Search Algorithm

Well, from what I have read you can simply remove what is referred to as vtermfrequency which means that when your algorithm searches for a term that you tell it you don’t care about how frequently the term appears in a document and your results are then not displaying based on the number of times a search term is present.

This would ensure that your returns have the term present but are not purely predicated on the number of times it appears.

Or you can actually have search results become negatively marked due to duplicate content. The number would have to be realistic but who really needs to reference the name of the item you’re selling more than 5 times?

You can also bias returns based on content length which would favor those who provide more information. This all depends on what standard Gameflip want to use to determine a “good” listing.

Personally I think listings should be ranked on relevance, content length and if the same user is displayed on a table multiple times then after their first return their other listings should be relegated to the bottom of the list. This would give buyers a better spread of their options. Filters then could be used to refine what is displayed in order of preference.

1 Like