Abstract: Technologies pertaining to slot filling are described herein. A deep neural network, a recurrent neural network, and/or a spatio-temporally deep neural network are configured to assign labels to words in a word sequence set forth in natural language. At least one label is a semantic label that is assigned to at least one word in the word sequence.

Abstract: An information processing method includes acquiring first text information from a storage apparatus in which the first text information representing one or more utterance sentences is stored as a learning data set, identifying one or more named entities included in the acquired first text information, replacing each of the one or more identified named entities with an abstract expression abstracted based on a predetermined rule thereby generating second text information from the first text information, and learning a model of a dialogue system using, as training data, the second text information generated in the replacing.

Abstract: A method whereby an audio content is analyzed for speech semantic content and speech vocal content to determine audio content metrics. Additionally, the method may also incorporate the similarity of the audio content to other audio contents and weight the analyses to determine audio content metrics.

Abstract: A computer-implemented method, computer program product, and apparatus are provided. The method includes generating a plurality of sequences of small unit tokens from a first language model that is trained with a small unit corpus including the small unit tokens, the small unit corpus having been derived by tokenization with a small unit. The method further includes tokenizing the plurality of sequences of small unit tokens by a large unit that is larger than the small unit, to create a derived large unit corpus including derived large unit tokens.

Abstract: A method for processing an input data matrix so as to provide co-clustering assignments of rows and columns of the input data matrix includes generating, from the input data matrix, a graph of rows in the data matrix and a graph of columns in the data matrix, determining an objective function for generating a compact representation of the input data matrix, and generating the compact representation of the input data matrix by performing alternating optimization of the objective function until a test compact representation that causes the objective function to converge is generated. The method additionally includes generating a new graph of rows and a new graph of columns from the compact representation of the input data matrix, and returning a connected component of the new graph of rows and a connected component of the new graph of columns as the co-clustering assignments of the rows and columns.

Abstract: Methods, systems, and computer-readable storage media for receiving a source domain data set including a set of source document and source label pairs, each source label corresponding to a source domain and indicating a sentiment attributed to a respective source document, receiving a target domain data set including a set of target documents absent target labels, processing documents of the source and target domains using a feature encoder of a DAS platform, to map the documents of the source and target domains to a shared feature space through feature representations, the processing including minimizing a distance between the feature representations of the source domain, and feature representations of the target domain based on a set of loss functions, providing an ensemble prediction from the processing, and providing predicted labels based on the ensemble prediction, the predicted labels being used by the sentiment classifier to classify documents from the target domain.

Abstract: A system and method for processing source data, which includes a translation processor configured on a server to translate a stream of source data received at the server into a stream of formatted data types based on one or more of parameter definitions for the source data and new parameter information for the source data having an unknown data type. The source data can be received at the server from multiple sources and the data format may not be consistent from one source to another. An analytics processor on the server can be configured to operate on the out-of-tolerance data type to perform one or more of trend data analytics, associated data analytics and preventative action analytics.

Abstract: Representative embodiments disclose mechanisms to compile documents into a timeline document that tracks the evolution of a topic over time. Social media documents can be used to identify importance or popularity of linked documents (i.e., documents shared by social media in a post, tweet, etc.). A collection of social media documents is analyzed and used to identify a series of n-grams and a ranked list of linked documents. A subset of the ranked list is selected based upon similarity to the series of n-grams. The subset is then summarized and captured, along with underlying supporting data, into an entry of a timeline document. Related entries in different timeline documents can be linked to create a pivot point that allows a user to jump from one timeline to another. Timeline documents can be made available as part of a search performed by a query system.

Abstract: According to an embodiment, a speech recognition apparatus includes a calculation unit that calculates, based on a speech signal, a score vector sequence including score vectors including an acoustic score for each of input symbols, a search unit that generates an input symbol string by searching for a path of the input symbol tracing the acoustic score having a high likelihood in the score vector sequence and that generates an output symbol representing a recognition result of the speech signal based on a recognition target symbol representing linguistic information as a recognition target among the input symbols, an additional symbol acquisition unit that obtains an additional symbol representing paralinguistic information and/or non-linguistic information from among the input symbols included in a range corresponding to the output symbol, and an output unit that outputs the output symbol and the obtained additional symbol in association with each other.

Abstract: A text analytics system may ascertain sentiment about multi-sectioned documents and may associate the sentiment with particular sections. The system may include at least one processor configured to scrape the Internet for text data associated with comments expressed by a plurality of individuals about a common multi-sectioned document. The comments may not be not linked to a particular section. The at least one processor may be further configured to analyze the text data in order to determine a sentiment associated with each comment; apply an association analysis filter to the text data in order to correlate at least a portion of each comment with one or more sections of the multi-sectioned document; and transmit for display to the system user a visualization of the sentiment mapped to one or more sections of the multi-sectioned document.

Abstract: Software that answers questions asked of question answering systems, by performing the following steps: (i) identifying a hierarchy of question and answer pairs, where a first answer paired to a first question in the hierarchy of question and answer pairs includes at least a second answer paired to a second question in the hierarchy of question and answer pairs; (ii) receiving, by a question answering system, an input question from a first user; (iii) identifying first user contextual information pertaining to the first user; and (iv) answering the input question, by the question answering system, based, at least in part, on the hierarchy of question and answer pairs and the first user contextual information.

Abstract: Symbol prediction can be implemented using a multi-task system trained for different tasks. The tasks may include a single symbol prediction, symbol category prediction, and symbol subcategory prediction. Categories of symbols can be generated by clustering sets of training data using a clustering scheme.

Abstract: A smartphone-based telephone translation system includes a translation machine that starts a call translation process when the call translation is set to ON. The system detects an incoming call or dials, translates a voice signal to obtain a translation text for the ON call state. The voice signal of the caller is received and recognized. A voice signal is translated to obtain a translation text for the ON call state. The localized voice signal is recognized, using a TTS machine to obtain a reply voice signal and sent to the caller. The system performs voice recognition and text translation on the call voice, translates the language into localized language, and performs the function of calling between different languages in real time, so as to realize a telephone communication between two persons speaking different languages. The system supports communication with a hearing-impaired person or deaf-mute person.

Abstract: There is provided a method that includes (a) receiving a request for access to data, (b) identifying a data store that stores the data, and (c) communicating with the data store, by way of an electronic communication, to access the data. There is also provided a system that performs the method, and a storage medium that includes a program module for controlling a processor to perform the method.

Abstract: Techniques to automatically correct or complete text are disclosed. An entered text and a context data indicating a context in which the entered text is used are received. Examples of context data include additional words and/or a phrase or sentence in which the entered text occurs. A replacement candidate to replace the entered text is determined based on the entered text and the context data.

Abstract: A method for translating a current electronic document is disclosed that includes storing previous translations of prior electronic documents for profiles of translation professionals, extracting terms from prior electronic documents, and generating glossaries that are each associated with one of the profiles and include a respective subset of terms. The method also includes receiving a request to translate the current electronic document, selecting one or more of the profiles based on proximity of the respective subset of terms to extracted terms of the current electronic document, evaluating qualities of the previous translations for each of the selected profiles, planning a workflow for translation of the current electronic document based on the selected profiles, and causing the current electronic document to be translated according to the planned workflow.

Abstract: Exemplary embodiments relate to techniques for improving a machine translation system. The machine translation system may include one or more models for generating a translation. The system may generate multiple candidate translations, and may present the candidate translations to different groups of users, such as users of a social network. User engagement with the different candidate translations may be measured, and the system may determine which of the candidate translations was most favored by the users. For example, in the context of a social network, the number of times that the translation is liked or shared, or the number of comments associated with the translation, may be used to determine user engagement with the translation. The models of the machine translation system may be modified to favor the most-favored candidate translation. The translation system may repeat this process to continue to tune the models in a feedback loop.

Abstract: Systems and methods for disambiguation of an ambiguous entity in a search query based on the gaze of a user. These systems and methods may be implemented by a media guidance application (e.g., executed by user equipment associated with the user). In some aspects, the media guidance application may monitor the gaze of the user and attempt to disambiguate the ambiguous entity based on an area of the screen the user viewed while issuing the search query. If the media guidance application receives an indication that it did not disambiguate the ambiguous entity in the search query correctly, the media guidance application may increase the area of the screen that the user viewed in order to find an additional entity. This may allow the media guidance application to quickly and accurately find the correct answer to the user's search query.

Abstract: Systems and methods are provided for scoring the use of financial market messages. Uses of messages that are efficient and/or improve market liquidity receive positive scores. Traders are charge fees or receive rewards that are dependent upon message quality scores.

Abstract: Methods and apparatus related to determining reply content for a reply to an electronic communication. Some implementations are directed generally toward analyzing a corpus of electronic communications to determine relationships between one or more original message features of “original” messages of electronic communications and reply content that is included in “reply” messages of those electronic communications. Some implementations are directed generally toward providing reply text to include in a reply to a communication based on determined relationships between one or more message features of the communication and the reply text.

Abstract: Knowledge map-based question-answer method, server, and storage medium are provided. The method includes obtaining a natural query sentence inputted by a user, and identifying a globally unique identifier, with respect to the knowledge map, of an entity in the natural query sentence. The knowledge map includes attributes and attribute values of entities and relationships thereof. The method also includes according to a context-free grammar rule, parsing the natural query sentence into a syntax tree, and according to the syntax tree, obtaining a logical expression corresponding to the natural query sentence. Moreover, the method includes according to the logical expression and the globally unique identifier of the entity, generating a machine query sentence corresponding to the knowledge map.

Abstract: The present invention enables a system to efficiently maintain an ancillary pane containing a projection of document content, and content derived therefrom, while the underlying document is being edited. In one embodiment, an acronym pane displays a list of acronyms used in a document and one or more expansions of those acronyms. A data store containing maps is used to index which paragraphs contain which acronyms, and which acronyms are associated with which expansions. As sections of content are added, modified, or deleted, acronyms and expansions are added to and/or removed from the data store, causing the acronym pane to be updated.

Abstract: Provided are a computer program product, system, and method for using remote words in data streams from remote devices to autocorrect input text. Remote words included in a at least one data stream are accessed from at least one remote device over a network to store in a remote word corpus. User input text is received in a text entry field in a user interface rendered by the computer device. Local words are maintained in a local word corpus stored in the computer device. The local words and the remote words are used to autocorrect and validate the user input text.

Abstract: A word generation model obtains textual content and a requested topic of interest, and generates a targeted summary of the textual content tuned to the topic of interest. To do so, a topic-aware encoding model encodes the textual content with a topic label corresponding to the topic of interest to generate topic-aware encoded text. A word generation model selects a next word for the topic-based summary from the topic-aware encoded text. The word generation model is trained to generate topic-based summaries using machine learning on training data including a multitude of documents, a respective summary of each document, and a respective topic of each summary. Feedback of the selected next word is provided to the word generation model. The feedback causes the word generation model to select subsequent words for the topic-based summary based on the feedback of the next selected word.

Abstract: Methods, apparatuses, and computer program products are described herein that are configured to generate an operator text in response to an alarm that is either received from an alarm or alert system or that is self-generated based on an analysis of one or more data feeds. The method of an example embodiment may include determining whether an operator text is to be generated in response to a received alert condition by performing data analysis operations comprising: analyzing, using a processor, a primary data feed and at least one confirmatory data feed to identify one or more features; and determining based on the detection of a feature in the primary data feed or the at least one confirmatory data feed satisfies at least one predetermined constraint. The method may further include generating an output text that is displayable in a user interface that describes at least a diagnosis for the feature that satisfied that at least one predetermined constraint.

Abstract: A method, system and computer program product for improving the quality of the posts in a social network. A topic of a post to a social network environment is identified using natural language processing. The tense of the post directed to the identified topic is detected using natural language processing. The post is transformed to a suggested post of a second tense directed to the same topic to the user using natural language processing. A suggested message to be posted by the user to the social network in a third tense (e.g., past tense) based on the posts directed to the identified topic in the first and second tenses (e.g., future and present tenses) is generated using natural language processing thereby forming a past/present/future pairing of posts. By forming a past/present/future pairing of posts, a more complete and understandable view of the user's personal history may be obtained.

Abstract: Methods, apparatuses, and computer program products are described herein that are configured to be embodied as and/or performed by a document planner. In some examples, a method is provided for generating a document plan. The method may include receiving a document plan template and a message store. The document plan template may include program code defining the structure and content of a document plan. The method may also include processing, by a processor, the document plan template to determine one or more messages from the message store for inclusion in the document plan. The method may also include generating the document plan. A structure and a content of the document plan may be determined at least based on the determined one or more messages and the document plan template. Apparatuses and computer readable media are also provided.

Abstract: Provided are a language conversion method and device based on artificial intelligence, and a terminal. The language conversion method includes: receiving a language conversion request, wherein the language conversion request comprises a sentence to be converted, and the sentence to be converted comprises an original language word and a target language word; determining a plurality of candidate words each corresponding to the original language word according to a preset rule; determining a score of each of the plurality of candidate words when used in the sentence to be converted; and determining a target word corresponding to the original language word according to the score of each of the plurality of candidate words.

Abstract: The present invention is related to an information extraction method. The information extraction method may comprise providing r semantic relationships, acquiring entity pairs corresponding to the semantic relationships, acquiring first instances based on the entity pairs, and eliminating instances that do not have the semantic relationships from the first instances to obtain second instances. r is a positive integer. Each of the entity pairs contains a pair of named entities. The first instances are sentences containing the entity pairs.

Abstract: An input method editor (IME) configured to provide language assistance across a plurality of applications is disclosed. In one example, the IME is adapted for use by English-as-a-second-language (ESL) users. In a specific example, language assistance may be provided by first detecting a need to suggest a substitute word to a user who is typing within an application. The detection may be based on a probability that a current word is in error. If a need is detected, a suggestion may be obtained for the word, such as from a cloud linguistic service or from a local lexicon and language-model, if network connectivity is poor. Once obtained, the suggestion may be displayed to the user in a non-intrusive manner by user interface element(s). Interaction with the user allows the user to accept or reject the suggestion, and perform other functions, such as relocating user interface elements utilized by the display.

Abstract: In one embodiment, a sequence of input words is received. Each of the input words is encoded as an indicator vector, wherein a sequence of the indicator vectors captures features of the sequence of input words. The sequence of the indicator vectors is then mapped to a distribution of a contextual probability of a first output word in a sequence of output words. For each subsequent output word, the sequence of the indicator vectors is encoded with a context, wherein the context comprises a previously mapped contextual probability distribution of a fixed window of previous output words; and the encoded sequence of the indicator vectors and the context is mapped to the distribution of the contextual probability of the subsequent output word. Finally, a condensed summary is generated using a decoder by maximizing the contextual probability of each of the output words.

Abstract: Methods for generating jokes include coupling one or more servers with one or more databases having words stored therein; receiving text at the server(s) from an external source communicatively coupled with the server(s) through a telecommunications network; in response to receiving the text at the server(s): selecting one or more topic keywords of the topic sentence using the server(s); generating one or more punch words with the server(s) using words stored in the database related to the topic keyword(s); adding bridges to the punch word(s), using the server(s), to generate one or more jokes; communicating a signal to a first computing device through the telecommunications network using the server(s); and in response to receiving the signal at the first computing device, displaying or speaking one of the one or more jokes using the first computing device. Systems for generating jokes include networked computer components configured to carry out the methods.

Abstract: A system for improved natural language understanding (NLU) provides pre-feature input to a named entity recognition (NER) component. Pre-features may include no-textually derived information associated with the circumstances describing a user command (such as time, location, etc.). A domain and/or intent may also be determined prior to NER processing and may be passed to the NER component as an input. The pre-features and/or domain or intent information may assist the NER processing by providing context to a textual input, thereby improving NER processing such as semantic tagging, which in turn may improve overall NLU processing quality.

Abstract: A method for information processing includes collecting input information, obtaining feedback information in response to the input information, and outputting the feedback information. The feedback information contains query sub-information.

Abstract: Presenting natural-language-understanding (NLU) results can include redundancies and awkward sentence structures. In an embodiment of the present invention, a method includes, responsive to receiving a result to a NLU query, loading a matching template of a plurality of templates stored in a memory. Each template has mask fields associated with at least one property. The method compares the properties of the mask fields of each of the templates to properties of the query and properties of the result, and selects the matching template. The method further completes the matching template by inserting fields of the result into corresponding mask fields of the matching template. The method may further suppress certain mask fields of the matching template to increase brevity and improve the naturalness of the response when appropriate based on the results of the NLU query. The method further presents the completed matching template to a user via a display.

Abstract: Methods, apparatus, systems and articles of manufacture are disclosed to improve feature engineering efficiency. An example method disclosed herein includes retrieving a log file in a first file format, the log file containing feature occurrence data, generating a first unit operation based on the first file format to extract the feature occurrence data from the log file to a string, the first unit operation associated with a first metadata tag, generating second unit operations to identify respective features from the feature occurrence data, the second unit operations associated with respective second metadata tags, and generating a first sequence of the first metadata tag and the second metadata tags to create a first vector output file of the feature occurrence data.

Abstract: Vocabulary consistency for a language model may be improved by splitting a target token in an initial vocabulary into a plurality of split tokens, calculating an entropy of the target token and an entropy of the plurality of split tokens in a bootstrap language model, and determining whether to delete the target token from the initial vocabulary based on at least the entropy of the target token and the entropy of the plurality of split tokens.

Abstract: A semantic translation model system is described along with various methods and mechanisms for administering the same. The semantic translation model system proposed herein creates an intermediate representation and a knowledge base in multiple languages, reducing the amount of time and expensive resources typically required for translation and automatic response to written communications. The system also removes the problem of a translation being influenced by a person's writing style and human misinterpretation and provides ongoing translation to keep the system current.

Abstract: The present disclosure provides a method and apparatus of discovering a bad case based on artificial intelligence, a device and a storage medium, wherein the method comprises: performing named entity recognition for a to-be-recognized query, and respectively obtaining a confidence level of each character in the query; respectively obtaining a probability value of each character of forming a word with a neighboring character in the query; determining whether there is a bad case according to the confidence level and the probability value. The solution of the present disclosure may be applied to save man power costs, and improve the processing efficiency and enhance a discovery rate of bad cases.

Abstract: Described herein are systems and methods for providing a natural language comprehension system (NLCS) that iteratively performs an alternating search to gather information that may be used to predict the answer to the question. The NLCS first attends to a query glimpse of the question, and then finds one or more corresponding matches by attending to a text glimpse of the text.

Abstract: Embodiments described herein provide a more flexible, effective, and computationally efficient means for determining multiple intents within a natural language input. Some methods rely on specifically trained machine learning classifiers to determine multiple intents within a natural language input. These classifiers require a large amount of labelled training data in order to work effectively, and are generally only applicable to determining specific types of intents (e.g., a specifically selected set of potential inputs). In contrast, the embodiments described herein avoid the use of specifically trained classifiers by determining inferred clauses from a semantic graph of the input. This allows the methods described herein to function more efficiently and over a wider variety of potential inputs.

Abstract: A computer-implemented method, a computer program product, and a computer system for transformation of security information and event management (SIEM) rules and deploying the SIEM rules in a network of event processors. A computer system or server converts the SIEM rules to formal representations. The computer system or server generates rule abstraction of the formal representations, by using an abstraction function. The computer system or server constructs a finite automaton based on the rule abstraction. The computer system or server eliminates irrelevant transitions in the finite automaton to generate an optimized finite automaton. The computer system or server generates optimized formal rules, based on the optimized finite automaton. The computer system or server converts the optimized formal rules to optimized SIEM rules. The computer or server deploys the optimized SIEM rules in the network of the event processors.

Abstract: A sensory eyewear system for a mixed reality device can facilitate user's interactions with the other people or with the environment. As one example, the sensory eyewear system can recognize and interpret a sign language, and present the translated information to a user of the mixed reality device. The wearable system can also recognize text in the user's environment, modify the text (e.g., by changing the content or display characteristics of the text), and render the modified text to occlude the original text.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer-readable storage medium, and including a computer-implemented method for providing creatives. The method comprises identifying, using one or more processors, a creative for processing, the creative including a title portion and a body portion, where the body portion includes a visual uniform resource locator (URL), the visual URL comprising a visual portion and a link to a resource. The method further comprises evaluating the visual URL for inclusion in the title portion, including determining when promotion of the visual URL satisfies one or more promotion criterion, and if so, promoting the visual URL for inclusion in the title portion. The method further comprises providing the creative including the title portion with the promoted visual URL.

Abstract: A computing system is operable as a virtual personal assistant (VPA). Among other things, a semantic preprocessor of the VPA recognizes natural language user input that is intended as a reference to a known semantic type, and creates a semantic interpretation of the natural language input. The semantic interpretation may be used by a natural language understanding module of the VPA.

Abstract: Methods, systems, and computer program products for extracting from a descriptive document the value of a slot associated with a target entity described in the descriptive document are disclosed. The descriptive document is split into a set of sentences, and the sentences are filtered to generate a set of candidate sentences. Candidate sentence-entity pairs are determined from the candidate sentences, where each candidate sentence-entity pair includes a candidate sentence and an associated entity of the same type as the slot. The candidate sentence-entity pairs are compared to a set of gold sentences associated with the slot to calculate a plurality of similarity measures. A candidate sentence-entity pair associated with a maximum similarity measure is determined, and the value of the associated entity of the determined candidate sentence-entity pair is assigned to the slot.

Abstract: A method for processing an electronic document (ED) to infer titles in the ED is provided. The method includes: generating a mark-up version of the ED comprising text-styling attributes, text-layout attributes, and text content information of characters included in the ED; generating statistical information of the text-styling and text-layout attributes; calculating, for each text-styling and text-layout attribute, a relative weight score; calculating, for each paragraph in the ED: a styling criteria score and a layout criteria score based on the statistical information and the relative weight scores; a text content score based on the text content information; and a title confidence score based on the styling criteria score, the layout criteria score, and the text content score; and generating a metadata for the ED that includes the title confidence score for each paragraph for use in inferring the titles in the ED.

Abstract: Compatibility can be maintained when program code and pre-computed data associated with the program code are distributed in separate modules by using a data structure of key/value pairs comprising a dictionary. The dictionary can include a fixed part, a floating part and a version number. The version number in the dictionary can be used to determine if a dictionary for the program is current or outdated by comparing the dictionary version number and the version number associated with a dictionary layout in the current shared library. The dictionary can be rebuilt dynamically if the version number of the dictionary does not match the version number of the dictionary layout in the current shared library. The version number of the dictionary can be updated to the version associated with the layout version and the version pointer in the dictionary can be updated to point to the updated version number.

Abstract: Systems for caller identification and authentication may include an authentication server. The authentication server may be configured to receive audio data including speech of a plurality of telephone calls, use audio data for at least a subset of the plurality of telephone calls to store a plurality of known characteristics each associated with a specific demographic, and/or use audio data for at least one of the plurality of telephone calls to identify the telephone caller making the telephone call based on determining a most similar known characteristic of the plurality of known characteristics to the audio data of the caller.

Abstract: A method of presenting a sentiment indicator with entered text is disclosed. The method comprises causing a presentation of a graphical user interface (GUI) comprising a screen, the screen including a first area for receiving text and a second area next to the first area for showing a sentiment indicator; receiving text via the first area; before receiving a request for submitting the text to a server for access by other hardware processors in a distributed computing environment: determining whether a triggering condition for estimating a sentiment is satisfied; in response to determining that the triggering condition is satisfied, estimating a sentiment of the text; and causing displaying the sentiment indicator in the second area with a sentiment indicator value corresponding to the estimated sentiment of the text.