Technical Report 2017-11-01
Internetworking and Media Communications Research Laboratories
Department of Computer Science, Kent State University
http://medianet.kent.edu/technicalreports.html

 

Dissertation Title:

A Computational Mimicry of the Knowledge Augmentation Process in Comprehension Based Learning

Amal Babour

ababour@kent.edu

Advisor: Prof. Javed I. Khan
Department of Computer Science
Kent State University
  November 2017

Abstract:

Comprehension is one of most dominate means of human learning. The process of human comprehension has long studied in psychology, yet no algorithm level model for comprehension is available to-date. In this thesis we explore a plausible computational model behind a special form of comprehension- prose comprehension. [definition- is a graph forest with multiple disjoint components indicate reference, prior knowledge are used implicity by human]. We suggest a comprehension engine consisting of two major cognitive processes involved; knowledge induction and distillation. The knowledge induction process seeks to increase the knowledge, in particular augmenting the associations by reading incremental external reference texts and finding the highest familiarity knowledge associations among the prose concepts and uses ontology engine to find lexical knowledge associations among each pair of concepts to obtain a knowledge graph with single giant component to establish a base model for the prose comprehension. We experiment using a version of Steiner Tree called Terminal-to-Terminal Steiner Tree (TTST) that mimic finding the highest familiarity knowledge associations (links) among the prose concepts through no or minimum number of external concepts and associations, which established The time complexity of the algorithm is O(C+E). The distillation process grades all the knowledge associations between each pair of the prose concepts and selects a subgraph which has the best familiar easy to understand knowledge associations that that can be useful for comprehending the relation between each pair of the prose concepts and present them to the reader as an enhanced text. We suggest to use an equivalent electrical circuit EEC for grading the knowledge associations and select the one which has the highest delivered current flow between the two concepts.  We conduct an experiment on three proses to evaluate the efficiency of the comprehension gained from the comprehension engine. The used reference text is Wikipedia and the used ontology-engine is Wordnet. In addition, we involved human readers to study the impact of the acquired knowledge gained form the reduction phase on the prose comprehension. We suggest a computational evaluation model to measure the quantitative insight of the acquired knowledge and the learning process on the prose comprehension obtained by the comprehension engine. The results of the experiment verify the efficiency of the compression engine for improving the quality of comprehension and saving time.

 

Last Modified: Sep 2017.