Hewlett-packard Development Company, L.p.

Hewlett Packard Enterprise Development LP

Conduent Business Services, LLC

Sandya Srivilliputtur Mannarswamy

Sujoy Saraswati

Shourya Roy

Robert Hundt

Prakash Sathyanath Raghavendra

Anoop Kumar Raveendran

Manjira Sinha

Anand Andaneppa Ganjihal

Virendra Kumar Mehta

Santigopal Mondal

Anirban Sen

Douglas L Voigt

Suparna Bhattacharya

Dhruva Ranjan Chakrabarti

Shin-Ming Liu

Simarjot Kaur

Mark Evan Mills

Sungdo Moon

Suresh Soundararajan

Jini Susan George

Kishore Kumar M

Sakshi Agarwal

Luis A Lozano

Hariharan Sandanagobalane

Easwaran Raman

Sachin Mittal

An autonomous automated intent mining system extracts intents from stored conversations in a text-based chat corpus for use in an AI-driven dialog system. The intent mining system includes an intelligent pre-processor for extracting the conversation data from a computer memory and operating on the conversation data to identify relevant utterances. The system additionally includes a representation processor for operating on the identified relevant utterances to create a numerical representation. The system also includes a clustering processor for generating clusters of utterances and classifying each of the identified relevant utterances into a cluster based on the numerical representation of each utterance. Each generated cluster represents a unique intent for use in the AI-driven dialog system.

Method and system for automated autonomous intent mining

Additional background information is used with a trained neural network based model to help classify whether text is a subtly and/or ambiguously offensive. This additional background information can come from different sources such as the article on which the comment was made, world knowledge about the external entities (e.g., Wikipedia, Urban Dictionary), phrases referenced in the text being classified, and, the context of the previous comments/text in the thread. The background information is retrieved based on key entities (e.g., people, places things) and/or key phrases in the comment. Sentence matrix encodings are built for both the comment and the background information. The background information encoding is used to condition the comment encoding. The background information encoding, and the conditioned comment encoding are fed to a trained multi-level perceptron to classify the comment as hate speech or non-hate speech.

Neural network architecture for subtle hate speech detection

Storage space may be allocated from a non-reserved zone of a file system when the file system is not undergoing an online file system check. When the file system is undergoing an online file system check, storage space is allocated from a soft-reserved zone.

Online file system check

Relevance optimized representative content associated with a data storage system is disclosed. One example is a system including a data summarization module, a clustering module, and a representative content selection module. The data summarization module associates, via a processor, each data object in a storage system with a derived data object. The clustering module determines clusters of similar data objects based on a similarity between associated derived data objects, and selects a representative data object for each determined cluster. The representative content selection module selects representative content associated with the storage system, where the representative content is based on the data objects, the derived data objects, and the representative data objects, and relevance optimizes of the selected representative content to an analytics application.

Relevance optimized representative content associated with a data storage system

Method, system, and apparatus for automatic stance classification. Propositions can be collected that are relevant to a query. A classifier can classify the stance of each proposition based on whether the proposition supports the query, opposes the query, or is neutral with respect to the query in order to thereafter provide substantive data for decision making based on and extracted from the query. The stance can be classified based on, for example, an SVM-SC (SVM Based Stance Classification) approach and/or an NN-SC (Neural Network Stance Classification Approach).

Stance classification of multi-perspective consumer health information

Dynamic updating of metadata associated with a file system is described herein. In accordance with an example, the updating includes creating a shadow file system metadata (FSM) index map in a shadow data block. The shadow data block is associated with a metadata storage of the file system and the metadata storage stores metadata files associated with the file system. The shadow FSM index map is associated with the metadata storage through an index map of the metadata storage, the index map being indicative of the metadata files in the metadata storage. The shadow FSM index map is updated to dynamically update the metadata associated with the file system, in the metadata storage.

Dynamically updating metadata

Example implementations relate to a storage system. An example implementation includes a storage system including a controller coupled to a storage medium. The controller is to receive block priority information from a file system via a communication agent. The block priority information and the file system are associated with the storage medium. Based on the block priority information, the controller is to selectively perform an operation on the storage medium. The operation includes a reconstruction operation, a data scrubbing operation, and a data migration operation.

Block priority information

A method and system for performing semantic analysis for electronic communication using a lexicon is provided. A neural network model is trained with a plurality of annotated text strings, the annotations comprising characteristic tuples that indicate characteristics for the text strings. An unannotated text string is received that comprises a plurality of words from a user. A characteristic matrix for the received text string is generated using a lexicon. The determined characteristic matrix is input into the trained neural network. And a characteristic tuple that indicates a characteristic for the received text string is received as output from the trained neural network.

Performing semantic analyses of user-generated text content using a lexicon

A method, and a corresponding system, for dynamically updating software while the software is running by automatically dividing a patch into a plurality of micro-updates using compiler analysis. The method includes providing the patch which includes updates to several variable units of the software, including functions, type definitions, and data stores. Next, an interference graph of the patch is generated by creating a node corresponding to each variable unit of the patch and creating connected components by adding edges connecting variable units having an impact expression with a non-zero intersection. The patch is divided into the micro-updates, where each micro-update corresponds to a connected component. The micro-updates are then applied to the software when the variable units of the micro-update are at a safepoint, and at least two of the micro-updates are applied at different times while the software is running.

Dynamic software updates

Systems, methods, and computer-readable and executable instructions are provided for managing shared memory. A method for managing shared memory can include statically assigning a first number of locks to the shared memory during compile-time and dynamically assigning a second number of locks to the shared memory during runtime.

Managing shared memory

One embodiment relates to a computer-implemented method of compiling a software program into an executable binary file, including determining a data layout in the binary file and a data layout in the executable's dynamically allocated memory. The method includes taking into account data types of data as a factor in determining a data layout for the binary file and for the executable's dynamically allocated memory, wherein the data types include a floating-point data type and a non-floating-point data type. Other embodiments, aspects and features are also disclosed herein.

Data layout using data type information

A system and method for improving run-time performance of applications with multithreaded and single threaded routines that are linked with libpthreads library is disclosed. In one embodiment, a method for running a mixed ST/MT application program linked with libpthreads library including creating an interceptor library containing pthread application programming interface (pthread_API) call interceptors and loading the interceptor library into the mixed ST/MT application program, and running the mixed ST/MT application program by using light weight (LW) and heavy weight (HW) synchronization routines based on determining switchovers between ST and MT program modes, respectively, during run-time using the interceptor library.

System and method for improving run-time performance of applications with multithreaded and single threaded routines

A method and system for data centric heap profiling is disclosed. In one embodiment, a method, implemented in a computing device, for data centric heap profiling includes generating a type table for data structure types in source code using a compiler of the computing device. The method also includes identifying each heap allocation site and a corresponding data structure type in the source code using the compiler. The method further includes generating a data centric view of a heap o fan application compiled from the source code based on the each heap allocation site and the corresponding data structure type using a debugger of the computing device when a snapshot of the heap is requested during an execution of the application.

Method and system for data centric heap profiling

A technique for reducing non-local access, in dynamically generated code that resides in a code buffer of a non-uniform memory access computer system including multiple nodes, for improving overall performance of dynamic optimization systems. In one example embodiment, this is accomplished by partitioning the code buffer into multiple smaller code buffers and assigning each of the multiple smaller code buffers to one of the multiple nodes. Statically determining which methods in the generated code are executed by a thread and then to place those methods in associated one of the multiple smaller code buffers to reduce memory latencies introduced by non-local accesses.

Growing community of inventors

Karnataka, India

Sandya Srivilliputtur Mannarswamy