.

Deliverables

Semantic Procedure Analyzer (SPA) software which examines the meaning of terms contained in procedures one-at-a-time or as-a-set to remove extraneous information; identify parts-of-speech such as actions, actors and things; generate tabular representations of instructions in procedures; and locate logical chunks of instructions by heuristics or commonalities.

Determination of information not contained in procedures (tacit knowledge) and how that varies with expertise.