Abstract: For confidence weighting relationships between complex entities in unstructured data an expressed relationship between a subset of a set of tokens is extracted from a knowledge graph corresponding to the unstructured data. The knowledge graph includes the set of tokens. The tokens in the subset of tokens are related in the expressed relationship by a set of predicates. A number of occurrences of the set of predicates in the unstructured data is determined. A number of occurrences of the expressed relationship in the unstructured data is determined. Using the number of occurrences of the set of predicates and the number of occurrences of the expressed relationship, a confidence value is computed and assigned to the expressed relationship.

Abstract: The present description is directed towards systems and methods for directing a user request for content over a network to a given content server on the basis of one or more rules. Methods and systems implemented in accordance with the present description comprise receiving a request for content form a user, the request for content including a profile of the user identifying one or more characteristics associated with the user. One or more rules are retrieved for identifying a content server to which a request for content is to be delivered, the one or more rules including at least one of business rules, network rules, and user profile rules. The one or more retrieved rules are applied to the request for content to identify a content server to which the request for content is to be delivered and the request for content is delivered to the identified content server.

Abstract: Machine-learning based detection (MLD) profiles can be used to identify sensitive information in documents. The MLD profile can be used to generate a confidence value for the document that expresses the degree of confidence with which the MLD profile can classify the document as sensitive or not. In one embodiment, a data loss prevention system provides or suggests a confidence level threshold to a user of the data loss prevention system by providing a confidence level threshold for the MLD profile to the user, the confidence level threshold to be used as the boundary between sensitive data and non-sensitive data. In one embodiment the provided confidence level threshold is determined by scanning a random data set using the MLD profile.

Abstract: Aspects of the present disclosure are directed toward evaluating an answer sequence. Aspects are directed toward receiving a set of answer sequences including a first answer sequence. The first answer sequence may have a first set of answers. Aspects are also directed toward identifying a set of scores coupled with the first set of answers. Aspects are also directed toward determining, based on a subject matter corresponding to the first answer sequence, a set of evaluation rules. Aspects are also directed toward generating, based on the set of scores and the set of evaluation rules, a sequence evaluation score for the first answer sequence.

Abstract: Briefly, embodiments of methods and/or systems of providing relevant and diverse recommendations are disclosed. For one embodiment, as an example, a system may extract structured and/or semi-structured parameters from web resources obtained from interaction logs comprising records of browsing sessions. Content from extracted parameters may be compared, using an ontology, to find relationships among web resources and query resources.

Abstract: A computing device can obtain usage data associated with the device. The usage data indicate how the computing device operates or how the device is used. The device can analyze the usage data to recognize usage patterns. The usage patterns can correspond to recurring actions or tasks initiated by the user using the device, such as actions or tasks initiated due to the user's habits and/or routines. Based on the usage patterns, the device can determine a task that has a sufficient likelihood of being performed using the device within a specified or determined time (e.g., 5 minutes from now, one year from now, etc.). The device can provide information (e.g., recommendations) associated with the task, and likely relevant to the user. The user can use the provided information to perform the task, thereby increasing the ease of access or efficiency associated with performing the task.

Abstract: A system and method are disclosed for classifying a record of data when the record of data does not contain class membership information. In one example, summary data is utilized with predictive modeling to predict the age range of a user of a line of service. A prior probability that a user within the summary data belongs to a respective class, a first conditional probability of a first predictor within the summary data and a second conditional probability of a second predictor within the summary data are calculated. For each record of data, a probability of class membership is calculated based on the calculated prior probability and the first and second conditional probabilities.

Abstract: An approach is provided to train a question answering (QA) system using answer keys based on forum content. In the approach, a question is selected from a post in a threaded discussion. An answer to the selected question is automatically identified from crowd-based sources, with the identified answer having a confidence level greater than a threshold. An answer key is built using the selected question and the identified answer. The QA system is automatically trained using the answer key.

Abstract: A method includes receiving a set of parameters for a given project and generating, using information from a knowledge database, a plurality of combinations of group members based at least in part on the set of parameters. The method also includes evaluating a set of metrics for each of the combinations of group members, the set of metrics comprising at least one novelty metric and at least one collective intelligence metric. The method further includes generating one or more strategy matrices for each of at least a subset of the combinations of group members using information from the knowledge database, evaluating the combinations of group members in the subset using the strategy matrices to determine respective predicted success values, and selecting a given one of the combinations of group members for the given project based at least in part on the sets of metrics and predicted success values.

Abstract: A method processes a signal represented as a graph by first determining a graph spectral transform based on the graph. In a spectral domain, parameters of a graph filter are estimated using a training data set of unenhanced and corresponding enhanced signals. The graph filter is derived based on the graph spectral transform and the estimated graph filter parameters. Then, the signal is processed using the graph filter to produce an output signal. The processing can enhance signals such as images by denoising or interpolating missing samples.

Abstract: Natural language management of online social network connections may comprise receiving natural language data associated with a user's social network interactions. The natural language data associated with a user's social network interactions is analyzed and features used in the user's social network interactions are determined based on the analysis. The contexts in which the features are used in the user's social network interactions may be also determined. Point values to the features used in the user's social network interactions are assigned. A fingerprint of the user is created at least based on the features and the point values. The created fingerprint may be compared with information associated with online communities, and based on the comparison, one or more of the online communities may be recommended as user's potential social network connections.

Abstract: A method optimizes a predictive computation model efficiently. The method includes (i) selecting model parameters that are expected to take real values within a one-sided predetermined range; and (ii) iteratively: (a) receiving a set of input values; (b) executing the computation model based on the input values; (c) updating the values of the model parameters to minimize a loss function; and (d) examining each of the model parameters, such that, when the examined model parameter attains or moves past a value that is idempotent to the computation model, removing the model parameter from the computation model. In one embodiment, the predetermined range is either the range between a predetermined positive real value and positive infinity or the range between a predetermined negative real value and negative infinity. The predetermined positive real value or the predetermined negative real value may be an idempotent value to the model computation.

Abstract: Computer-implemented methods and systems are disclosed for making a recommending providing a post on a social media forum. One exemplary embodiment involves utilizing machine-learning techniques to produce a model capable of determining optimal post recommendations from various posting factors. The model may be produced from historical post information regarding various posts made by, for instance, marketers on a social media forum and corresponding community interest responses to the posts made by the community of users associated with the social media forum. The model may be provided to a recommendation engine.

Abstract: A method for diagnosis assistance exploits similarity between a new medical case and existing medical cases and experts when embedded in a common embedding space. Different types of queries are provided for, including a query-by-cases and a query-by-experts. These may be associated with different cost structures that encourage the requester to use the query-by-cases first and seek expert assistance if this proves unsuccessful. Depending on whether the query-by-cases or query-by-experts is requested, a subset of the existing cases or experts is identified based on the similarity of their representations, in the embedding space, with a representation of the new case in the embedding space. There may then be provision for communicating the new case to a selected one or more of the subset of experts for the expert to attempt to provide a diagnosis.

Abstract: Methods, systems, and articles of manufacture for annotating of an image are disclosed. These include scoring the image using a plurality of trained classifiers, wherein each of the trained classifiers corresponds to at least one of a plurality of image groups clustered based upon image similarity, and wherein each image group is associated with a set of weighted labels; selecting one or more of the image groups based upon the scoring; aggregating one or more sets of weighted labels associated with the selected one or more image groups; and annotating the image using the aggregated one or more sets of weighted labels.

Abstract: Disclosed is a novel system, and method to evaluate a prediction of a possibly unknown outcome out of a plurality of predictions of that outcome. The method begins with accessing a particular prediction of an outcome out of a plurality of predictions of that outcome in which the outcome may be unknown. Next, a subsample of the plurality of predictions of the outcome is accessed. The subsample can possibly include the particular prediction. A consensus prediction of the outcome based on the subsample of the plurality of predictions is determined. A proximity of the particular prediction to the consensus prediction is determined. Each prediction is ranked out of the plurality of predictions in an order of a closest in proximity to the consensus prediction to a farthest in proximity to the consensus prediction.

Abstract: A method for predicting a patient's outcome variable, such as a probability of mortality/recovery, includes accessing at least one of a plurality past patients' data fields including physiological and/or laboratory data, and a time of stay indicating how long each patient had been under care at the time. An outcome variable estimation algorithm is generated by data mining from a plurality of past patients' physiological and/or laboratory data, corresponding time of stay, and associated outcome variables. A current patient's outcome variable is determined from the current patient's physiological and/or laboratory data, the current time of stay, and the outcome variable estimation algorithm.

Abstract: A parallel convolutional neural network is provided. The CNN is implemented by a plurality of convolutional neural networks each on a respective processing node. Each CNN has a plurality of layers. A subset of the layers are interconnected between processing nodes such that activations are fed forward across nodes. The remaining subset is not so interconnected.

Abstract: A method for detecting abnormal behavior of users is disclosed. Processors identify from a log of user activity, a first number of actions performed by a user over a first time period that match a pattern of user activity for a task associated with one or more roles of the users. Processors also identify from the log of user activity, a second number of actions performed by the user over a second time period that match the pattern of user activity. Processors calculate an amount of deviation between the first number of actions and the second number of actions. The deviation identifies a difference between amounts of time spent in the one or more roles. Processors then determine whether the amount of deviation between the first number of actions and the second number of actions exceeds a threshold for abnormal behavior.

Abstract: A computer-implemented method can include receiving, at a computing device having one or more processors, questions and answers, each question having one or more answers, and each question and each answer being associated with a particular user. The method can include receiving, at the computing device, evaluations of the answers from users. The method can include identifying, at the computing device, at least one of textual and contextual features for each answer to obtain sets of features. The method can include generating, at the computing device, a user preference graph indicating relationships between users associated with at least one of the questions, the answers, and the evaluations. The method can also include detecting, at the computing device, whether each specific answer is a deceptive answer based on its set of features and the user preference graph.