Skip to Main Content
 

Global Search Box

 
 
 
 

ETD Abstract Container

Abstract Header

Measuring Goal Similarity Using Concept, Context and Task Features

Abstract Details

2018, Master of Science (MS), Wright State University, Computer Science.
Goals can be described as the user’s desired state of the agent and the world and are satisfied when the agent and the world are altered in such a way that the present state matches the desired state. For physical agents, they must act in the world to alter it in a series of individual atomic actions. Traditionally, agents use planning to create a chain of actions each of which altering the current world state and yielding a new one until the final action yields the desired goal state. Once this goal state has been achieved, the goal is said to have been satisfied. Since these goals involve physical actions, we can describe these goals as being physical goals. Our work focuses on a special type of goal that doesn’t exist physically and are knowledge goals. Much like physical goals, knowledge goals also have a desired state but this desired state is of the user’s understanding. Once the user has learned the missing information, the knowledge goal has been satisfied. While physical goals are given to agents who must then produce a plan of actions to alter the world, knowledge goals are given to an agent who must then produce a sequence of intermediate knowledge goals to alter the user’s state of knowledge. Much like how individual actions comprise a plan to alter the physical world, individual questions comprise a goal trajectory and alter the state of a user’s knowledge. This overall path of inquiry is much like that of an investigation for knowledge not unlike those of a detective or investigator. Given that not all users learn the same way, creating a plan to solve a knowledge goal is not a trivial task. Furthermore, in complex domains, it is not immediately clear to user themselves what their knowledge goal is as they continue to understand how to phrase the correct questions. As the user continues to refine their questions, their search grows in length and often in complexity as questions become increasingly specific. To address these issues, we created and evaluated a case-based goal reasoning system with the ability to measure similarity between goals.
Michelle Cheatham, Ph.D. (Advisor)
Michael Cox, Ph.D. (Committee Member)
Michael Raymer, Ph.D. (Committee Member)
79 p.

Recommended Citations

Citations

  • Eyorokon, V. (2018). Measuring Goal Similarity Using Concept, Context and Task Features [Master's thesis, Wright State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=wright1534084289041091

    APA Style (7th edition)

  • Eyorokon, Vahid. Measuring Goal Similarity Using Concept, Context and Task Features. 2018. Wright State University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=wright1534084289041091.

    MLA Style (8th edition)

  • Eyorokon, Vahid. "Measuring Goal Similarity Using Concept, Context and Task Features." Master's thesis, Wright State University, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=wright1534084289041091

    Chicago Manual of Style (17th edition)