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, system, and storage device storing a computer program, for generating questions based on provided content, such as, for example, a document having words. The method comprises automatically estimating the probability of interesting phrases in the provided content, and generating a question in natural language based on the estimating. In one example embodiment herein, the estimating includes predicting the interesting phrases as answers, and the estimating is performed by a neural model. The method further comprises conditioning a question generation model based on the interesting phrases predicted in the predicting, the question generation model generating the question. The method also can include training the neural model. In one example, the method further comprises identifying start and end locations of the phrases in the provided content, and the identifying includes performing a dot product attention mechanism parameterizing a probability distribution.

Abstract: Seamless electronic discovery systems with an enterprise data portal are disclosed herein. An example system includes a document collection system that collects relevant documents from one or more target sources, an investigation platform that processes the relevant documents and a portal system that interfaces the document collection system and the investigation platform to provide end-to-end electronic discovery using tracking units that provide chain-of-custody proof.

Abstract: A feature engineering tool automatically generates a group of features for training a machine learning model. The feature engineering tool selects primitives for a dataset and synthesizes a plurality of features based on the selective primitives and the dataset. The feature engineering tool iteratively applies the plurality of features to a different portion of the dataset to evaluate usefulness of the plurality of features. Based on the evaluation, it removes some of the plurality of features to obtain the group of features. The feature engineering tool also determines the importance factor for each feature in the group. The machine learning model is generated based on the features and their importance factors and can be used to make a prediction based on new data.

Abstract: A multi-modal pre-training model acquisition method, an electronic device and a storage medium, which relate to the fields of deep learning and natural language processing, are disclosed. The method may include: determining, for each image-text pair as training data, to-be-processed fine-grained semantic word in the text; masking the to-be-processed fine-grained semantic words; and training the multi-modal pre-training model using the training data with the fine-grained semantic words masked.

Abstract: A computer system is provided, including a processor coupled to a mass storage device that stores instructions, which, upon execution by the processor, cause the processor to store an original string formed of a plurality of characters, perform a wordbreak algorithm on the original string, and tokenize the original string to generate a processed string including a plurality of word tokens separated by spaces. The processor is further configured to generate an offset map between locations within the word tokens in the processed string and corresponding locations in the original string and classify a portion of the processed string as a target. The processor is further configured to identify target characters in the original string that correspond to the target using the offset map and perform a predetermined action on the target characters in the original string.

Abstract: A computer-implemented method is provided for estimating output confidence of a black box Application Programming Interface (API). The method includes generating paraphrases for an input text. The method further includes calculating a distance between the input text and each respective one of the paraphrases. The method also includes sorting the paraphrases in ascending order of the distance. The method additionally includes selecting a top predetermined number of the paraphrases. The method further includes inputting the input text and the selected paraphrases into the API to obtain an output confidence score for each of the input text and the selected paraphrases. The method also includes estimating, by a hardware processor, the output confidence of the input text from a robustness of output scores of the input text and the selected paraphrases.

Abstract: Method for managing a promises management apparatus comprises automatically transcribing, using an automatic speech recognition engine, a call between a caller and an agent, based on real-time call stream of the call, wherein the caller calls the agent for resolution of a concern using the promises management apparatus. The method comprises automatically extracting, using a natural language processing engine, in real-time a promise made by the agent to the caller based on the transcript of the call, generating promises data comprising a timeline and an action, and displaying the promises data in a graphical user interface (GUI).

Abstract: Disclosed embodiments relate to aligning pairs of sentences. Techniques can include receiving a plurality of sentences; generating a graph for each of at least two sentences of the plurality of sentences, wherein generating a graph for each sentence of the at least two sentences comprises: identifying one or more tokens for the sentence; and connecting via edges the one or more tokens; generating a combined graph for the at least two sentences wherein generating a combined graph comprises: aligning the identified tokens of the at least two sentences of the plurality of sentences; identifying matching and non-matching tokens between the at least two sentences based on the alignment; and merging matching tokens into a combined graph node.

Abstract: In an approach for a smart reading assistant that helps users learn to read a new language, a processor detects a language of a digital book opening by a user on a computing device. A processor identifies a user reading capability level for the language of the digital book from a user profile for the user. A processor inputs at least one in context meaning of potential unknown words based on the user reading capability level. A processor detects a user action by the user on the computing device. A processor updates the user reading capability level based on the user action. A processor identifies, using collocation extraction, related words to a word involved in the user action. A processor inputs at least one in context meaning of at least one of the related words.

Abstract: An example controller device that manages a plurality of network devices includes one or more processing units implemented in circuitry and configured to receive, via an application programming interface (API) framework, an indication of an intent. The intent includes data indicating an update to a data structure including a plurality of nodes representing the plurality of network devices and a plurality of edges connecting the plurality of nodes. The one or more processing units are further configured to process the intent to select a topology compiler from a plurality of topology compilers and invoke, via the API framework, the selected topology compiler using a role of a network device of the plurality of network devices and an indication of the network device as input to generate abstract configuration information. The one or more processing units are further configured to configure the network device based on the abstract configuration information.

Abstract: A method, computer system, and a computer program product for coordinating supervision of at least one document processing pipeline is provided. The present invention may include receiving one or more documents. The present invention may then include parsing the received one or more documents to identify one or more performance indicators associated with the received one or more documents. The present invention may also include processing the parsed one or more documents based on a series of processor nodes. The present invention may further include identifying one or more deviations associated with the identified one or more performance indicators. The present invention may also include transferring the identified one or more deviations to a supervisor component. The present invention may then include generating at least one deviation escalation. The present invention may then further include reprocessing the generated at least one deviation escalation after a human response.

Abstract: Procedural optimization is facilitated by receiving user input for creating or modifying a body of text comprising a procedure, detecting one or more procedural steps associated with the procedure using a procedural step detection module, automatically searching within a corpus of references for one or more related procedural steps using a related procedural step extraction module, automatically identifying one or more outcomes within the corpus of references associated with the one or more related procedural steps using an outcome extraction module, automatically determining whether the one or more outcomes comprise detrimental results using an outcome analysis module, and, in response to determining a set of detrimental outcomes from the one or more outcomes that comprise detrimental results, presenting a detriment indicator within the user interface in association with the one or more procedural steps.

Abstract: The present system and method for analysing sentiment is a response to the growing availability of informal, opinionated texts like blog posts and product review websites, a field of Sentiment Analysis has sprung up in the past decade to address the interrogation about how people feel about certain topics. Sentiment Analysis expands the traditional fact-based text analysis to enable opinion-oriented information systems. The present system and method for analysing sentiment presents a novel learning system and method for classifying text as positive, negative, or objective. It is mainly based on fuzzy set theory and clustering.

Abstract: Examples herein involve detection of entities in unstructured data. Terms are extracted from unstructured data. Entities scores for the terms are calculated using information from a name probability source, a known entity database, and historical context information. The entity scores indicate a probability that the respective terms refer to entities. The presence of detected entities are indicated based on the entity scores.

Abstract: Text is intelligently annotated by first creating a topic map summarizing topics of interest of the user. A data structure is created. The topic map is used to create two linked user dictionaries, a topic dictionary reflecting topic names and a traversal dictionary reflecting the knowledge structure of a topic. Actions may be linked with topic types. When the text to be annotated is being read, the topic data structure of the topics found in the text are automatically instantiated using the dictionaries and any actions previously linked to topic types. Instantiated topic data structures are automatically attached to the text being annotated. A user GUI may be created to allow the user to access and interact with the text annotations.

Abstract: A method includes: initializing a list of token embeddings, each of the token embeddings corresponding to a tokenized word from text in a corpus of text; generating a graph for a group of consecutive words s from the text, said graph including binary relations between pairs of tokenized words of the group of consecutive words; selecting the token embeddings representing the words of the group of consecutive words from the list of token embeddings; computing a tensor of binary relations as the product between a matrix of the selected token embeddings and a tensor representing discourse relations, the computed tensor representing the binary relations between the pairs of tokenized words; computing a loss using the computed tensor; optimizing the list of token embeddings using the computed loss. The above may be repeated until the computed loss is within a predetermined range.

Abstract: Embodiments include methods, systems and computer program products for performing named entity recognition. Aspects include obtaining a text having a plurality of words and comparing each of the plurality of words to a dictionary. Aspects also include creating, based on the comparison, an annotation for at least one of the plurality of words that the least one of the plurality of words refers to a named entity. Aspects further include parsing the text to identify a part of speech for each of the plurality of words and removing the annotations from each of the at least one of the plurality of words that has a part of speech that is not one or a noun and a noun supporting adjective.

Abstract: A technique manages collaborative web sessions (CWS). The technique receives graphical content of a CWS. The technique translates a set of portions of the graphical content into text output. The technique provides the text output to a set of text application services. The set of text application services associate the text output with the CWS.

Abstract: The embodiments herein disclose methods and systems for passive wakeup of a user interaction device and configuring a dynamic wakeup time for a user interaction device, a method includes detecting an occurrence of at least one first non-voice event associated with at least one device present in an Internet of Things (IoT) environment. The method includes detecting an occurrence of at least one successive event associated with the at least one device. The method includes estimating a contextual probability of initiating at least one interaction by a user with the user interaction device on detecting the occurrence of at least one of the at least one first event and the at least one successive event. On determining the estimated contextual probability is above a pre-defined threshold value, the method includes configuring the dynamic wakeup time to switch the user interaction device to a passive wakeup state.

Abstract: The described technology is generally directed towards coordinating the generation, validation and enabling of content selection graphs in an in-memory content selection graph data store. When a set of content selection graphs is requested, a coordinator starts the generation of the relevant graphs. Upon successful generation, the coordinator starts a validation of the generated graphs against rules for the nodes/response data in the graphs. If the generated graphs pass validation, the coordinator enables the graph set for use in an in-memory cache, whereby when a request to return content selection data is received, an active graph that corresponds to the request and the current time is accessed to obtain and return the response data as the requested content selection data.

Abstract: A computer-implemented method identifies one or more concepts from a document, wherein the document comprises information on completing a task. The method computes a relevance value of an utterance with respect to completing the task using the one or more identified concepts. The method removes the utterance from a dialog model to be used for completing the task when the relevance value of the utterance is below a given threshold value.

Abstract: A system includes a communication interface, at least one processor, and at least one memory storing information about a plurality of chatbots and a plurality of categories of the chatbots. The memory may further store instructions that, when executed, cause the processor to receive from a client device including a user interface (UI), data associated with a voice-based or text-based user input associated with or indicative of a request to perform at least one task, to select at least one chatbot from the plurality of chatbots for at least one category for performing the at least one task among the plurality of categories of the plurality of chatbots, and to transmit information on the selected at least one chatbot and the at least one category via the communication interface to the client device such that the client device provides the information through the UI.

Abstract: A method includes obtaining, using at least one processor of an electronic device, a base model trained to perform natural language understanding. The method also includes generating, using the at least one processor, a first model expansion based on knowledge from the base model. The method further includes training, using the at least one processor, the first model expansion based on first utterances without modifying parameters of the base model. The method also includes receiving, using the at least one processor, an additional utterance from a user. In addition, the method includes determining, using the at least one processor, a meaning of the additional utterance using the base model and the first model expansion.

Abstract: A speech synthesizer using artificial intelligence includes a memory configured to store a first ratio of a word classified into a minor class among a plurality of classes, a second ratio of the word which is not classified into the minor class, and a synthesized speech model and a processor configured to change a first class classification probability set of the word to a second class classification probability set, based on the first ratio, the second ratio and the first class classification probability set, and learn the synthesized speech model using the changed second class classification probability set.

Abstract: Semantic consolidation of data is disclosed. One example is a system including a data processor, a semantic relations module, and a consolidation module. The data processor identifies at least two data objects, each data object including a plurality of data elements. The semantic relations module identifies, via a processor, semantically similar data elements of the plurality of data elements. The consolidation module merges the identified semantically similar data elements via the processor, and generates a consolidated data object based on the merged data elements.

Abstract: Methods and systems for performing a language processing task include setting an angular coordinate for a vector representation of each of a set of words, based on similarity of the words to one another. A radial coordinate is set for the vector representation of each word, according to hierarchical relationships between the words. A language processing task is performed based on hierarchical word relationships using the vector representations of the words.

Abstract: A method for analyzing constituents of a material under test is disclosed. The method includes obtaining a plurality of impedance spectroscopy measurements of the material under test over a predetermined frequency range. A data set is passed to a machine learning system, the machine learning system having been trained to return an output related to one or more constituents in the material under test based on impedance spectroscopy measurement data over the predetermined frequency range. An output of the machine learning system is received and outputted. Related systems, methods and devices are also disclosed.

Abstract: Systems and methods for automatic generation of templates for information extraction rules to extract information objects from natural language texts.

Abstract: A system and method are disclosed, to distinguish fraudulent transactions from a legitimate transaction, predicated on the notion that the card is considered likely to be in state of fraud. The disclosed system and method can be activated as soon as an account has suspicious activity that causes a high score for potential fraud, but before a bank either can or needs to confirm fraud. The system or method is able to pinpoint the actual fraudulent transactions inside a window of potential fraudulent activity, using a specialized model referred to as the pinpoint model.

Abstract: Techniques are disclosed for processing natural language communications. A computing device receives a set of natural language communications. The computing device generates a word index from the communications and generates, from the word index, a set of topics, each topic including two or more words. For each topic, the computing device generates a score indicative of an amount of semantic information represented by the topic. The computing device then discards topics that are supersets or subsets of other topics. The computing device presents the remaining topics based on to the score of each topic.

Abstract: The present invention provides a method and a system utilizing an AI entity for confirming an agreement has been entered between a first entity and a second entity during a verbal communication, capturing the portions of the communication that constitute the elements of an agreement and storing the portions for later verification of the agreement.

Abstract: An information processing device includes a divider and an identifier. The divider divides a plurality of sentences included in a processing target sentence set into a plurality of groups based on a supervised topic list representing a list of supervised topic name that is set in advance. The identifier identifies a topic name for each of the groups based on the supervised topic list.

Abstract: A system to convert sequences of human thought representations into coherent stories, in association with a language understanding system is disclosed. Said system comprises: an entity dereferencing and enrichment module, an anomaly detecting unit that comprises: a context anomaly module, and a meaning anomaly and reinforcement module; an inter thought representation reasoning and transformation unit; an entity knowledge base; a thought representation knowledge base; and an output thought representation cloud. The system takes sequences of thought representations as input and tries to make sense out of them. The system is used in association with any type of language understanding system for creating meaning out of the sequence of thoughts.

Abstract: In one embodiment, a method includes receiving a first user input by a first user at a client system associated with the first user, wherein the first user input is associated with a first language register, selecting a first language-register model from a plurality of language-register models based on the first language register by the client system, wherein the plurality of language-register models are each personalized with respect to the first user, and presenting a personalized first communication content generated based on the selected first language-register model at the client system.

Abstract: An unidentified media content item may be received by a processing device. A set of features of the unidentified media content item may be determined. Metadata associated with the unidentified media content item may be determined. A first similarity between the metadata associated with the unidentified media content item and additional metadata associated with a known media content item from a media content repository may be determined. A second similarity between the set of features of the unidentified media content item and an additional set of features associated with the known media content item may be determined. The unidentified media content item may be identified as a cover of the known media content item based on the first similarity and the second similarity by the processing device.

Abstract: An article generation method and device, and a computer storage medium. According to an example of the method, after a content source is mined based on requirement information inputted by a user, at least one topic dimension vector may be extracted from the mined content source by using a particular topic generation model. Then, for each extracted topic dimension vector, topic sentence mining is performed on the content source according to the topic dimension vector, to obtain topic sentences corresponding to the topic dimension vector. Finally, the topic sentences corresponding to the at least one topic dimension vector are spliced and synthesized, to generate an article conforming to the requirement information.

Abstract: The technology described herein provides for a match fix-up stage that removes matching documents identified for a search query that don't actually contain terms from the search query. A representation of each document (e.g., a forward index storing a list of terms for each document) is used to identify valid matching documents (i.e., documents containing terms from the search query) and invalid matching documents (i.e., documents that don't contain terms from the search query). Any invalid matching documents are removed from further processing and ranking for the search query.

Abstract: Techniques for limiting natural language processing performed on input data are described. A system receives input data from a device. The input data corresponds to a command to be executed by the system. The system determines applications likely configured to execute the command. The system performs named entity recognition and intent classification with respect to only the applications likely configured to execute the command.

Abstract: Embodiments may provide techniques to generate training data for summarization of complex documents, such as scientific papers, articles, etc., that are scalable to provide large scale training data. For example, in an embodiment, a method may be implemented in a computer system and may comprise collecting a plurality of video and audio recordings of presentations of documents, collecting a plurality of documents corresponding to the video and audio recordings, converting the plurality of video and audio recordings of presentations of documents into transcripts of the plurality of presentations, generating a summary of each document by selecting a plurality of sentences from each document using the transcript of the that document, generating a dataset comprising a plurality of the generated summaries, and training a machine learning model using the generated dataset.

Abstract: A boundary of a highlight of audiovisual content depicting an event is identified. The audiovisual content may be a broadcast, such as a television broadcast of a sporting event. The highlight may be a segment of the audiovisual content deemed to be of particular interest. Audio data for the audiovisual content is stored, and the audio data is automatically analyzed to detect one or more audio events indicative of one or more occurrences to be included in the highlight. Each audio event may be a brief, high-energy audio burst such as the sound made by a tennis serve. A time index within the audiovisual content, before or after the audio event, may be designated as the boundary, which may be the beginning or end of the highlight.

Abstract: In one embodiment, a method includes receiving a first user input by a first user at a client system associated with the first user, wherein the first user input is associated with a first language register, selecting a first language-register model from a plurality of language-register models based on the first language register by the client system, wherein the plurality of language-register models are each personalized with respect to the first user, and presenting a personalized first communication content generated based on the selected first language-register model at the client system.

Abstract: Input context for a statistical dialog manager may be provided. Upon receiving a spoken query from a user, the query may be categorized according to at least one context clue. The spoken query may then be converted to text according to a statistical dialog manager associated with the category of the query and a response to the spoken query may be provided to the user.

Abstract: Embodiments are directed to organizing conversation information. Two or more machine learning (ML) models and a plurality of sentences provided from a conversation may be employed to generate insight scores for each sentence such that each insight score correlates to a probability that its sentence includes one or more of an action or a question. In response to one or more sentences having insight scores that exceed a threshold value an information score and a definiteness score may be determined for the one or more sentences. And one or more insights associated with the conversation may be generated based on the one or more sentences. A report may be generated that associates the one or more insights with one or more portions of the conversation that include the one or more sentences that are associated with the insights.

Abstract: Method for managing a promises management apparatus comprises automatically transcribing, using an automatic speech recognition engine, a call between a caller and an agent, based on real-time call stream of the call, wherein the caller calls the agent for resolution of a concern using the promises management apparatus. The method comprises automatically extracting, using a natural language processing engine, in real-time a promise made by the agent to the caller based on the transcript of the call, generating promises data by assigning an agent, a type, a timeline, an action, and a mode of execution to the promise, and storing the promises data in a datastore.

Abstract: A method for expanding software application functionalities processing is disclosed herein. The method comprises receiving a user request within a software application. The software application is enhanced with annotations, which include metadata. The method continues with loading the metadata from the annotations to a dialog system located on a remote device, sending at least a part of the user request to the dialog system, and receiving a dialog system response to the user request from the dialog system. At least one of the annotations associated with the dialog system response is identified. A code associated with the identified annotation is invoked within the software application.

Abstract: Methods, systems, and apparatus for receiving user input that invokes digital assistant functionality; obtaining screen data indicating content displayed on a screen of the computing device; determining a classification for an entity referenced in the content indicated by the screen data; determining a suggested request that refers to the entity based on the determined classification; and providing the suggested request in response to receiving the user input that invokes the digital assistant functionality.

Abstract: Users consume a wide variety of content from various sources, such as videos accessible through websites. As provided herein, content recommendations that are contextually and/or semantically relevant to current content consumed by a user may be identified and provided to the user. For example, metadata for a video being watched by the user may be identified (e.g., terms extracted from a description, user reviews, a category, and/or other information). The metadata may be used to identify content recommendations based upon the metadata corresponding to terms grouped into a set of refined topic groupings of a graph comprising terms and relationships between terms extracted from a content corpus. The metadata may be matched to relevant terms within the set of refined topic groupings, and content recommendations comprising content corresponding to the relevant terms may be suggested to the user.

Abstract: Methods, systems, and computer-readable media for generating a statistically covaried machine learning model for performance measurement of service providers. The method receives a configuration file that includes one or more parameters associated with a plurality of individuals and parses it to generate and executing the database query on input data to generate sets of tabulated data of individuals of the plurality of individuals. The method next determines one or more measures of service providers listed in the configuration file using two or more tabulated data of individuals from the sets of tabulated data of individuals. The method finally generates a covaried machine learning model by training a machine learning model by statistically covarying measures and using them as training data.

Abstract: Techniques and systems are provided for authenticating a user of a device. For example, input biometric data associated with a person can be obtained. A similarity score for the input biometric data can be determined by comparing the input biometric data to a set of templates that include reference biometric data associated with the user. The similarity score can be compared to an authentication threshold. The person is authenticated as the user when the similarity score is greater than the authentication threshold. The similarity score can also be compared to a learning threshold that is greater than the authentication threshold. A new template including features of the input biometric data is saved for the user when the similarity score is less than the learning threshold and greater than the authentication threshold.