Abstract: A method of operating a messaging platform, including: obtaining, for a first profile, a first and a second topic of interest, a first intra-profile (IP) weight for the first topic of interest, and a second IP weight for the second topic of interest; obtaining a first plurality of trending entities for the first topic and a first plurality of intra-topic (IT) weights for the first plurality of trending entities; obtaining a second plurality of trending entities for the second topic and a second plurality of IT weights for the second plurality of trending entities; selecting a subset of the first plurality of trending entities and the second plurality of trending entities based on the first IP weight, the second IP weight, the first plurality of IT weights, and the second plurality of IT weights; and sending content associated with the subset for display to a user of the first profile.

Abstract: A three-dimensional interactive/digital locker room system that is developed for and accessible through the Internet, which accesses a database of profile/user data that is displayed in the three-dimensional digital, interactive locker room the “Interactive Locker Room System”. Users in the Interactive Locker Room System can move around in first person movements, chat with users, open lockers to view profile data such as equipment and statistics, search for users, share their profiles with college coaches, and receive training, nutrition, and equipment recommendations. Profile data includes videos, images, and biographical data. The 3D interactive locker room system generates and displays featured athletes and videos, and favorite athletes. The interactive locker room system is incorporated into a web application, mobile application, and a virtual reality application. Coaches are able to manage player and team statistics, manage schedules and attendance, and receive training and nutrition recommendations.

Abstract: A method for bidirectional hyperlink management of a hypertext associated with an on-line media is provided. The method may include searching the on-line media for at least one keyword associated with the hypertext. The method may also include scanning a website associated with the hypertext based on the search of the at least one keyword. The method may further include locating at least one dead-link uniform resource locator (URL) associated with the scanned website. Additionally, the method may include managing the at least one located dead-link based on a set of pre-defined rules associated with the on-line media.

Abstract: A method for determining a crawling schedule is disclosed, the method being executable at a crawling server coupled to a first web resource server and a second web resource server. The method comprises: acquiring a first new web page associated with the first web resource server; acquiring a second new web page associated with the second web resource server; determining a first crawling benefit parameter for the first new web page, the first crawling benefit parameter being based on a predicted popularity parameter and a predicted popularity decay parameter thereof; determining a second crawling benefit parameter for the second new web page, the second crawling benefit parameter being based on a predicted popularity parameter and a predicted popularity decay parameter thereof; based on the first crawling benefit parameter and the second crawling benefit parameter, determining a crawling order for the first new web page and the second new web page.

Abstract: A system, method and computer program product for automatic document classification, including an extraction module configured to extract structural, syntactical and/or semantic information from a document and normalize the extracted information; a machine learning module configured to generate a model representation for automatic document classification based on feature vectors built from the normalized and extracted semantic information for supervised and/or unsupervised clustering or machine learning; and a classification module configured to select a non-classified document from a document collection, and via the extraction module extract normalized structural, syntactical and/or semantic information from the selected document, and generate via the machine learning module a model representation of the selected document based on feature vectors, and match the model representation of the selected document against the machine learning model representation to generate a document category, and/or classificatio

Abstract: Techniques for coordinating the sharing of content among applications, provided that said applications and their users are registered to the same sharing coordinator service (SCS), are described. An embodiment of the present invention can enable an application to define, to the SCS it is registered to, the types of content it is able to receive and the application routes where to receive them. Said definition provides the application users with the ability to easily select content of those types from share sources and send it to specific routes on the application, undergoing an overall personalized sharing experience. Content types can be defined on the SCS and made available to other applications to consume, centralizing the parsing of content and the interchange of information into a secure robust strongly typed system.

Abstract: The present invention provides a model generation method, word weighting method, device, apparatus, and computer storage medium. In one aspect, the embodiments of the present invention obtain other images that are identical to a specified image, and use the specified image and said other images as sample images; and then obtain a text clustering based on the texts of the sample images; and then obtain text features based on the text clustering and obtaining visual features based on the sample images; and carry out machine learning based on the text features and the visual features to generate a first regression model and a first sorting model for word weighting on words in the text of an image. The embodiments of the present invention therefore can improve the accuracy of the word weighting result in an image text, thus improving the accuracy of an image search result.

Abstract: The illustrative embodiments provide for a computer-implemented method of automatically measuring an emotion or a tone in received electronic text. A selected color scheme chosen for the electronic text is received at a processor, the color scheme is selected to indicate an intended emotion or an intended tone intended for future electronic text that is to be received. Actual electronic text is received at the processor. An estimated emotion or an estimated tone for the actual electronic text is measured by the processor. The estimated emotion or the estimated tone is compared by the processor to the intended emotion or the intended tone. Responsive to a mismatch between the estimated emotion or the estimated tone and the intended emotion or the intended tone, an alert is transmitted by the processor to a display device that indicates the mismatch.

Abstract: Method, system, and programs for measuring user engagement with content items. In one example, a query is received. A set of content items related to the query is obtained. A presentation of at least one content item of the set of content items is provided on a user interface. A user activity related to the at least one content item is determined. An amount of time between a time at which the presentation of the at least one content item is provided on the user interface and a time at which the user activity occurred is determined. A score associated with the content item is determined based on the amount of time. Information related to user engagement with the set of content items is generated based on the score.

Abstract: This invention relates to a system to search and play music/video as per user's mood and preferences through web or a user's net-enabled mobile communication device. The user can select his mood from the options available on the screen. The user can further filter/refine his/her mood-based music discovery results on the basis selection of the category/language of song/video; followed by selection of era and tempo of song/video. The software also includes a gamification module which awards the redeemable virtual coins and achievement badges to a user for outdoing others in content discovering, creating play-lists, saving discoveries, sharing content, etc.

Abstract: A method, system and computer program product are disclosed for searching for data. In one embodiment, the invention provides a method comprising identifying a query and a search scope including a set of specified entities; and for each of these entities, estimating a number of documents that would be identified in a search through the entity to answer the query. On the basis of this estimating, a subset of the entities is formed. The query and this subset of entities are sent to a search engine to search the subset of entities to answer the query. In one embodiment, the estimating includes collecting statistical information from queries to build up a historical cache using heuristics or machine learning techniques, wherein the query includes a key word and a scope, and the historical cache contains a maximum number of returned results for an entity given the queries executed.

Abstract: A search engine is enhanced to employ user preferences to customize answer output by, receiving a first plurality of candidate answers and weights to a first search query for a first user from a first stage of a non-semantic search engine routing and processing path; augmenting the weights for the plurality of candidate answers according to a set of preference coefficients according to a first set of preferences for the first user; selecting one or more of the first plurality of candidate answers according to the augmented weights; and producing an output including the selected one or more candidate answers.

Abstract: Electronic content such as an electronic book (e-book) that is referenced in or otherwise related to other content such as website content is identified, and a visual representation of the electronic content is generated. The visual representation is presented in conjunction with the website content, and a reader application for previewing the electronic content is launched via user interaction with the visual representation. Exiting the reader application causes a browser application that is rendering the website content to be automatically returned to a foreground state. A variety of actions are enabled via interaction with the reader application such as purchasing the electronic content, sharing the electronic content, or downloading a sample preview of the electronic content for later consumption.

Abstract: Content pre-render and pre-fetch techniques are described. In one or more implementations, a likelihood is predicted of subsequent navigation to content associated with a network address based at least in part on monitored navigation performed by a computing device to one or more network addresses. A determination is then made whether to cause the computing device to pre-fetch and pre-render the content based at least in part on the prediction.

Abstract: Systems, methods, and non-transitory computer readable media can determine a value of a boosting parameter, wherein the boosting parameter is indicative of a value of feedback to a first account type. Respective scores of a plurality of content items that are candidates for a feed of a user can be determined, wherein the plurality of content items are associated with at least one of the first account type or a second account type. Respective scores of content items associated with the first account type in the plurality of content items can be adjusted based on the value of the boosting parameter. The plurality of content items can be ranked based on the respective scores of the plurality of content items.

Abstract: In one embodiment, a method includes accessing multiple posts of an online social network, each post being associated with a first entity of the online social network. The method also includes generating multiple candidate keywords by extracting one or more n-grams from content of the accessed posts and calculating a rank for each candidate keyword. The method also includes storing one or more candidate keywords in association with the first entity, each stored candidate keyword having a calculated rank higher than a threshold rank and sending, to a second entity of the online social network, one or more suggested queries associated with the first entity for display on the client system, each suggested query comprising one or more of the stored candidate keywords associated with the first entity.

Abstract: In some examples, a facility chooses documents for recommendation to people interested in a distinguished subject. The facility accesses a data resource indicating, for each of a plurality of documents, (1) a subject semantically determined for the document, (2) a number of people who have interacted with the document, and (3) an aggregate level of intensity with which those people have interacted with the document. Based on contents of the data structure, the facility identifies one or more documents (1) for which the distinguished subject has been semantically determined, and (2) with which a relatively large number of people have interacted and/or with which people have interacted with a high aggregate level of intensity. The facility transmits to another computing system data configured to cause the first computing system to present a recommendation for some or all of the identified documents.

Abstract: In one embodiment, a method includes receiving a query associated with a trending topic selected by a user of an online social network from multiple trending topics and rewriting the query into a query command including multiple query constraints. The method also includes identifying one or more posts matching the query command, where each identified post has privacy settings making the post visible to all users of the online social network, and calculating, for each of the identified posts, a score for the post based on one or more post-quality features, where the score is calculated using a machine-learning model that assigns a particular weight to each of the one or more post-quality features. The method also includes sending to the user a commentary module including at least a portion of each of one or more of the identified posts having scores higher than a threshold score.

Abstract: A method for intelligently connecting seekers to advisors and providing recorded or synthesized answers to the seekers using artificial intelligence is provided. Receiving a topic of interest from a seeker or retrieving advisors with expertise in the topic from a database of advisors, filtering and sorting the advisors based on ratings associated with each of the advisors related to the topic using machine learning and a plurality of criteria specified by the seeker, routing and establishing a call between the seeker and advisor, when the advisor is unavailable the call is rescheduled for alternate time windows of the advisor or with an alternate advisor and recording and storing the call for automating the answers to the similar questions, obtaining feedback from the seeker for the advisor, and updating the intelligent call routing system and the rankings of the advisor based on the plurality of criteria.

Abstract: Disclosed are systems and methods for behavioral modeling based on content genre and utilizing results for content recommendation and other network handling and storage of the content. Viewing events with respect to a content item are aggregated. An affinity is calculated based on the viewing events. Additional viewing events occurring during the delivery of the content item and associated with other content items are also selected. A sampling bonus is added to the affinity if these additional viewing events have a duration below a threshold and the other content items share a same genre as the content item.

Abstract: Media suitable for a designated style of use are recommended from among a database of media objects. A prediction engine is trained using a plurality of media object lists, each of the media object lists containing metadata associated with a plurality of media objects, the media object lists corresponding to the designated style of use, and the prediction engine being trained to calculate the likelihood that a media object is suitable for the designated style of use. The prediction engine includes a binomial classification model trained with feature vectors that include behavioral data, acoustic data and cultural data for the plurality of media objects. The trained prediction engine is applied to media objects in the database of media objects so as to calculate likelihoods that the media objects are suitable for the designated style of use. One or more media objects are recommended using the calculated likelihoods.

Abstract: This disclosure describes, at least in part, techniques for identifying referral candidates to provide referrals for applicants for employment vacancies for merchants. For instance, the techniques include systems and methods for identifying referral candidates, such as customers of the merchant, who have connections with an applicant for the employment vacancy for the merchant using various types of information, such as merchant transaction history, employment history, etc. The systems and methods may further include determining a strength of a connection between the referral candidates and the applicants, and use the strength of the connections to select referral candidates for referrals and/or weight the referrals provided. In various examples, the systems and methods provide the referrals, as well as other employment-relevant information for the applicant, to the merchant to help in determining whether to hire an applicant.

Abstract: Users may consume and/or share information through various types of content items. For example, user may post a family photo through a social network, create a running blog through a microblogging service, etc. Because users may be overwhelmed by the amount of available content items, it may be advantageous to recommend content items, such as blogs to follow, to users. Accordingly, inductive matrix completion is used to evaluate user interactions with content items (e.g., a user following a blog), content item features (e.g., text and/or images of a blog is evaluated to identify a topic of the blog), and/or user features (e.g., a user liking or reblogging a blog, user demographics, user interests, etc.) to determine whether to recommend a content item to a user. Additionally, graph proximity is used to recommend content items based upon weights of edges connecting user nodes to content item nodes within a directed graph.

Abstract: A method for processing and routing client communications based on information contained in the communications, the method comprising: receiving client communications in one or more formats selected from verbal, electronic, and physical; converting the received client communications into formatted electronic records; creating system records for the formatted electronic records; identifying clusters from the information in the formatted electronic records; categorizing the identified clusters; identifying clients from the information in the formatted electronic records; saving the system records and the formatted electronic records in data files for the identified clients; identifying business departments charged with handling identified clusters; and routing the system records and formatted electronic records to the identified business departments, wherein the system records and formatted electronic records are added to workflow of the business departments.

Abstract: A system, computer-readable storage medium storing at least one program, and a computer-implemented method for re-ranking ranked search results is presented. Ranked search results satisfying a search query are obtained, where the ranked search results include a first search result corresponding to a first document associated with a first entity and a second search result corresponding to a second document associated with a second entity, and where the first search result is ranked higher than the second search result. The first document and the second document are determined to satisfy a similarity criterion. The second entity is determined to satisfy a predefined authorship differential with respect to the first entity. Responsive to determining that the second entity satisfies the predefined authorship differential with respect to the first entity, the second search result and the first search result in the ranked search results are swapped to produce re-ranked search results.

Abstract: An apparatus comprises a segmenter and a pre-processor. The segmenter segments a first stream of data into P-1 substreams, P being an integer greater than 1. The pre-processor, having N input ports, N being an integer greater than P-1, receives the P-1 substreams at P-1 ports of the N input ports and a second stream at a Pth input port. The pre-processor generates P output streams from the P-1 substreams and the second stream using a P-to-P K-muxing operation. The P output streams are uploaded to a cloud storage via a network. Redundancy of the P output streams recovered by a user from the cloud storage is based on whether the second stream is accessible to the user.

Abstract: Methods and systems for automatically generating a media clip library comprising preferred portions of media assets for a user. In one aspect, the system retrieves a profile of the user from memory. Based on the profile, the system identifies a preferred content type for the user. The system identifies a portion of a media asset matching the preferred content type and generates for display the portion of the media asset. The system receives a user input to interrupt the display of the portion of the media asset. Based on receiving the user input, the system retrieves a media clip library associated with user and updates the media clip library to include the portion of the media asset.

Abstract: A location-aware search assist capability identifies location-aware search query suggestions using location information associated with the location-aware search query suggestions. A user's search query input and location and a location associated with each location-aware search query suggestion candidates may be used to identify a set of search query suggestions for presentation to the user. Location-aware search query suggestion candidates may be ranked in accordance with a closeness of each one's location to the user's location. The ranking may be performed using a score, such as a popularity score associated with each search query suggestion candidate. The location-aware search query suggestion candidates having a location closer to the user's location may be promoted by adjusting each candidate's popularity score upward, and the search query suggestion candidates that are farther away from the user's location may be demoted by adjusting each such candidate's popularity score downward.

Abstract: Methods and apparatus for providing query suggestions to a user based on one or more past queries submitted by the user. Candidate query suggestions responsive to a current query may be identified. A candidate query similarity measure may be determined for a given candidate query suggestion based on matching entities related to the given candidate query suggestion and the one or more past queries. In some implementations the similarity measure may be based on cumulative entity scores of the matching entities that may optionally be based on at least one of inverse popularity of the matching entities, and how recent a past query is. In some implementations a ranking of the candidate query suggestions may be adjusted based on the similarity measure. A given candidate query suggestion may be selected as a query suggestion based on at least one of the adjusted ranking and the candidate query similarity measure.

Abstract: The present disclosure relates to assigning labels based on matching electronic activities to record objects. Electronic activities of one or more data source providers may be accessed. Each electronic activity may be matched with a record object of a system of record of the one or more data source providers. Record objects matching electronic activities may be identified. Values from object field-value pairs of the matching record objects may be extracted. Labels may be selected based on the extracted values. Associations between electronic activities and the selected labels may be stored.

Abstract: Adaptive Sampling. Data comprising pairings of data value with lists of data keys are received. The range of possible values of the data keys is partitioned into unbalanced buckets, with at least two of the unbalanced buckets representing different fractions of the range. Each unbalanced bucket is assigned to a respective processing unit selected from a plurality of processing units. The pairings are processed by the processing units, with each processing unit generating an intermediate result. The intermediate results are combined to generate a comprehensive result. A sampling error is determined by scaling an unbalanced bucket's intermediate result according to its corresponding fraction and comparing the scaled intermediate result to the comprehensive result. An unbalanced bucket having a sampling error less than a sampling error threshold is selected. The selected unbalanced bucket's corresponding fraction is selected as a sampling rate for a second data processing job.

Abstract: The present disclosure relates to a method for updating labels assigned to electronic activities. The method includes accessing electronic activities transmitted or received via electronic accounts associated with data source providers. The node profiles can be updated using the electronic activities. The method includes matching one of the electronic activities to participant node profiles, each corresponds to one of the electronic accounts via which the electronic activity is transmitted or received. The method includes assigning tags to the electronic activity based on the participant node profiles. The method includes detecting a change to the participant node profiles. The method includes determining whether to update the tags assigned to the electronic activity based on the detected change to the participant node profile. The method includes modifying the tag assigned to the electronic activity responsive to determining to update the tag in accordance with the detected change.

Abstract: Disclosed are various embodiments for navigating a workflow using a text messaging interface. A client device executes an application. The application renders the messaging interface on a display. The application also renders within the messaging interface a response to a previous selection of a first user interface element. The application further renders a second user interface element within the messaging interface that, when manipulated, causes the application to remove the response and the second user interface element from the messaging interface and also to rerender the first user interface element within the messaging interface.

Abstract: A system and method of recommending a bundle of content items to a user, including storing a plurality of content items in a computer system, determining a respective co-selection score for each pair of content items among the plurality of content items, the co-selection score indicating a probability that a given pair of content items among the plurality of content items will both be downloaded by a user of the computer system, and outputting, to a first user, a plurality of content items comprising a sub-set of the plurality of content items.

Abstract: The present disclosure is directed to a technique for reduction of latency in network traffic data transmissions. The system parses an online document to determine a messaging identifier used to communicate over an asynchronous network-based communication channel with a content provider device. The system assembles a first instance of the online content item with the messaging identifier. An intermediary appliance located on the asynchronous network-based communication channel in between the first computing device and the content provider device intercepts data packets including the electronic message The system determines a quality of the asynchronous network-based communication channel. The system blocks insertion of the messaging identifier in a second instance of the online content item prior to transmission of the second instance of the online content item to a second computing device to reduce latency by preventing additional messages from being sent to the messaging identifier.

Abstract: A method includes obtaining a graph representative of a given network, sampling the graph a given number of times to estimate a level of noisiness for one or more edges in the graph, and annotating the one or more edges of the graph with the respective level of noisiness.

Abstract: Reductions in latencies and improvements in computational efficiency when analyzing data stored in a relational graph by integrating analytical capabilities into graph queries. Instead of a user having to run a graph query and then perform analytics on the resulting subgraph via separate requests, the user is enabled to run analytics at the time the graph query is run via a single request to the database maintaining the relationship graph, which improves the computationally efficiency of analyzing relational graphs and thereby improves the functionality of the computing devices hosting the relational graphs and running the queries and analytics.

Abstract: According to examples, software integration testing with an unstructured database may include retrieving a configuration file stored in memory, and parsing the configuration file to identify configuration details of an unstructured database. A connection may be established between an integration testing tool and the unstructured database based on the configuration details. Software integration testing with an unstructured database may further include identifying a transaction file specifying a database operation to be performed by the unstructured database to retrieve data stored in the unstructured database responsive to the application performing a function. A query may be generated based on the database operation. The query may be sent, via an interface, to the unstructured database for execution. Results of the query may be received via the interface. The query results may be compared to validation data to determine whether the function operates in a determined manner.

Abstract: Methods and systems for speech signal processing an interactive speech are described. Digitized audio data comprising a user query from a user is received over a network in association with a user identifier. A protocol associated with the user identifier is accessed. A personalized interaction model associated with the user identifier is accessed. A response is generated using the personalized interaction model and the protocol. The response is audibly reproduced by a voice assistance device.

Abstract: An order (or priority) for reassigning data portions of a database from a first map to a second map can be determined. It should be noted that the first map assigns the data portions to a first configuration of processors for processing the data portions, and the second map assigns the data portions to a second configuration of processors, different than the first configuration, for processing the data portions in a database system and/or environment. The order can be determined by grouping the data in multiple groups and ordering the groups in accordance with a determined priority for reassignment. In doing so, one or more relationships between the data portions and/or cost of use for each one the groups can be considered.

Abstract: A method and system for dynamically providing contextually relevant news on an article. The method includes extracting a list of key terms from the article on a web page. The method also includes filtering a set of posts associated with a key term in the list of key terms. The set of posts is also filtered from one of a plurality of user accounts and a plurality of editor programmed user accounts on one or more networking websites. The method further includes identifying a plurality of uniform resource locators comprised in the set of posts. Further, the method includes extracting the contextually relevant news from the plurality of uniform resource locators. The contextually relevant news is related to the key term. Moreover, the method includes displaying the contextually relevant news as news feed on the web page. The system includes electronic devices, a communication interface, a memory, and a processor.

Abstract: An apparatus and a method thereof for determining a control condition set of a production line. The apparatus divides several historical control condition sets into several groups, wherein the historical control conditions corresponding to the same control factor are the same in each group. For each group, the apparatus calculates a measurement of central tendency according to the historical yield related values in the group. The apparatus decides a subset of the groups. For each group in the subset, the apparatus calculates a degree of variation and a number regarding the different control conditions between the control condition set and the group. The apparatus calculates weight scores. Based on the measurements of central tendency and the weight scores, the apparatus selects one of the groups as a selected group and assigns the historical control conditions of the selected group as the control conditions of the control condition set.

Abstract: In general, the subject matter described in this disclosure can be embodied in methods, systems, and program products for receiving, by a computing system, a first query that was specified by user input at a computing device. The computing system requests, as a result of having determined that one or more datasets of structured data do not have data for answering the first query, a search of one or more datasets of unstructured data that is based on the first query. The computing system requests multiple results to the search of the one or more datasets of unstructured data, wherein the multiple results are each associated with a query relevance score. The computing system provides the multiple results for presentation by the computing device, in a ranked order that is based on the query relevance score for each of the multiple results.

Abstract: Techniques described herein relate to generating new assessment items and updating existing assessment items. Input data may be received corresponding to the addition, removal, or modification of assessment components within assessment items, and may cause immediate generation and validation of corresponding markup language data blocks, thereby allowing for interactive construction and automated encoding of assessment items. Additional techniques described herein relate to determining compatible scoring types for assessment items and generating and embedding markup language data blocks corresponding to assessment item scoring data.

Abstract: A method of determining component dependencies in an enterprise application architecture implemented over an enterprise network, including identifying structures of hosts connected to the network; wherein the structures includes applications executed by the host, obtaining configuration parameters from said hosts, determining dependencies between applications of hosts based on the configuration parameters, receiving a template for identifying dependencies between components in the enterprise network, identify candidate mappings of hosts based on the template, further identifying candidate mappings of host pairs based on the determined dependencies between hosts and the template, analyzing the candidate mappings to calculate a probability score for the candidate mappings, output the candidate mappings having the highest score.

Abstract: In particular embodiments, a method includes receiving, from a client device associated with a first user of an online social network, a text query comprising one or more character strings, identifying one or more data objects that match at least a portion of one or more of the character strings, wherein each of the identified data objects is connected to the first user within the online social network, generating one or more recommended queries that each comprise the character strings of the text query and references to one or more of the identified data objects, and sending, to the client device associated with the first user in response to receiving the text query, one or more of the recommended queries for display to the first user.

Abstract: Provided is a process, including: receiving resume data from a computing device of a job candidate, the resume data including: a plurality of roles the job candidate has occupied, for each of the roles, a plurality of skills used by the job candidate in the respective roles, for each of the skills, a measure of usage of the respective skill in the respective role; combining the measures of usage of the each skill across the plurality of roles to form a skill density score for each of the skills; receiving a request for information about job candidates from a computing device of a recruiter; and sending data indicative of the skill density scores for at least some of the skills to the computing device of the recruiter.

Abstract: Sketch and style based image retrieval in a digital medium environment is described. Initially, a user sketches an object (e.g., with a stylus) to be searched in connection with an image search. Styled images are selected to indicate a desired style of images to be returned by the search. A search request is generated based on the sketch and selected images. Responsive to the request, an image repository is searched to identify images having the desired object and styling. To search the image repository, a neural network is utilized that is capable of recognizing the desired object in images based on visual characteristics of the sketch and independently recognizing the desired styling in images based on visual characteristics of the selected images. This independent recognition allows desired styling to be specified by selecting images having the style but not the desired object. Images having the desired object and styling are returned.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for classifying user behavior as anomalous. One of the methods includes obtaining user behavior data representing behavior of a user in a subject system. An initial model is generated from training data, the initial model having first characteristic features of the training data. A resampling model is generated from the training data and from multiple instances of the first representation for a test time period. A difference between the initial model and the resampling model is computed. The user behavior in the test time period is classified as anomalous based on the difference between the initial model and the resampling model.

Abstract: A data property recognition apparatus, includes a storage unit; a model data acquisition processor acquiring a plurality of model sets of data entries, each data entry individually representing an identified property common to the model set and being of a data type common to the model set; a feature vector generation processor receiving an input set of data entries, recognizing a data type common to the input set of data entries from among a plurality of supported data types, selecting a set of statistical characteristics representing the input set of data entries in dependence upon the recognised data type, generating a value of each of the selected set of statistical characteristics from the input set of data entries, and outputting a feature vector composed of the generated values of statistical characteristics.