Abstract: A computer identifies data detailing usage of a first set of one or more applications by a first user and a first set of users, and information associated with usage of a second set of one or more applications by the first set of users. The computer determines one or more similarity coefficients between the first user and the first set of users and determines an overall predicted taste score for each content associated with the second set of one or more applications based on the determined one or more similarity coefficients. The computer determines one or more recommendations for one or more contents of the second set of one or more applications based on the determined one or more overall predicted taste scores.

Abstract: A system and method are provided for enabling a user to search for documents that the user has previously viewed on its local machine. The system includes three main components: the desktop integration module, the index module, and the graphical user interface module. The desktop integration module is an application which monitors documents with which the user interacts for predetermined events, and obtains content data and metadata from the monitored documents. The index module indexes the content data and metadata received from the desktop integration module. The graphical user interface module then permits a user to utilize the desktop integration module and index module by allowing a user to search for a document.

Abstract: A calendared event may be visually indicated. A proposed time for the calendared event can be identified and each of a plurality of event participants can be associated with the calendared event. For each of the participants, a local time of the participant that corresponds to the proposed time for the calendared event can be identified and displayed.

Abstract: Methods, products, apparatus, and systems may identify one or more relevant social communities for one or more questions. Additionally, a user-question affinity value between a user and a question may be determined. In addition, a user-community affinity value between the user and each of a plurality of candidate social communities may be determined. Moreover, a question-community affinity value between the question and each of the plurality of candidate social communities may be determined based on the user-question affinity value and the user-community affinity value. The question-community affinity value determination may involve calculating a running average using the user-question affinity value and the user-community affinity value. The question-community affinity value may identify the one or more relevant social communities from the plurality of candidate social communities.

Abstract: A method for training a machine learning model for open domain question answering includes receiving trained classifiers for question answering. The received trained classifiers are used to generate a set of candidate answers to a question. Second trained classifiers are used for scoring each of the candidate answers. The scoring indicates a measure of how well each candidate answer answers the question. Using the second trained classifiers for scoring each of the candidate answers includes comparing each candidate answer to a first ground truth corresponding to the question. A set of top-scoring candidate answers is presented to a human operator who marks each as correct or incorrect. The correct candidate answers are treated as additional ground truths for further training the first trained classifiers.

Abstract: A method for training a machine learning model for open domain question answering includes receiving trained classifiers for question answering. The received trained classifiers are used to generate a set of candidate answers to a question. Second trained classifiers are used for scoring each of the candidate answers. The scoring indicates a measure of how well each candidate answer answers the question. Using the second trained classifiers for scoring each of the candidate answers includes comparing each candidate answer to a first ground truth corresponding to the question. A set of top-scoring candidate answers is presented to a human operator who marks each as correct or incorrect. The correct candidate answers are treated as additional ground truths for further training the first trained classifiers.

Abstract: A set of tags can be identified from a first set of tagged documents in a first repository. A set of tags can be identified from a second set of tagged documents in a second repository. Access to documents in the second repository can be more restrictive than access to documents in the first repository. For each of a subset tags in the first set and/or the second set, a number of steps can occur. A ratio can be determined of tag instances in the second repository compared to tag instances in the first repository. It can be determined whether the ratio exceeds a previously determined threshold. When the threshold is exceeded, an indicator of at least one tagged document associated with the tag can be changed to indicate that the tagged document is likely to contain sensitive content.

Abstract: Methods, products, apparatus, and systems may identify one or more relevant social communities for one or more questions. Additionally, a user-question affinity value between a user and a question may be determined. In addition, a user-community affinity value between the user and each of a plurality of candidate social communities may be determined. Moreover, a question-community affinity value between the question and each of the plurality of candidate social communities may be determined based on the user-question affinity value and the user-community affinity value. The question-community affinity value determination may involve calculating a running average using the user-question affinity value and the user-community affinity value. The question-community affinity value may identify the one or more relevant social communities from the plurality of candidate social communities.

Abstract: Methods, products, apparatus, and systems may identify one or more relevant social communities for one or more questions. Additionally, a user-question affinity value between a user and a question may be determined. In addition, a user-community affinity value between the user and each of a plurality of candidate social communities may be determined. Moreover, a question-community affinity value between the question and each of the plurality of candidate social communities may be determined based on the user-question affinity value and the user-community affinity value. The question-community affinity value determination may involve calculating a running average using the user-question affinity value and the user-community affinity value. The question-community affinity value may identify the one or more relevant social communities from the plurality of candidate social communities.

Abstract: A calendared event may be visually indicated. A proposed time for the calendared event can be identified and each of a plurality of event participants can be associated with the calendared event. For each of the participants, a local time of the participant that corresponds to the proposed time for the calendared event can be identified and displayed.

Abstract: Embodiments of the present invention provide a method, system and computer program product for interest contour computation and management based upon user generated content and associated meta-data. In an embodiment of the invention, an interest contour computation and management method is provided. The method includes crawling content sources disposed about a computer communications network for authored content created by an end user. The method further includes identifying meta data provided for the authored content and adding the meta data to a user interests profile of the end user. The meta-data further can include extracted text from the content. Of note, the method can further include receiving from the end user a specified time period and limiting the addition of the meta data to meta data applied to the authored content during the specified time period.

Abstract: A calendared event may be visually indicated. A proposed time for the calendared event can be identified and each of a plurality of event participants can be associated with the calendared event. For each of the participants, a local time of the participant that corresponds to the proposed time for the calendared event can be identified and displayed.

Abstract: A system and method are provided for enabling a user to search for documents that the user has previously viewed on its local machine. The system includes three main components: the desktop integration module, the index module, and the graphical user interface module. The desktop integration module is an application which monitors documents with which the user interacts for predetermined events, and obtains content data and metadata from the monitored documents. The index module indexes the content data and metadata received from the desktop integration module. The graphical user interface module then permits a user to utilize the desktop integration module and index module by allowing a user to search for a document.

Abstract: A method for processing reviews includes identifying reviews that match a request criterion in a request from a user; filtering the identified reviews using preferences and characteristics of the user; and outputting a compilation of only those reviews filtered according to preference and characteristics of the user.

Abstract: Embodiments of the present invention provide a method, system and computer program product for ranking search results based upon content creation trends. In an embodiment of the invention, a method for ranking search results based upon content creation trends includes first receiving a set of results from a search engine produced in response to a search engine query of one or more search terms. The method additionally includes ranking the results of the result set according to at least one ranking factor. The method yet further includes retrieving trend data for terms associated with either or both of the results and/or the search terms. Finally, the method includes weighting the ranked results in the result set according to the retrieved trend data.

Abstract: Embodiments of the present invention provide a method, system and computer program product for ranking search results based upon content creation trends. In an embodiment of the invention, a method for ranking search results based upon content creation trends includes first receiving a set of results from a search engine produced in response to a search engine query of one or more search terms. The method additionally includes ranking the results of the result set according to at least one ranking factor. The method yet further includes retrieving trend data for terms associated with either or both of the results and/or the search terms. Finally, the method includes weighting the ranked results in the result set according to the retrieved trend data.

Abstract: A system and method are provided for enabling a user to search for documents that the user has previously viewed on its local machine. The system includes three main components: the desktop integration module, the index module, and the graphical user interface module. The desktop integration module is an application which monitors documents with which the user interacts for predetermined events, and obtains content data and metadata from the monitored documents. The index module indexes the content data and metadata received from the desktop integration module. The graphical user interface module then permits a user to utilize the desktop integration module and index module by allowing a user to search for a document.

Abstract: A method is provided for enabling a user to search for documents that the user has previously viewed on its local machine. The method includes maintaining, on a workstation of a user, an index that includes one or more entries for only one or more documents the user has previously accessed; receiving from the user a request to search the index; and presenting a user interface providing a search results listing and one or more graphical visualizations characterizing the search results. In response to an input from the user, usage metrics for a selected documents are displayed.

Abstract: Embodiments of the present invention provide a method, system and computer program product for interest contour computation and management based upon user generated content and associated meta-data. In an embodiment of the invention, an interest contour computation and management method is provided. The method includes crawling content sources disposed about a computer communications network for authored content created by an end user. The method further includes identifying meta data provided for the authored content and adding the meta data to a user interests profile of the end user. The meta-data further can include extracted text from the content. Of note, the method can further include receiving from the end user a specified time period and limiting the addition of the meta data to meta data applied to the authored content during the specified time period.

Abstract: A system is provided for enabling a user to search for documents that the user has previously viewed on its local machine. The system may include three main components: the desktop integration module, the index module, and the graphical user interface module. The desktop integration module is an application which monitors documents with which the user interacts for predetermined events, and obtains content data and metadata from the monitored documents. The index module indexes the content data and metadata received from the desktop integration module. The graphical user interface module then permits a user to utilize the desktop integration module and index module by allowing a user to search for a document.