Abstract: A method and server for determining rank positions of non-native items by a ranking system is provided. Native items are rankable by the ranking system and non-native items are non-rankable thereby. The method comprises acquiring user interactions associated with items. Each item is associated with a visual position on a SERP such that the native items and the non-native item are arranged in a visual 2-D arrangement having rows, each row has at least two horizontally-arranged items. The method comprises generating a usefulness score for each non-native item by analyzing user interactions associated with (i) the non-native item, (ii) items being horizontally-arranged after the non-native item and (iii) with items being vertically-arranged after the non-native item. The method comprises training the ranking system to predict a predicted usefulness score. The method also comprises determining an optimal ranked position for a non-native item by the ranking system during an in-use phase.

Abstract: A method and server for determining rank positions of non-native items by a ranking system is provided. Native items are rankable by the ranking system and non-native items are non-rankable thereby. The method comprises acquiring user interactions associated with items. Each item is associated with a visual position on a SERP such that the native items and the non-native item are arranged in a visual 2-D arrangement having rows, each row has at least two horizontally-arranged items. The method comprises generating a usefulness score for each non-native item by analyzing user interactions associated with (i) the non-native item, (ii) items being horizontally-arranged after the non-native item and (iii) with items being vertically-arranged after the non-native item. The method comprises training the ranking system to predict a predicted usefulness score. The method also comprises determining an optimal ranked position for a non-native item by the ranking system during an in-use phase.