A new algorithm to find most similar items with respect to target item by using ahp and fuzzy membership functions

Finding most similar commodities with respect to selected one is a core subject in today's competitive business environment. This paper introduces a new method and a new algorithm which can be used by commercial web sites or trading centres. The algorithm is based on finding and listing most similar goods/services related to customer preferences. In other words, when the user selects an item, this algorithm sorts other items depending on each item's similarity to the selected main item and lists them to the customer. The aims of the algorithm are to make customer selection easier, to increase his/her willingness to buy and to give a sense of satisfaction of sufficient searching before the buying decision. Our algorithm consists of 4 main steps. Determining item categories, finding item property set for each category, Finding priority weights of the properties by using Analytical Hierarchy Process, Defining fuzzy variables and calculating membership function values, using the values found on previous steps, calculating the similarity score of each alternative item with respect to the main item. An empirical study was done on printers category. Among the alternatives, the algorithm found the most similar item with %96,1 similarity score.

,

A New Algorithm ,

Fuzzy Membership Functions ,

AHP ,