Abstract: A claims preprocessor processes claim data to identify claims that are to be adjudicated. Each claim includes at least one claim exception. The claims preprocessor further prioritizes the claim exception of each identified claim based on the claim data. A robotic process automator then orchestrates adjudication of the identified claims based on claim data. Further, a rules engine adjudicates the identified claims based on pre-defined rules. Subsequently, a fall out handler determines if any of the identified claims are incorrectly adjudicated and identify an issue associated with incorrect claims adjudication on determining that any of the identified claims are incorrectly adjudicated. A self learner then provides feedback to rules engine based on a decision tree and information received from fall out handler, the feedback being usable to resolve the issue. The information received from fall out handler is indicative of issue associated with incorrect claims adjudication.

Abstract: A machine learning document processing system performs natural language processing (NLP) and machine learning to determine a subset of documents from a document dataset based on the structural features and semantic features. The system facilitates an interactive process, e.g., through a client application, to receive user input from a user to identify documents with a specific document feature category. The user input may be provided from a user as speech or text, and NLP is performed on the user input to determine user intent, the document features, and document feature category. Using the user intent and the additional document feature category, the system identifies subsets of the document dataset that matches the document feature category for display.

Abstract: In some examples, image content moderation may include classifying, based on a learning model, an object displayed in an image into a category. Further, image content moderation may include detecting, based on another learning model, the object, refining the detected object based on a label, and determining, based on the another learning model, a category for the refined detected object. Further, image content moderation may include identifying, based on the label, a keyword associated with the object, and determining, based on the identified keyword, a category for the object. Further, image content moderation may include categorizing, based on a set of rules, the object into a category, and moderating image content by categorizing, based on aforementioned analysis the object into a category. Yet further, image content moderation may include tagging, based on fusion-based tagging, the object with a category and a color associated with the object.

Abstract: A system for reconciliation comprises a determination engine to determine whether data is structured or unstructured, a data structuring engine to structure the data, and a rule extraction engine to determine relations between pairs of values of a first set and a second set of data. The system further comprises a matching engine to generate a confidence score for each pair of the values, a categorization engine to classify the pairs of values into matched pairs and unmatched pairs, a validation engine to validate matching and classification of the pairs based on a user feedback, and a learning engine to store details pertaining to the validation of the matching and the classification over a period of time. The learning engine forwards the details to the rule extraction engine and the categorization engine to determine the relations between subsequent pairs of values and classify the pairs based on the stored details.

Abstract: A narrative response generator receives a user data query specifying variables and data sources from which to extract information desired by a user. The narrative response presents the information desired by the user in a non-textual format such as graphs and a textual format such as one or more paraphrases that are automatically generated by a sentence struct model. The sentence struct model generates context free grammar (CFG) which provides templates for generating word sequences that contain natural language words and placeholders. The placeholders are replaced with values obtained from the user data query for generating grammatically-accurate, complete paraphrases. The narrative response may additionally include information extracted from external data sources.

Abstract: In some examples, target object color analysis and tagging may include ascertaining an attribute of an image, and determining, based on the ascertained attribute, a target object that is to be identified and color tagged in the image. Based on a learning model, a plurality of objects may be extracted from the image. Based on a comparison of the target object and the plurality of extracted objects, the target object may be identified in the image. Color information may be extracted from the identified target object, and a plurality of color tags associated with the identified target object may be ascertained. A plurality of color distances may be determined between the color information and the plurality of color tags. Based on a determination of a minimum color distance from the plurality of color distances, a color tag that is to be assigned to the identified target object may be determined.

Abstract: Implementations are directed to providing an electronic document, and include receiving text content including a plurality of segments, the text content being received from data sources, determining a set of topics to be included in the electronic document, for each topic in the set of topics, providing a set of contextual words associated with a respective topic, contextual words being determined from a lexical database, each contextual word having a respective frequency, determining a score for each segment and topic pair, the score indicating a relevance of a respective topic to a respective segment, each score being determined based on respective contextual words of the respective topic and frequencies of the respective contextual words, for each topic, providing, by the one or more processors, a summary including at least one segment based on respective score, and providing, to a user device, the electronic document including one or more summaries.

Abstract: A curator captures input data corresponding to service tasks from an external source. Further, a browser extension collects intermediate service delivery data for the service tasks from the external source. Subsequently, a learner stores the input data and the intermediate service delivery data as training data. Then, a receiver receives a service request from a client. The service request is indicative of a service task to be performed and information associated with the service task. Further, an advisor processes the service request to generate an intermediate service response. Thereafter, the advisor determines a confidence level associated with the intermediate service response and ascertains whether the confidence level associated with service response is below pre-determined threshold level. If the confidence level is below a pre-determined threshold level, the advisor automatically generates a final service response corresponding to service request based on training data.

Abstract: Techniques are described for routing service requests in a computer-implemented service environment. A received service request may be initially analyzed to determine a priority of the request. In some implementations, one or more actions may be automatically performed to provide an initial response to the requester. The text of the request may be analyzed to automatically determine a category of the request. In some implementations, a classification engine may determine the category of the request through use of a classification model that has been trained using one or more machine learning (ML) techniques and/or that employs Natural Language Processing (NLP). Based on the category, the request may be routed to agent(s) for handling. Routing may include generating a ticket that includes the request, the category, the priority, and/or other information, and the ticket may be provided to the appropriate agent(s) through a ticketing service.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for predictive modeling for unintended outcomes are disclosed. In one aspect, a method includes the actions of accessing an order history that, for each of one or more past orders, indicates (i) one or more order details associated with the order, and (ii) a fulfillment outcome associated with the order. The actions further include selecting one or more particular past orders that are associated with the particular unintended order fulfillment outcome. The actions further include generating a predictive model. The actions further include receiving one or more order details associated with a subsequently received order. The actions further include providing the one or more order details as input to the predictive model. The actions further include, identifying a remedial action. The actions further include providing data indicating the remedial action.

Abstract: In the pharmaceutical research and development process, it may be necessary to process large amounts of medical records or clinical literature, to ensure safety of patients consuming a drug. A pharmacovigilance system may assist in this process by efficiently and automatically processing medical records to extract information and relationships contained therein and may also form a preliminary assessment regarding a medical or clinical judgment. The pharmacovigilance system may automatically generate reports based on this information, which may be validated by trained clinicians and medical experts.