TY - JOUR AU - Zhang, Yihong AU - Hara, Takahiro PY - 2021/07/02 Y2 - 2024/03/29 TI - Predicting E-commerce Item Sales With Web Environment Temporal Background JF - Business Information Systems JA - Bus. Inf. Sys. VL - 1 IS - SE - Social Media DO - 10.52825/bis.v1i.37 UR - https://www.tib-op.org/ojs/index.php/bis/article/view/37 SP - 233-243 AB - <p><span style="left: 206.362px; top: 469.835px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.01183);">In this paper, we study the effect of Web environment temporal background in pre-</span><span style="left: 117.6px; top: 492.416px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(0.984366);">dicting e-commerce item sales, especially&nbsp; those in temporary sales. Temporary sales nowadays</span><span style="left: 117.6px; top: 514.998px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.01503);">are a popular strategy for quickly clearing inventories. For traditional&nbsp; recommender systems,</span><span style="left: 117.6px; top: 537.58px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.04026);">predicting the sales of an item is done based on its past purchase records. For temporary</span><span style="left: 117.6px; top: 560.163px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.02883);">sales items, however, such records are not available. In order to make recommendation for</span><span style="left: 117.6px; top: 582.745px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.01002);">such items, contextual information, such as product descriptions, is usually used. We investi-</span><span style="left: 117.6px; top: 605.326px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(0.993539);">gate whether temporal background in the Web environment can be additional useful contextual</span><span style="left: 117.6px; top: 627.908px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.03037);">information in recommender systems. It is assumed that items consistent with the temporal</span><span style="left: 117.6px; top: 650.49px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.00284);">background would have higher demands. We propose a method for representing the temporal</span><span style="left: 117.6px; top: 673.073px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(0.995914);">background using word embeddings of e-commerce activities and social media data, and eval-</span><span style="left: 117.6px; top: 695.655px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.00309);">uate their effect on sales prediction. Through empirical analysis with real-world data, we found</span><span style="left: 117.6px; top: 718.236px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(1.01669);">that temporal background does have positive effects for sales prediction. The findings in this</span><span style="left: 117.6px; top: 740.818px; font-size: 18.1818px; font-family: sans-serif; transform: scaleX(0.998749);">paper can be conveniently incorporated into future recommender system designs.</span></p> ER -