Abstract: An electronic apparatus for compressing a language model is provided, the electronic apparatus including a storage configured to store a language model which includes an embedding matrix and a softmax matrix generated by a recurrent neural network (RNN) training based on basic data including a plurality of sentences, and a processor configured to convert the embedding matrix into a product of a first projection matrix and a shared matrix, the product of the first projection matrix and the shared matrix having a same size as a size of the embedding matrix, and to convert a transposed matrix of the softmax matrix into a product of a second projection matrix and the shared matrix, the product of the second projection matrix and the shared matrix having a same size as a size of the transposed matrix of the softmax matrix, and to update elements of the first projection matrix, the second projection matrix and the shared matrix by performing the RNN training with respect to the first projection matrix, the second p

Abstract: Client instance data including a plurality of incidents is obtained, each incident including a plurality of fields. A target field and an evaluation field are selected from among the plural fields. The plurality of incidents are grouped into a plurality of clusters based on a degree of a natural language text similarity of respective target fields in the plurality of incidents. A quality value is determined for each of the plurality of clusters based on the degree of the natural language text similarity of respective target fields in grouped incidents of the cluster from among the plurality of incidents, and based on respective evaluation fields. Each of the plurality of clusters is ranked based on the respective quality value of the cluster and a number of the grouped incidents of the cluster. At least one of the ranked plurality of clusters is identified to perform a service management operation.

Abstract: Systems, devices, and methods of the present invention detect argumentation in text. In an example, an application executing on a computing device accesses text comprising fragments. The application creates a discourse tree from the text. The discourse tree includes nodes, each nonterminal node representing a rhetorical relationship between two of the fragments and each terminal node of the nodes of the discourse tree is associated with one of the fragments. The application matches each fragment that has a verb to a verb signature, thereby creating a communicative discourse tree. The application determines whether the communicative discourse tree represents text that includes argumentation by applying a classification model trained to detect argumentation to the communicative discourse tree.

Abstract: A method of translating information using crowdsourcing includes evaluating a metric related to a content to be translated, determining a priority for the content, queuing the content for the crowdsourcing based on the priority determined from the metric, and translating the information from a language to another language using the crowdsourcing.

Abstract: An apparatus is provided with an ambiguous keystroke disambiguation and/or word autocompletion text editor application that uses a common language dictionary. The apparatus is also provided with one or more lexica that contain a vocabulary relating to a specific subject matter. The ambiguous keystroke disambiguation and/or word autocompletion text editor application uses one or more of the lexica in combination with a language dictionary for ambiguous keystroke disambiguation and/or word autocompletion text editing. The user can determine which of the lexica are to be used by the ambiguous keystroke disambiguation and/or word autocompletion text editor application. The user can also determine the priority with which the lexica are used. The lexica can be downloaded to the apparatus from the Internet or transferred from any other device that the apparatus is connected to and the lexica that are stored on the apparatus can be edited by a user.

Abstract: A translation device includes an input unit, a controller, a notification unit, an output unit, and a memory. The input unit inputs an utterance of a speaker to generate utterance data. The controller determines accuracy of a translation result when translated utterance data from the utterance data is obtained. The notification unit notifies the speaker of the determination result of the controller. The output unit outputs a translated utterance according to the translated utterance data. The memory stores dictionary data associating utterance data with translated utterance data. The controller performs translation based on the dictionary data. When the controller determines that the accuracy of the translation result is lower than a predetermined value, the controller controls the output unit not to output the translated utterance, and controls the notification unit to issue a notification regarding the determination result.

Abstract: Embodiments of the inventive system and methods are directed to a computer program that employs a drag-and-drop user interface for managing dialogue states, tracking dialogue context, understanding dialogue utterances, and managing dialogue sessions. Each dialogue element is defined in a “node” that can be dragged and dropped into a canvas of the user interface. An embodiment provides wiring mechanisms to freely link nodes. Dialogue utterances are contained in messages that flow through the wires linking different nodes until exiting the canvas to an end user. An executable image of a conversational agent is then generated by compiling the source code associated with the nodes based on their connections. A conversational agent can be deployed in an electronic device such as a home device, which configured to perform an action in response to a user verbal command or request using the conversational agent deployed therein.

Abstract: A system provides on a computer display screen, a graphical image of an initial parsing as generated by a computerized algorithm. Individual clauses are encased in expandable graphical elements that a human user can manipulate on the display screen using an input device in furtherance of improving the initial parsing. The expandable graphical elements have particular spatial and visual interrelationships in keeping with the initial text parsing, human user interactions with the expandable graphic elements captured so as to convert the human-based interactions with these graphical elements into additional machine-readable text and mark-up that can be usable for other machine-based text processes. The system, in addition to being a parsed text/editor interface, thereby functions as an intuitive, visual conduit for dynamic human-machine text analysis collaboration.

Abstract: Various embodiments described herein facilitate multi-lingual communications. The systems and methods of some embodiments enable multi-lingual communications through different modes of communication including, for example, Internet-based chat, e-mail, text-based mobile phone communications, postings to online forums, postings to online social media services, and the like. Certain embodiments implement communication systems and methods that translate text between two or more languages. Users of the systems and methods may be incentivized to submit corrections for inaccurate or erroneous translations, and may receive a reward for these submissions. Systems and methods for assessing the accuracy of translations are described.

Abstract: Systems, devices, and methods of the present invention uses noisy-robust discourse trees to determine a rhetorical relationship between one or more sentences. In an example, a rhetoric classification application creates a noisy-robust communicative discourse tree. The application accesses a document that includes a first sentence, a second sentence, a third sentence, and a fourth sentence. The application identifies that syntactic parse trees cannot be generated for the first sentence and the second sentence. The application further creates a first communicative discourse tree from the second, third, and fourth sentences and a second communicative discourse tree from the first, third, and fourth sentences. The application aligns the first communicative discourse tree and the second communicative discourse tree and removes any elementary discourse units not corresponding to a relationship that is in common between the first and second communicative discourse trees.

Abstract: Asymmetrical formatting of word spaces according to the uncertainty between words includes an initial filtering process and subsequent text formatting process. An equivocation filter generates a mapping of keys and values (output) from a corpus or word sequence frequency data (input). Text formatting process for asymmetrically adjusts the width of spaces adjacent to keys using the values. The filtering process, which generates a mapping of keys and values can be performed once to analyze a corpus and once generated, the key-value mapping can be used multiple times by a subsequent text processing process.

Abstract: In one embodiment, a computer-implemented method for proactively improving machine translation in real time by processing a translation request includes: receiving the translation request from a mobile device; and either generating or retrieving translation information based on a result of either or both of: determining whether the translation request includes image data suitable for generating a machine translation; and determining whether the translation request includes metadata suitable for generating an estimated translation. In another embodiment, a computer-implemented method for proactively improving machine translation in real time by generating and submitting a translation request includes: capturing image data; evaluating one or more conditions corresponding to the captured image data; generating metadata corresponding to the captured image data; and generating and submitting the translation request to a machine translation platform.

Abstract: Techniques are described for data transformation performed based on a current emotional state of the user who provided input data, the emotional state determined based on biometric data for the user. Sensor(s) may generate biometric data that indicates physiological characteristic(s) of the user, and an emotional state of the user is determined based on the biometric data. Different dictionaries and/or dictionary entries may be used in translation, depending on the emotional state of the sender when the data was input. In some implementations, the emotional state of the sending user may be used to infer or otherwise determine that a translation was incorrect. The input data may be transformed to include information indicating the current emotional state of the sending user when they provided the input data. For example, the output text may be presented in a user interface with an icon and/or other indication of the sender's emotional state.

Abstract: Presented herein are systems and methods for question answering (QA). In embodiments, extractive question answering (QA) is cast as an iterative search problem through the document's structure: select the answer's sentence, start word, and end word. This representation reduces the space of each search step and allows computation to be conditionally allocated to promising search paths. In embodiments, globally normalizing the decision process and back-propagating through beam search makes this representation viable and learning efficient. Various model embodiments, referred to as Globally Normalized Readers (GNR), achieve excellent performance. Also introduced are embodiments of data-augmentation to produce semantically valid examples by aligning named entities to a knowledge base and performing swaps new entities of the same type. This methodology also improved the performance of GNR models and is of independent interest for a variety of natural language processing (NLP) tasks.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a target sequence from an input sequence. In one aspect, a method comprises maintaining a set of current hypotheses, wherein each current hypothesis comprises an input prefix and an output prefix. For each possible combination of input and output prefix length, the method extends any current hypothesis that could reach the possible combination to generate respective extended hypotheses for each such current hypothesis; determines a respective direct score for each extended hypothesis using a direct model; determines a first number of highest-scoring hypotheses according to the direct scores; rescores the first number of highest-scoring hypotheses using a noisy channel model to generate a reduced number of hypotheses; and adds the reduced number of hypotheses to the set of current hypotheses.

Abstract: Provided are systems and methods for creating custom dialog system engines. The system comprises a dialog system interface installed on a first server or a user device and a platform deployed on a second server. The platform is configured to receive dialog system entities and intents associated with a developer profile and associate the dialog system entities with the dialog system intents to form a custom dialog system engine associated with the dialog system interface. The web platform receives a user request from the dialog system interface, activates the custom dialog system engine based on identification, and retrieves the dialog system entities and intents. The user request is processed by applying the dialog system entities and intents to generate a response to the user request. The response is sent to the dialog system interface.

Abstract: Technology is disclosed for snippet pre-translation and dynamic selection of translation systems. Pre-translation uses snippet attributes such as characteristics of a snippet author, snippet topics, snippet context, expected snippet viewers, etc., to predict how many translation requests for the snippet are likely to be received. An appropriate translator can be dynamically selected to produce a translation of a snippet either as a result of the snippet being selected for pre-translation or from another trigger, such as a user requesting a translation of the snippet. Different translators can generate high quality translations after a period of time or other translators can generate lower quality translations earlier. Dynamic selection of translators involves dynamically selecting machine or human translation, e.g., based on a quality of translation that is desired.

Abstract: An agent automation system includes a memory configured to store a natural language understanding (NLU) framework, and a processor configured to perform actions, including: generating a meaning representation from an annotated utterance tree of an utterance, wherein a structure of the meaning representation indicates a syntactic structure of the utterance and one or more subtree vectors of the meaning representation indicate a semantic meaning of one or more intent subtrees of the meaning representation; searching the meaning representation of the utterance against an understanding model to extract intents/entities of the utterance based on the one or more subtree vectors of the meaning representation, wherein the understanding model includes a plurality of meaning representations derived from the intent/entity model; and providing the intents/entities of the utterance to a reasoning agent/behavior engine (RA/BE) of the agent automation system that performs one or more actions in response to the intents/entit

Abstract: With reference to information storing a co-occurrence probability of each of plural words in association with each of distribution-destinations, the apparatus extracts, from a message to be distributed, an unknown-word that is not included in the plural words, where the co-occurrence probability indicates a probability that each word is included in a message to be distributed to each distribution-destination.

Abstract: A translations editor for translating content is provided on at least one user terminal. The translations editor includes a content data display unit for displaying text data and image data together, wherein the text data and the image data are extracted from content data in which the text data and the image data are stored separately; a text data editor unit configured to display a first-language text included in the text data, and to input a second-language text for the first-language text; and a scroll controller for controlling one of the content data display unit and the text data editor unit to be scrolled together with the other, according to a request to scroll the content data display unit or the text data editor unit.