Abstract: Techniques for analyzing recommendations may be described. In particular, an item offered from a network-based resource may be selected based at least in part on past orders for the item. Recommended items offered from the network-based resource may also be identified. A determination may be made as to whether the recommended items may include the item. Based on this determination, an indication of an issue associated with recommending the item may be generated. Based on the indication, a workflow to identify a potential cause of the issue may be initiated.

Abstract: Disclosed are various embodiments for identifying correlations between child nodes of a taxonomy. A correlation service identifies a correlation between parent nodes of the taxonomy. A data set such as an interaction history is filtered according to the parent node correlation. The filtered data set is then used to identify correlations between the respective child nodes of the correlated parent nodes.

Abstract: Features are disclosed for identifying randomized latent feature language modeling, such as a recurrent neural network language modeling (RNNLM). Sequences of item identifiers may be provided as the language for training the language model where the item identifiers are the words of the language. To avoid localization bias, the sequences may be randomized prior to or during the training process to provide more accurate prediction models.

Abstract: Features are disclosed for identifying and routing items for tagging using a latent feature model, such as a recurrent neural network language model (RNNLM). The model may be trained to identify latent features for catalog items such as movies, books, food items, beverages, and the like. Based on similarities in latent features, tags previous assigned to items may be applied to untagged items. Application may be manual or automatic. In either case, resources need to be balances to ensure efficient tagging of items. The included features help to identify and direct these limited tagging resources.

Abstract: The present technology may identify item category affinities by identifying a plurality of classifications of an item. An accuracy of the plurality of classifications relative to one another for the item may be identified. A category affinity of the item may be determined based on the accuracy of the plurality of classifications relative to one another.