Ahh... I understand now. To rephrase what I understand your problem to be is that you basically want to use the cart as the user and recommend based on a history composed of items already in the cart.
That should work just fine. I would recommend combining that with user level recommendations by using search abuse. This would give you two levels of recommendations in one step. These levels would be cart level recommendations and (presumably weaker) user level recommendations. Similarly, you could cross-recommend from item meta-data to item purchases or to purchased item meta-data. This last would be tricky to do in a single search engine request. On Thu, Feb 14, 2013 at 8:41 PM, Pat Ferrel <[email protected]> wrote: > Sure, we have cart/session IDs, items IDs, and user IDs when purchases are > made or when asked for a recommendation from the cart page. > > We currently don't get the add-to or remove-from cart actions. We could > get them. > > Are you thinking that we can use the add-to-cart user x item matrix and > purchase user x item matrix to get purchase recs from add-to-cart actions? > Interesting idea. This could be combined with the purchase recs from > show-details to get even better recs given an item context. > > In this case I was looking for a way to do use the purchases cart x item > matrix to get recommendations by finding the cart in the training matrix > most similar to the cart contents at runtime. In which case we have to use > the current cart contents as the query to find the most similar cart in the > matrix, then return recs for that cart. At least that was what I was > thinking. > > On Feb 14, 2013, at 6:09 PM, Ted Dunning <[email protected]> wrote: > > Do you see the contents of the cart? > > Is the cart ID opaque? Does it persist as a surrogate for a user? > > On Thu, Feb 14, 2013 at 10:30 AM, Pat Ferrel <[email protected]> wrote: > > > I thought you might say that but we don't have the add-to-cart action. We > > have to calculate cart purchases by matching cart IDs or session IDs. So > we > > only have cart purchases with items. > > > > If we had the add-to-cart and the purchase we could use your cross-action > > method for getting recs by training only on those two actions. > > > > Still without the add-to-cart the method below should work, right? The > > main problem being finding a similar cart in the training set quickly. > Are > > there other problems? > > > > On Feb 14, 2013, at 9:19 AM, Ted Dunning <[email protected]> wrote: > > > > I think that this is an excellent use case for cross recommendation from > > cart contents (items) to cart purchases (items). The cross aspect is > that > > the recommendation is from two different kinds of actions, not two kinds > of > > things. The first action is insertion into a cart and the second is > > purchase of an item. > > > > On Thu, Feb 14, 2013 at 9:53 AM, Pat Ferrel <[email protected]> > wrote: > > > >> There are several methods for recommending things given a shopping cart > >> contents. At the risk of using the same tool for every problem I was > >> thinking about a recommender's use here. > >> > >> I'd do something like train on shopping cart purchases so row = cartID, > >> column = itemID. > >> Given cart contents I could find the most similar cart in the training > > set > >> by using a similarity measure then get recs for this closest matched > > cart. > >> > >> The search for similar carts may be slow if I have to check for pairwise > >> similarity so I could cluster and find the best cluster then search it > > for > >> the best cart. I could create a decision tree on all trained carts and > > walk > >> as far as I can down the tree to find the cart with the most > > cooccurrences. > >> There may be other cooccurrence based methods in mahout??? With the id > of > >> the cart I can then get recs from the training set. I could also fold-in > >> the new cart contents to the training set and ask for recs based on it > >> (this seems like it would take a long time to compute). This last would > >> also pollute the trained matrix with partial carts over time. > >> > >> This seems like another place where Lucene might help but are there > other > >> mahout methods to look at before I diving into Lucene? > > > > > >
