Reputation and feedback systems in online marketplaces are often biased, making it difficult to ascertain the quality of sellers. We use post-transaction, buyer-to-seller message traffic to detect signals of unsatisfactory transactions on eBay. We posit that a message sent after the item was paid for serves as a reliable indicator that the buyer may be unhappy with that purchase, particularly when the message included words associated with a negative experience. The fraction of a seller's message traffic that was negative predicts whether a buyer who transacts with this seller will stop purchasing on eBay, implying that platforms can use these messages as an additional signal of seller quality.

We present an algorithm for language identification, in particular of short documents, for the case of an Internet domain with sites in multiple countries with differing languages. The algorithm is significantly faster than standard language identification methods, while providing state-of-the-art identification. We bootstrap the algorithm based on the language identification based on the site alone, a methodology suitable for any supervised language identification algorithm. We demonstrate the bootstrapping and algorithm on eBay email data and on Twitter status updates data. The algorithm is deployed at eBay as part of the back-office development data repository.

Data mining techniques are routinely used by fundraisers to select those prospects from a large pool of candidates who are most likely to make a financial contribution. These techniques often rely on statistical models based on trial performance data. This trial performance data is typically obtained by soliciting a smaller sample of the possible prospect pool. Collecting this trial data involves a cost; therefore the fundraiser is interested in keeping the trial size small while still collecting enough data to build a reliable statistical model that will be used to evaluate the remainder of the prospects.We describe an experimental design approach to optimally choose the trial prospects from an existing large pool of prospects. Prospects are clustered to render the problem practically tractable. We modify the standard D-optimality algorithm to prevent repeated selection of the same prospect cluster, since each prospect can only be solicited at most once.We assess the benefits of this approach on the KDD-98 data set by comparing the performance of the model based on the optimal trial data set with that of a model based on a randomly selected trial data set of equal size.