Abstract: A method, system and apparatus for selecting, displaying, managing, tracking tagging and transferring access to specific content of Internet web pages and other sources, permitting the user or recipient to navigate quickly to a source of a content item in the custom selection, and posting content to web logs. A method and system of assisted content selection provides temporary borders around content items on mouseover; of Dynamic Content Display provides updating of selected content items; of Content Marks allows users to tag specific content items in web pages with a keyword or keywords, store a definition of the content item along with the keyword(s), and then search the stored definitions using search keywords; and a method of Multiple Clip Mode allows users to select specific content items from disparate source web pages and then tag, save, search, retrieve and print the selection.

Abstract: The technology disclosed relates to an end-to-end neural network for question answering, referred to herein as “dynamic coattention network (DCN)”. Roughly described, the DCN includes an encoder neural network and a coattentive encoder that capture the interactions between a question and a document in a so-called “coattention encoding”. The DCN also includes a decoder neural network and highway maxout networks that process the coattention encoding to estimate start and end positions of a phrase in the document that responds to the question.

Abstract: The technology disclosed provides a quasi-recurrent neural network (QRNN) encoder-decoder model that alternates convolutional layers, which apply in parallel across timesteps, and minimalist recurrent pooling layers that apply in parallel across feature dimensions.

Abstract: Disclosed are examples of systems, apparatus, methods and computer program products for providing user-customizable permissions governing user access to computing resources in a computing system. For example, a database storing data objects identifying permissions of users, sets of the permissions, and users can be maintained. One or more fields can be displayed on a display of a user device. The one or more fields can be configured to receive input to perform one or more operations. Input can be received from a user via the user interface on the display of the user device. One or more operations can be performed. The database can be updated such that the data objects identify a custom permission or a permission set to which a custom permission has been assigned.

Abstract: Disclosed are some examples of systems, apparatus, methods and storage media for creating groups in a social networking database system, and more specifically, to creating groups based on network feed items. In some implementations, a database system is capable of maintaining a database including data associated with a plurality of users and groups to which the users can be subscribed. The system is configurable to provide a feed for display to a first user, and to receive input entered in a publication field by the first user. The system is configurable to create a feed item for display to the first user and to at least one second user based on the received input. The system is configurable to receive second input associated with the feed item from the second user. The system is additionally configurable to provide a selectable user interface (UI) element for display to the first user.

Abstract: In accordance with embodiments, there are provided mechanisms and methods for query optimization in a database system. These mechanisms and methods for query optimization in a database system can enable embodiments to optimize OR expression filters referencing different logical tables. The ability of embodiments to optimize OR expression filters referencing different logical tables can enable optimization that is dynamic and specific to the particular tenant for whom the query is run and improve the performance and efficiency of the database system in response to query requests.

Abstract: The technology disclosed relates to maintaining throughput of a stream processing framework while increasing processing load. In particular, it relates to defining a container over at least one worker node that has a plurality workers, with one worker utilizing a whole core within a worker node, and queuing data from one or more incoming near real-time (NRT) data streams in multiple pipelines that run in the container and have connections to at least one common resource external to the container. It further relates to concurrently executing the pipelines at a number of workers as batches, and limiting simultaneous connections to the common resource to the number of workers by providing a shared connection to a set of batches running on a same worker regardless of the pipelines to which the batches in the set belong.

Abstract: A method for building and managing send jobs with restricted information, the method comprising constructing at least one email with at least one reference to a restricted information and injecting each of the at least one emails to one or more send centers, wherein each of the one or more send centers is authorized to receive the restricted information.

Abstract: The technology disclosed relates to automatic generation of tuples from a record set for outlier analysis. Applying this new technology, user need not specify which 1-tuples to combine into n-tuples. The tuples are generated from structured records organized into features (that also could be fields, objects or attributes.) Tuples are generated from combinations of feature values in the records. Thresholding is applied to manage the number of tuples generated. The technology disclosed further relates to indexing and searching high dimensional tuple spaces in a computer-implemented system.

Abstract: An encryption platform may identify language characters in database fields and determine which of multiple encoding schemes more efficiently encodes the language characters. The encryption platform may use the selected encoding scheme to encode plaintext into a byte array and then encrypt the byte array into ciphertext. The encryption platform may assign encoding flags to the ciphertext identifying the encoding scheme used for encoding the plaintext. The encryption platform then may use the encoding flag to determine which encoding schemes to use for decoding the data back into plaintext. Using the different encoding schemes allows encrypted data to fit better into database storage structures. The encryption platform also may enable users to select different fields for different selectable objects for encoding and encrypting.

Abstract: Systems and methods are described to validate user connections to one or more application servers within a multi-tenant application system. A domain-level cookie at the client identifies any active connections for that client. As the client requests a connection to a particular application, the cookie is provided to a validation server that determines if any previously-established sessions with the multi-tenant system exist, and/or if such sessions remain active. If an active session already exists, then the client can be redirected to a particular server to continue the previously-established session. If no valid prior sessions are available, then the client can be validated and a new connection to an appropriate server can be established, as appropriate.

Abstract: The technology disclosed provides a so-called “joint many-task neural network model” to solve a variety of increasingly complex natural language processing (NLP) tasks using growing depth of layers in a single end-to-end model. The model is successively trained by considering linguistic hierarchies, directly connecting word representations to all model layers, explicitly using predictions in lower tasks, and applying a so-called “successive regularization” technique to prevent catastrophic forgetting. Three examples of lower level model layers are part-of-speech (POS) tagging layer, chunking layer, and dependency parsing layer. Two examples of higher level model layers are semantic relatedness layer and textual entailment layer. The model achieves the state-of-the-art results on chunking, dependency parsing, semantic relatedness and textual entailment.

Abstract: The technology disclosed provides a so-called “joint many-task neural network model” to solve a variety of increasingly complex natural language processing (NLP) tasks using growing depth of layers in a single end-to-end model. The model is successively trained by considering linguistic hierarchies, directly connecting word representations to all model layers, explicitly using predictions in lower tasks, and applying a so-called “successive regularization” technique to prevent catastrophic forgetting. Three examples of lower level model layers are part-of-speech (POS) tagging layer, chunking layer, and dependency parsing layer. Two examples of higher level model layers are semantic relatedness layer and textual entailment layer. The model achieves the state-of-the-art results on chunking, dependency parsing, semantic relatedness and textual entailment.

Abstract: The disclosed systems and methods include pre-calculation, per object, of object feature bin values, for identifying close matches between objects, such as text documents, that have numerous weighted features, such as specific-length word sequences. Predetermined feature weights get scaled with two or more selected adjacent scaling factors, and randomly rounded. The expanded set of weighted features of an object gets min-hashed into a predetermined number of feature bins. For each feature that qualifies to be inserted by min-hashing into a particular feature bin, and across successive feature bins, the expanded set of weighted features get min-hashed and circularly smeared into the predetermined number of feature bins. Completed pre-calculated sets of feature bin values for each scaling of the object, together with the scaling factor, are stored for use in comparing sampled features of the object with sampled features of other objects by calculating an estimated Jaccard similarity index.

Abstract: The technology disclosed provides a so-called “joint many-task neural network model” to solve a variety of increasingly complex natural language processing (NLP) tasks using growing depth of layers in a single end-to-end model. The model is successively trained by considering linguistic hierarchies, directly connecting word representations to all model layers, explicitly using predictions in lower tasks, and applying a so-called “successive regularization” technique to prevent catastrophic forgetting. Three examples of lower level model layers are part-of-speech (POS) tagging layer, chunking layer, and dependency parsing layer. Two examples of higher level model layers are semantic relatedness layer and textual entailment layer. The model achieves the state-of-the-art results on chunking, dependency parsing, semantic relatedness and textual entailment.

Abstract: The disclosed systems and methods include pre-calculation, per object, of object feature bin values, for identifying close matches between objects, such as text documents, that have numerous weighted features, such as specific-length word sequences. Predetermined feature weights get scaled with two or more selected adjacent scaling factors, and randomly rounded. The expanded set of weighted features of an object gets min-hashed into a predetermined number of feature bins. For each feature that qualifies to be inserted by min-hashing into a particular feature bin, and across successive feature bins, the expanded set of weighted features get min-hashed and circularly smeared into the predetermined number of feature bins. Completed pre-calculated sets of feature bin values for each scaling of the object, together with the scaling factor, are stored for use in comparing sampled features of the object with sampled features of other objects by calculating an estimated Jaccard similarity index.

Abstract: A computer implemented method includes receiving, at the browser application, a first request for web resources based on user selection of a first hyperlink outside of the web application; determining if the first hyperlink is associated with a record accessible by the web application from the database system; determining, when the first hyperlink is associated with the record for the web application, if a web application interface for the web application is active in at least one navigation tab of the browser application; and presenting, when the first hyperlink is associated with the record for the web application and the web application interface for the web application is active in the at least one navigation tab, the record within a console tab of the web application interface in the at least one navigation tab.

Abstract: Disclosed are systems, methods, apparatus and computer program products for facilitating text-based interaction by a mobile device with an enterprise database. In some implementations, a first one or more text messages sent from a mobile device during a session is parsed to identify a command and a type of enterprise work record. It is determined that the identified command and record type were previously identified. One or more fields of the record type required by the enterprise database to be populated to carry out the identified command is determined. A request message requesting user input to populate the one or more fields can be sent to the mobile device, and instructions can be sent to the enterprise database to carry out the identified command with respect to the identified type of enterprise work record using the field data.