Xerox Corporation

Eric H Cheminot

Denys Proux

John Allott Moore

Francois Ragnet

Jean-Luc Meunier

Stefania Castellani

Frederique Segond

Thierry Armand Lehoux

Nicolas Raphaël Saubat

Hervé Déjean

Jean-Michel Renders

Yves Hoppenot

Caroline Privault

Nicolas Guerin

Caroline Hagege

Caroline Hagège

Philippe Rerole

This disclosure provides an exemplary method and system for extracting structured label and value pairwise textual data from a textual document. According to an exemplary method, initially a layout analysis is performed resulting in one or more alternatives for grouping and ordering the textual elements of interest. Next, textual elements are tagged as including a label term, a value term or a label and value term. Finally, a sequence-based method is applied to the tagged elements to generate one or more sequence listings representative of the label and value pairwise data structure(s) and label:value pairwise data is extracted.

Method and system of extracting label:value data from a document

A computer implemented system and method are disclosed for updating an electronic calendar. The method includes receiving an electronic message in a natural language in which a change in role is expressed and, with a natural language processor implemented by a computer processor, automatically detecting the change in role within the email message, optionally storing the change in role in a contacts database, and proposing updates for entries in an electronic calendar based on the detected change in role.

System and method for updating an electronic calendar

A system and method for generating separations between jobs in a batch of scanned pages are provided. The method includes, for each of a set of jobs, capturing an image of a representative page of the job. Thereafter, the method includes scanning the set of jobs as a batch to generate a set of scanned pages. The captured images are compared with scanned pages in the set of scanned pages to identify, for at least one captured image, a respective scanned page which matches the captured image. At least one separator is generated for separating the set of scanned pages based on a location of the matching scanned page(s).

Document separation by document sequence reconstruction based on information capture

A system, method, and apparatus for mark recognition in an image of an original document are provided. The method/system takes as input an image of an original document in which at least one designated field is provided for accepting a mark applied by a user (which may or may not have been marked). A region of interest (RoI) is extracted from the image, roughly corresponding to the designated field. A center of gravity (CoG) of the RoI is determined, based on a distribution of black pixels in the RoI. Thereafter, for one or more iterations, the RoI is partitioned into sub-RoIs, based on the determined CoG, where at a subsequent iteration, sub-RoIs generated at the prior iteration serve as the RoI partitioned. Data is extracted from the RoI and sub-RoIs at one or more of the iterations, which allows a representation of the entire RoI to be generated which is useful in classifying the designated field, e.g., as positive (marked) or negative (not marked).

Optical mark classification system and method

A method for separating and categorizing documents includes receiving a scanned batch of documents. The batch includes scanned documents to which document separator stamps have been applied before scanning. Each stamp includes machine recognizable patterns applied on a same page of a document, spaced by a designated field for receiving a user-applied category code. The scanned batch of documents is processed to identify pages that contain a document separator, including identifying at least one of two spaced patterns. For a document page for which a document separator is identified, the the corresponding designated field is located and the category code associated with the designated field identified. The document containing the is separated from other documents in the batch based the identified separator and a document category is assigned to the document, based on the identified category code.

Method for one-step document categorization and separation using stamped machine recognizable patterns

A computer implemented electronic out-of-office message analysis system and method are disclosed. The method includes, for each of a plurality of users, receiving a user-generated electronic out-of-office message in a natural language in which a time window of absence and at least one alternate named contact are expressed and, based on the out-of-office message, generating a structured representation of the out-of-office message which links the alternate contact to a normalized representation of the time window. The structured representation of the out-of-office message is stored in a database. From the database it can be determined whether a current user's out-of-office message conflicts with another user's out-of-office message. If a conflict is detected, the current user can be notified.

Detecting conflicts in email messages

A textual representation of a voice message is provided to a communication device, such as a mobile phone, for example, when the mobile phone is operating in a silent mode. The voice message is input by a caller and the voice message converted to phonemes. A text representation of the voice message is transmitted to the mobile phone. The representation includes characters based on the phonemes with well known words being represented in an easily understood shorthand format.

System and method for providing a textual representation of an audio message to a mobile device

A calibrated categorizer comprises: a multi-class categorizer configured to output class probabilities for an input object corresponding to a set of classes; a class probabilities rescaler configured to rescale class probabilities to generate rescaled class probabilities; and a resealing model learner configured to learn calibration parameters for the class probabilities rescaler based on (i) class probabilities output by the multi-class categorizer for a calibration set of class-labeled objects, (ii) confidence measures output by the multi-class categorizer for the calibration set of class-labeled objects, and (iii) class labels of the calibration set of class-labeled objects, the class probabilities rescaler calibrated by the learned calibration parameters defining a calibrated class probabilities rescaler. In a method embodiment, class probabilities are generated for an input object corresponding to a set of classes using a classifier trained on a first set of objects, and are rescaled to form rescaled class probabilities using a resealing algorithm calibrated using a second set of objects different from the first set of objects. The method may further entail thresholding the rescaled class probabilities using thresholds calibrated using the second set of objects or a third set of objects.

Categorizer with user-controllable calibration

A method for detection of phishing attempts in received electronic mail messages includes retrieving the source code, displayed text, and a list of all specified addresses contained within the source code of a received electronic message. Visual character normalization is applied to each specified address to develop all possible address combinations and to form a normalized address list. The specified addresses are removed from the normalized address list to create a revised address list, upon which comparison tests are performed to determine if each address in the revised address list is from a valid source. The recipient of the electronic message is informed of any message found to be from an invalid source.

Growing community of inventors

Meylan, France

Eric H Cheminot