Abstract: A method for execution by a computing device includes obtaining a phrase that includes string of words and generating a valid sequence of words utilizing the phrase. The method further includes identifying a set of identigens for each word of the valid sequence of words to produce sets of identigens. The method further includes identifying, for each identigen of the sets of identigens, a word type associated with phrase structure grammar rules to produce sets of identigen-type associations. The method further includes interpreting, utilizing the phrase structure grammar rules, the sets of identigen-type associations to produce an entigen group. The entigen group represents a most likely interpretation of the phrase.

Abstract: A system and method for translating questions into database queries are provided. A text to database query system receives a natural language question and a structure in a database. Question tokens are generated from the question and query tokens are generated from the structure in the database. The question tokens and query tokens are concatenated into a sentence and a sentence token is added to the sentence. A BERT network generates question hidden states for the question tokens, query hidden states for the query tokens, and a classifier hidden state for the sentence token. A translatability predictor network determines if the question is translatable or untranslatable. A decoder converts a translatable question into an executable query. A confusion span predictor network identifies a confusion span in the untranslatable question that causes the question to be untranslatable. An auto-correction module to auto-correct the tokens in the confusion span.

Abstract: A system and method for automatically parsing an unstructured conversation to determine target entities for an engagement activity are provided. The method includes (i) obtaining the unstructured conversation associated with an asset using a dialogue manager, (ii) parsing the unstructured conversation using a natural language processing (NLP) model to obtain a target information associated with the asset, (iii) extracting attributes of one or more target entities from the target information using a Natural Language Understanding (NLU) model of domain of the unstructured conversation, (iv) generating, using the attributes of the one or more target entities, a definition of the cohort and at least one criteria for cohort curation for the asset by converting the unstructured conversation to a structured target information, and (v) determining a size of the one or more target entities based on the definition and the criteria for cohort curation for the asset for an engagement activity.

Abstract: A method for generating a relationship of events includes: obtaining a statement of a first event and a statement of a second event; generating a word sequence vector with first granularity and a word sequence vector with second granularity based on the statement of the first event; generating a word sequence vector with third granularity and a word sequence vector with fourth granularity based on the statement of the second event; generating a first fusion vector based on the word sequence vector with first granularity and the word sequence vector with second granularity; generating a second fusion vector based on the word sequence vector with third granularity and the word sequence vector with fourth granularity; and determining a relationship between the first event and the second event based on the first fusion vector and the second fusion vector.

Abstract: A method of automatically creating new intents and assigning examples to those new intents in a dialog system includes generating at least one predicted intent from an intent concept graph by combining two or more nodes of the intent concept graph. A graph embedding is generated of the at least one predicted intent. A sentence embedding is generated from examples from a call log. The graph embedding of the at least one predicted intent is then matched with the examples from the sentence embedding and the at least one predicted intent, with the examples, is connected into a dialog system.

Abstract: There is provided a technique for transforming a confusion network to a representation that can be used as an input for machine learning. A confusion network distributed representation sequence generating part that generates a confusion network distributed representation sequence, which is a vector sequence, from an arc word set sequence and an arc weight set sequence constituting the confusion network is included.

Abstract: Unstructured text is identified as larger than a threshold size. Named-entity recognition analysis is executed on the unstructured text. One or more anchor entities of the unstructured text are determined that each occur more than a threshold amount of times within the unstructured text. Two or more instances of the one or more anchor entities that are separated by at least a threshold amount of text of the unstructured text are identified. The unstructured text is partitioned into at least three sections. The unstructured text is partitioned at respective natural language demarcation points associated with each of the two or more instances such that each of the at least three sections is smaller than the threshold size. Separate coreference analyses are performed in parallel on each of the at least three sections.

Abstract: Techniques for determining an intent of a subsequent user input in a dialog are described. The system processes historic interaction data that is structured based on natural language understanding (NLU) hypotheses, with each NLU hypothesis being associated with one or more past user inputs received by the system, one or more sample inputs, and one or more past system responses. Based on processing of the historic interaction data and dialog data of previous turns of the dialog, the system determines candidate intents for the subsequent turn of the dialog. The system also uses context data to determine the candidate intents.

Abstract: Emotion information is obtained by an electronic device in order to improve communication between a person and the electronic device. Multimedia data is obtained regarding a person, predicted values for the person are obtained by applying the multimedia data to neural network models, and emotion information of the person is obtained by applying the predicted values to a weight model. Then, feedback information is obtained from the person with respect to the first emotion information of the person. Finally, the weight model is updated by using the feedback information. Subsequently, when multimedia data are again obtained regarding the person, new predicted values for the person are obtained by applying later multimedia data the plurality of neural network models, and emotion information of the person is again obtained, but this time using the weight model updated using the feedback information.

Abstract: Techniques for targeted partial re-enrichment include determining that at least one natural language processing (NLP) request is associated with at least one surface form, the NLP request being for a corpus, a database comprising preexisting annotations associated with the corpus. An index query related to the at least one surface form is performed to generate index query results, the index query results including identification of portions of the corpus affected by the NLP request. A scope of the NLP request related to the database is determined based on the index query results, the scope including identification of impacted candidate annotations of the preexisting annotations affected by the NLP request. An NLP service is performed on the corpus according to the scope and the portions, thereby resulting in updates. The updates are committed to the database associated with the corpus.

Abstract: A first example playback device includes one or more processors and a non-transitory computer-readable medium storing instructions that, when executed by the one or more processors, cause the first playback device to perform functions. The functions include generating a mixed audio signal comprising one or more components of a first audio stream and one or more components of a second audio stream; and playing the mixed audio signal, where the one or more components of the second audio stream of the mixed audio signal played by the first playback device are played in synchrony with a second playback device configured to play the second audio stream. An example non-transitory computer-readable medium and an example method, both related to the first example playback device, are also disclosed herein.

Abstract: In one embodiment, a method includes receiving a first user request to execute a first task comprising one or more sub-tasks from a client system associated with a user, receiving a second user request to suspend the first task from the client system, storing a dialog state for the first task in response to the second user request, wherein the dialog state comprises results of executed sub-tasks and a status indicating the first task is suspended, receiving a third user request to resume the first task, accessing the dialog state for the first task, and providing a prompt to resume the first task at the client system.

Abstract: According to one embodiment, an information processing apparatus includes a processing circuit. The processing circuit generates each of keywords stored in a frequently asked question (FAQ) collection as an utterance sentence. The processing circuit generates dialogue data for each of the keywords, the dialogue data generated by performing a dialogue for each of the keywords at least once, the dialogue obtained by generating a response sentence to the utterance sentence based on a result of searching the FAQ collection by use of the utterance sentence. The processing circuit generates aggregation data representing how often each of the keywords is used in the dialogue, based on the dialogue data.

Abstract: A system, program product, and method for use with an artificial intelligence (AI) platform to dynamically amend a knowledge base responsive to query evaluating and processing. A received or detected query is subject to natural language processing to identify, annotate, and map one or more query tokens against a knowledge base. The query tokens are evaluated against the knowledge base to identify one or more query tokens absent from the knowledge base and leverage a neural network to predict a probability relationship between the query tokens absent from the knowledge base and one or more tokens populated in the knowledge base. The natural language (NL) query is translated to a structured query language (SQL) and the SQL query is executed and evaluated, and the knowledge base is selectively and dynamically amended subject to the SQL evaluation.

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: Systems and methods for generating and annotating cell documents include extracting tables from a document using a table extraction engine. Headers are extracted for each of the tables using a header detection engine. Cells are extracted from each of the tables using a cell extraction engine. A cell document is generated for each of the cells which are each correlated to corresponding portions of the headers, each cell document recording the correlation between the cells and the headers. Each cell document is annotated to generate annotated cell documents with a cell recognition model trained to perform natural language processing on the cell documents by classifying each term in each of the cell documents and extracting relationships between the terms of each of the cell documents.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for automatically adjusting strategies. One of the methods includes: determining one or more characteristics of a plurality of complaints, wherein each of the complaints corresponds to an order; classifying the plurality of complaints into a plurality of categories based on the one or more characteristics by using a trained classifier; selecting a category from the plurality of categories based on a number of complaints in the selected category; from a group of strategies each associated with one or more conditions and one or more actions, identifying a candidate strategy causing the complaints of the selected category, wherein the one or more actions are executed in response to the one or more conditions being satisfied; and optimizing the candidate strategy using a reinforcement learning model at least based on a plurality of historical orders.

Abstract: Methods and systems for parsing log records. A method involves receiving a log record including data regarding a network device's operation and providing the log record to a natural language processing model. The natural language processing model may analyze the log record to identify items in the log record and relationships between items in the log record.

Abstract: An autonomous agent creates a first semantic tree from a question and second semantic tree from a candidate answer. The agent identifies, between the first semantic tree and the second semantic tree, common subtrees and calculates a semantic alignment score from a sum of sizes of each of the common subtrees. The agent forms a first syntactic tree for the question and a second syntactic tree for the candidate answer. The agent identifies a number of common syntactic nodes between the first syntactic tree and the second syntactic tree. The agent calculates a syntactic alignment score based on the number of common syntactic nodes. Responsive to determining that a sum of the semantic alignment score and the syntactic alignment score is greater than a threshold, the agent outputs the candidate answer to a device.

Abstract: The present disclosure relates to the technical field of natural language understanding, and provides a method, a terminal and a medium for generating a text based on a self-encoding neural network. The method includes: obtaining a text word vector and a classification requirement of a statement to be input; reversely inputting the text word vector into a trained self-encoding neural network model to obtain a hidden feature of an intermediate hidden layer of the self-encoding neural network model; modifying the hidden feature according to a preset classification scale and the classification requirement; defining the modified hidden feature as the intermediate hidden layer of the self-encoding neural network model, and reversely generating a word vector corresponding to an input layer of the self-encoding neural network model by the intermediate hidden layer; and generating the corresponding text, according to the generated word vector.