Abstract: A computer-implemented method includes receiving a transaction, where the transaction includes a plurality of operations and is applicable to a graph database. The transaction is represented by a transaction graph, which is a dependency graph representing dependencies among the plurality of operations of the transaction. The transaction graph is partitioned, by a computer processor, into two or more transaction subgraphs. Each of the two or more transaction subgraphs includes two or more operations of the transaction, and each of the two or more transaction subgraphs is a dependency graph representing dependencies among the two or more operations of the transaction subgraph. The two or more transaction subgraphs are independent of one another. The two or more transaction subgraphs are applied to the graph database in parallel, where applying each transaction subgraph to the graph database includes applying the two or more operations of the transaction subgraph to the graph database.

Abstract: A pseudo-relevance feedback (PRF) system is disclosed that determines an optimized relevance model for a search query by utilizing a posterior relevance model to estimate the likelihood that an initial set of top-K retrieved documents would be retrieved given the posterior relevance model, re-ranking the top-K documents based on their respective estimates of likelihood of retrieval, determining a rank similarity between the initial ranking of the top-K documents and the re-ranking of the top-K documents, updating one or more model parameters of the posterior relevance model based on the rank similarity, and iteratively performing the above process until the rank similarity is maximized, at which point, the optimized relevance model is obtained.

Abstract: A method includes associating, in a graph including graph nodes connected via of edges, a respective node weight with each of the graph nodes, and organizing the graph nodes into ancestor nodes, each of the ancestor nodes having one or more descendent nodes so that the ancestor and the descendent nodes include all the graph nodes. For a given descendent node, a respective path to one or more of the ancestor nodes is identified, each of the respective paths including one or more edges, and a given ancestor node having a shortest of the identified paths is determined. A respective edge weight is assigned to each of the one or more edges in the shortest path, and, for the given descendent node, a node loss value is calculated based on the node weight and the respective edge weight of the each of the one or more edges.

Abstract: A search query is received from a user. Multiple search results in response to the search query are received from multiple search engines. Each of the search engines provides one or more search results of the multiple search results. One or more features relating to the uniqueness of the multiple search results of each of the search engines are determined. A ranking of the multiple search results is determined based on the one or more features. The multiple search results are displayed to the user based on the ranking.

Abstract: A tool for determining a query answer selection. The tool receives one or more queries. The tool retrieves one or more strong searcher answers and one or more weak searcher answers, wherein the one or more strong searcher answers are provided by one or more strong searcher engines and the one or more weak searcher answers are provided by one or more weak searcher engines. The tool determines one or more pseudo-query answers, wherein determining the one or more pseudo-query answers includes utilizing the one or more strong searcher answers as a query to the one or more weak searcher engines. Responsive to determining one or more quality predictors, the tool determines whether at least one of the one or more quality predictors for the one or more pseudo-query answers exceeds at least one of the one or more quality predictors for the one or more weak searcher answers.

Abstract: The method includes receiving, by one or more processors, an initial query term. The method further includes generating, by one or more processors, an expanded query based on the received initial query term and one or more related terms to the received initial query. The method further includes determining, by one or more processors, weights corresponding to terms in the received initial query term and the generated expanded query term based on a predicted effect on query performance.

Abstract: Machine logic that traces increases (or decreases) in query drift by performing the following steps: retrieving a first response list responding to a first query; retrieving a second response list responding to a second query, wherein the second query is based, at least in part, on a topic related to the first query; determining a set of scores for each of the first response list and the second response list, calculating a normalized query commitment (“NQC”) for each of the first response list and the second response list; and responsive to determining that a ratio representing a relative difference between the NQC for the first response list and the NQC for the second response list is above a predetermined threshold, determining to use the second response list to respond to the first query.

Abstract: A method includes associating, in a graph including graph nodes connected via of edges, a respective node weight with each of the graph nodes, and organizing the graph nodes into ancestor nodes, each of the ancestor nodes having one or more descendent nodes so that the ancestor and the descendent nodes include all the graph nodes. For a given descendent node, a respective path to one or more of the ancestor nodes is identified, each of the respective paths including one or more edges, and a given ancestor node having a shortest of the identified paths is determined. A respective edge weight is assigned to each of the one or more edges in the shortest path, and, for the given descendent node, a node loss value is calculated based on the node weight and the respective edge weight of the each of the one or more edges.

Abstract: A method, including submitting, to a search engine, a first query including, and receiving, in response to the first query, a first list including first results, each of the first results having a respective first ranking. Keywords are derived from the first query, and for each keyword, a respective second query is submitted to the search engine, the respective second query including the first query term and the derived keyword. In response to each of the respective second queries, a respective second list including respective second results is received, each of the respective second results having a second ranking and a corresponding first result, and for each given second result, one or more stability scores are computed based on the second ranking of the given second result and the first ranking of the corresponding first result. The second results are ranked based on their respective one or more stability scores.

Abstract: A computer-implemented method includes receiving a transaction, where the transaction includes a plurality of operations and is applicable to a graph database. The transaction is represented by a transaction graph, which is a dependency graph representing dependencies among the plurality of operations of the transaction. The transaction graph is partitioned, by a computer processor, into two or more transaction subgraphs. Each of the two or more transaction subgraphs includes two or more operations of the transaction, and each of the two or more transaction subgraphs is a dependency graph representing dependencies among the two or more operations of the transaction subgraph. The two or more transaction subgraphs are independent of one another. The two or more transaction subgraphs are applied to the graph database in parallel, where applying each transaction subgraph to the graph database includes applying the two or more operations of the transaction subgraph to the graph database.

Abstract: A method comprising using at least one hardware processor for receiving a request of a user for media, retrieving a first metadata set relating to said media, comparing between: (a) a media usage profile of the user, which profile reflects knowledge of the user, and (b) the first metadata set relating to said media, to identify gaps in the knowledge of the user with respect to said media, and providing to the user information relating to said identified gaps.

Abstract: Methods, computing systems and computer program products implement embodiments of the present invention that include selecting a training dataset including training instances having respective training features, and applying a classifier to the training dataset, thereby generating a training classification that assigns, to each of the training instances, one of a plurality of categories, the classifier having an expected classification. A classification bias is detected in the training classification relative to the expected classification, and in response to the classification bias, a calibration matrix is defined based on the training features, and the classification bias. A production dataset including production instances is selected, and the classifier and the calibration matrix are applied to the production dataset, thereby generating a production classification quantification that assigns, to each of the production instances, one of the plurality of categories.

Abstract: According to some embodiments of the present invention there is provided a computerized method for labeling a cluster of text documents. The method comprises receiving a document cluster and producing automatically multiple document sub-clusters determined by randomly changing some documents. The method applies multiple cluster labeling algorithms on the cluster and on each sub-cluster, to generate ordered lists. The method comprises generating a ranked label list for each cluster labeling algorithm by computing automatically label values, one for each cluster label in the lists of the selected algorithm, and re-ranking the ordered list. The method combines the re-ranked label lists using a label fusing algorithm to produce a fused label list.

Abstract: Methods, computing systems and computer program products implement embodiments of the present invention that include selecting a training dataset including training instances having respective training features, and applying a classifier to the training dataset, thereby generating a training classification that assigns, to each of the training instances, one of a plurality of categories, the classifier having an expected classification. A classification bias is detected in the training classification relative to the expected classification, and in response to the classification bias, a calibration matrix is defined based on the training features, and the classification bias. A production dataset including production instances is selected, and the classifier and the calibration matrix are applied to the production dataset, thereby generating a production classification quantification that assigns, to each of the production instances, one of the plurality of categories.

Abstract: Methods, computing systems and computer program products implement embodiments of the present invention that include selecting a dataset that includes instances, each of the instances having respective features, and determining an expected distribution of the instances among multiple categories. A first classification is generated that includes, for each of the instances based on their respective features, one or more first categories and a corresponding confidence score for each of the one or more categories. One or more of the instances classified into the given category are allocated to each given category, based on their respective confidence scores, using the allocated one or more instances, a second classification is generated.

Abstract: Identifying influencers in a computer network by adjusting influence weights of corresponding participants of a computer network, where the influence weights relate to a topic, where each of the influence weights is adjusted in accordance with a predefined adjustment function, and where the predefined adjustment function uses a) the influence weight of any of the participants that disseminated content via the computer network, where the content relates to the participant whose influence weight is being adjusted, b) a participant topic similarity value of any of the participants that disseminated the content, where the participant topic similarity value relates to the topic, and c) a relationship topic similarity value of any relationship between the participants that disseminated the content and the participant whose influence weight is being adjusted, where the relationship topic similarity value relates to the topic, and then ranking the participants by their influence weights.

Abstract: A method comprising using at least one hardware processor for receiving a request of a user for media, retrieving a first metadata set relating to said media, comparing between: (a) a media usage profile of the user, which profile reflects knowledge of the user, and (b) the first metadata set relating to said media, to identify gaps in the knowledge of the user with respect to said media, and providing to the user information relating to said identified gaps.

Abstract: Method, system, and computer program product for access control for entity search are provided. The method includes: representing entity-relationship data in a conceptual model; representing entities in a search system as documents containing the entity's searchable content and metadata; defining authorization rules for searchers over entities and their relationships; and extending an entity document to include searchable tokens defining the authorization rules. Defining authorization rules may include: identifying query predicate constraints for entity search; and defining searchable tokens as paths for query predicates and permissible searchers; wherein the permissible searchers are permitted access to data based on a query that contains the predicate. Defining authorization rules may further include: defining searchable document files for a free-text predicate with a field name as a token of permissible searchers and the field value as the searchable content.

Abstract: Machines, systems and methods for providing an authentication challenge are provided. The method comprises analyzing data stored in a computing system equipped with a graphical user interface (GUI), wherein the data stored is related to identity and relationships among items that have a profile; and based on the analysis, issuing a challenge to authenticate access to one or more content or services available by way of the computing system, in response to a user interaction with the computing system, wherein the response to the challenge is known to a user who has personal knowledge of the identity and relationships among the items with a profile, and wherein the user successfully authenticates against the challenge by interacting with visual presentations of the items by placing the items in an order that indicates a correct relationship between at least two or more of the items.

Abstract: Method, system, and computer program product are provided for visualization of user sentiment for one or more product features. The method may include: providing one or more product image templates, a product image template having a location representing a product feature; obtaining an aggregated sentiment score for a product feature from user generated content; mapping the aggregated sentiment score to a score visualization on a visualization scale; and representing the location in the product image template relating to the product feature with the score visualization for the aggregated sentiment score to provide a visualization of the product. The method may also include: collecting one or more text expressions from user generated content relating to a product feature; representing one or more text expressions in relation to the product feature in the product image template.