Analysis of Item Selection Behavior in Online Shopping

Opportunities to purchase products from online shopping sites continue to increase. One feature of online shopping sites is searching for products by entering queries on input forms. Analyzing user search behavior is critical for appropriate search support and sales improvement. In this paper, we assigned five search behavior codes from continuous logs: replacing keywords (R), modifying keywords (M), adding keywords (A), deleting keywords (D), and continuing same keywords (C). Using code sequences, we analyzed frequent search patterns and the relationship between initial search patterns and search length. As a result of evaluation experiments, we found that behavior, which moves to another search result page without changing the query, occupies a majority of search behavior. Search behavior also becomes longer when it follows the initial stage of the search.