Improving Educators’ Search Engine Experience: A Quantitative Analysis of Search Terms

Abstract

K-12 educators within the University of Missouri’s Network for Educator Effectiveness (NEE) utilize the EdHub Library for professional development (PD), hosting over 500 self-paced activities since 2014. Despite the common use of click-through rates (CTR) to gauge user engagement in search engines, this metric needs to be more accurate due to their lack of ordinality in analyzing search terms. Also, CTR calculations of search terms ranging from 0% to 100% do not represent user engagement with search results that satisfy educators’ search goals. This study proposes a model predicting CTR, using SelectKBest for feature selection and ranking search terms by the Chi-square statistic. Out of 1,317 search terms, 296 (22.5%) were relevant. After outlier removal of extreme CTR values and one-hot encoding of relevant search terms, two regression models achieved over 0.98 accuracy, categorizing 47 search terms into four groups representing the NEE teacher evaluation system components, PD resources for teachers, Teacher Standards, and PD resources for school administrators. The study suggests that 36 of the most searched terms need optimization to reduce users’ cognitive load in search results. Notably, the most searched terms had around 40 search results, oscillating between 0.04 and 0.87 CTR, revealing educators’ search priorities and tolerance for browsing search results. Overall, this research contributes ordinality to the search engine dataset, shedding light on educators’ preferences and guiding improvements in search result relevance and usability.

Keywords

• Educational data mining
• information retrieval
• teacher professional development