Abstract: Systems for determining a merchant name as used in mobile expense reporting applications. Embodiments commence upon receiving an image of an expense item receipt from a mobile user device. In addition to the image, embodiments receive metadata derived from a travel app running on the user device, which metadata is sent together with an association to the image. Further processing extracts a set of characteristics of text-containing regions of the expense item receipt, and then forms a feature vector comprising the extracted characteristics of the text-containing regions as well as aspects of the metadata received from the user device. The feature vector is input into a predictor to determine at least one predicted merchant name. The metadata can comprise information selected from a mobile app that is operational on the user terminal. Such information can comprise user travel profile information and/or user itinerary information and/or user mileage tracker app information.

Abstract: Systems for determining a merchant name as used in mobile expense reporting applications. Embodiments commence upon receiving an image of an expense item receipt from a mobile user device. In addition to the image, embodiments receive metadata derived from a travel app running on the user device, which metadata is sent together with an association to the image. Further processing extracts a set of characteristics of text-containing regions of the expense item receipt, and then forms a feature vector comprising the extracted characteristics of the text-containing regions as well as aspects of the metadata received from the user device. The feature vector is input into a predictor to determine at least one predicted merchant name. The metadata can comprise information selected from a mobile app that is operational on the user terminal. Such information can comprise user travel profile information and/or user itinerary information and/or user mileage tracker app information.