Knowledge-Based Recommendation Systems Effect on Likelihood of Purchase

Sammanfattning

The fragmented market for academic literature presents search friction for students, hinderingthem from finding the correct material in a reasonable timeframe. These inefficienciescould lower the chance of students purchasing said material, illustrating the need forbetter systems for this purpose. This work investigates the relationship between the implementationof a constraint-based knowledge-based recommender systems (KBRS) andusers’ likelihood of purchase. Through a controlled experiment, two online store versionswere compared: one search-based and one utilizing a KBRS. Tests involved students atLinköping University studying Industrial Engineering and Management, measuring findabilityand cognitive load via time-on-task and NASA-TLX, respectively. Results demonstratedthat implementing a KBRS increased the median likelihood of purchase by 10 percentagepoints, with 94% of participants preferring the store with a KBRS. Analysis revealsthat increased findability and decreased cognitive load each separately correlated positivelywith increased purchase likelihood (r = 0.6). Statistical analysis confirms that a positivecorrelation exists on a population level with 95% confidence. The findings indicatethat the implementation of a KBRS is associated with an increased likelihood of purchase,improved findability, and a reduction in cognitive load.