Abstract: One embodiment is a method that observes a first object within an enterprise and then determines that use of the first object by an enterprise is potentially unauthorized. The method then alters a computer model based on the first object and determines based on the model that a second object within the enterprise is potentially unauthorized.

Abstract: Provided are, among other things, systems, methods and techniques for decryption key management. In one implementation, a decryption key is managed within a computer processing system by (a) creating within the computer system an association between an access token and retrieval information, the access token being a specified function of an identifier for a data object, and the retrieval information including (1) a first entry that corresponds to a value generated by encrypting a decryption key for the data object using a symmetric encryption/decryption key, and (2) a second entry that corresponds to a value generated by encrypting the symmetric encryption/decryption key using an asymmetric public key; and (b) repeating step (a) for a number of different data objects, keeping the symmetric encryption/decryption key identical across repetitions.

Abstract: A method for estimating similarity between two collections of information is described herein. The method includes comparing a first Bloom filter representing a first collection of information and a second Bloom filter representing a second collection of information, and determining a measure of similarity between the first collection of information and the second collection of information based on the comparing.

Abstract: Provided are, among other things, systems, methods, apparatuses and techniques for storing access grants. In one implementation, a blinding factor and access information for accessing a restricted object are obtained; blinded access information is generated for the restricted object based on the access information and the blinding factor; and an anchor node is stored into a data store, with the anchor node being accessible by submission of an identifier, the anchor node at least one of containing or referring to sufficient information to obtain access to the blinding factor and the blinded access information, and the identifier for the anchor node being independent of the blinding factor.

Abstract: Provided are, among other things, systems, methods and techniques for managing associations between keys and values within a computer processing system. In one exemplary implementation, requests to store associations between keys and data values corresponding to the keys are input, and the associations are stored in entry nodes within a data structure represented as a hash-based directed acyclic graph (HDAG). Upon receipt of a data value request and accompanying request key, together with satisfaction of any additional access criterion, a return data value corresponding to the request key automatically is provided, the return data value having been generated based on at least one of the associations that involve the request key.

Abstract: A first distribution of cases over a first group of categories is received. A categorizer trained using a search-and-confirm technique classifies the cases into a second group of categories. A second distribution of the cases over the second group of categories is generated using results of the classifying. The first and second distributions are compared to identify differences between the first and second distributions.

Abstract: A categorizer is trained for plural categories according to a machine-learning algorithm. The categorizer classifies cases in a set of cases into the plural categories. One or more quantification measures regarding cases in the data set are computed based on output from the categorizer, and information is provided to enable ranking of the categories based on the one or more quantification measures. Action is effected to address an issue associated with at least one of the categories based on the provided information.

Abstract: A computer executable method includes receiving word senses represented as semantic positions in a semantic space. The word senses are grouped into clusters based on semantic positions of the word senses in the semantic space. Labels for the respective clusters are provided that correspond to topics of the topic list.

Abstract: Chunks are stored in a data store, where respective collections of chunks form respective files. An index that maps digests of chunks to pages containing information to recreate the chunks is provided, where the index is stored in persistent storage.

Abstract: To provide a data store, nodes representing respective chunks of files are stored in a predefined structure that defines relationships among the nodes, where the files are divided into the chunks. The nodes are collected into plural groups stored in persistent storage, where some of the nodes are collected into a particular one of the groups according to a locality relationship of the some of the nodes.

Abstract: In response to a query relating to a category, identify at least one case. An indication is received regarding whether the at least one case belongs to the category. A categorizer is trained based on the received indication. Cases from a data set are allocated into plural bins. For respective bins, quantification measures are computed for cases belonging to the category based on output from the categorizer. Information is provided regarding a trend of the category over the plural bins based on the computed quantification measures.

Abstract: A partial order among a set of classifiers is constructed, where the partial order indicates which classifiers can be input as features for other classifiers. Based on the partial order, a function based on an output of one classifier is selected as a feature for another classifier.

Abstract: To build a model, an expression related to a task to be performed with respect to a collection of cases is received, where the task is different from identifying features for building the model. A feature is produced from the expression, and a model is constructed based at least in part on the produced feature.

Abstract: To calculate an aggregate of attribute values associated with plural cases, at least one parameter setting that affects a number of cases predicted positive by a classifier is selected. At least one measure pertaining to the plural cases is calculated, where the at least one measure is dependent upon the selected at least one parameter setting. An estimated quantity of the plural cases relating to at least one class is received.

Abstract: A computer executable method includes determining semantic positions in a semantic space of portions of a text. Once the semantic positions of portions of the text have been determined, an action is performed with respect to the text based on the determined semantic positions.

Abstract: A computer executable method includes receiving word senses represented as semantic positions in a semantic space. The word senses are grouped into clusters based on semantic positions of the word senses in the semantic space. Labels for the respective clusters are provided that correspond to topics of the topic list.

Abstract: A computer executable method includes receiving word senses associated with respective definitions, and providing a position-based dictionary that assigns the word senses to respective semantic positions in a semantic space.

Abstract: A system and method for customer-side market segmentation and categorization. This segmentation is done without disclosing sensitive private customer information to the business. A customer downloads a categorization module to a portable device (PDA, wireless cellular phone, etc.) or personal computer. A business defines a decision procedure corresponding to a set of defined customer categories. The business sends their rule set to the customer's device, which uses the rules and a set of stored customer-specific historical and demographic information to determine into which of the business-specific customer categories the customer falls. The categorization module may use any of a variety of methods, such as decision trees, neural networks, Bayesian belief networks, k-nearest neighbor, genetic algorithms, or rule sets. The customer category is sent to the business without other personal data for the business to prepare appropriate promotional material or initiate specific actions.

Abstract: An arrangement for determining a predicted health of a system that would result from an application of a proposed intervention to an existing system. The predicted health is determined by determining a set of modifications involved in the proposed intervention wherein each modification involves one or more components of the existing system. For each modification, a set of component information that pertains to the modification is obtained from a knowledge base. Each set of component information may specify inter-dependencies among the components involved in the modification. For each modification, it is determined whether the inter-dependencies specified in the component information are satisfied. The determination of whether the inter-dependencies are satisfied enables a determination of the predicted health of individual components of the system as well as the predicted health of the overall system.

Abstract: Disclosed herein are systems and methods for combining valuations of multiple evaluators. One disclosed method embodiment comprises (for each evaluator): sorting the choices into an order indicative of preference; determining interval sizes between adjacent choices; obtaining a rank for each interval size; and using interval size ranks to determine an evaluator score for each choice. The evaluator scores can be combined to determine an overall score for each choice.