Date of Conferral
Doctor of Information Technology (D.I.T.)
The purpose of this correlational quantitative study was to examine the possible relationship between user-driven parameters, user ratings, and ranking algorithms. The study’s population consisted of students and faculty in the information technology (IT) field at a university in Huntington, WV. Arrow’s impossibility theorem was used as the theoretical framework for this study. Complete survey data were collected from 47 students and faculty members in the IT field, and a multiple regression analysis was used to measure the correlations between the variables. The model was able to explain 85% of the total variability in the ranking algorithm. The overall model was able to significantly predict the algorithm ranking discounted cumulative gain, R2 = .852, F(3,115) = 220.13, p < .01. The Respondent DCG and Search term variables were the most significant predictor with p = .0001. The overall findings can potentially be useful to content providers who focus their content on a specific niche. The content created by these providers would most likely be focused entirely on that subgroup of interested users. While it is necessary to focus content to the interested users, it may be beneficial to expand the content to more generic terms to help reach potential new users outside of the subgroups of interest. User’s searching for more generic terms could potentially be exposed to more content that would generally require more specific search terms. This exposure with more generic terms could help users expand their knowledge of new content more quickly.
Taylor, Gary Michael, "Search Results: Predicting Ranking Algorithms With User Ratings and User-Driven Data" (2020). Walden Dissertations and Doctoral Studies. 8337.