Abstract: A system and method for improving computer functionality by retrieving answers/responses to questions/input from a cache such as those used with chatbots and generative AI systems. Disclosed is a multi-layered caching strategy that focuses on the relevance of a cache hit by improving the quality of the answer. The approach demonstrates that response latency is significantly reduced when using caching and how a caching strategy could be applied in various layers of increasing relevance for a simple Question-and-Answer system with the possibility of extending to more complex generative AI interactions.

Abstract: In accordance with an embodiment, described herein is a system and method for use with a data analytics or other computing environment, for on-demand fetching of backend server logs into a frontend environment, such as for example a browser. Such on-demand log fetching can be specific to the working context that is for current session and current request; and can be accomplished by appending a parameter or flag to a current request. For each step associated with an instruction being performed, the method can create a timestamp within one or more log files associated with the instruction; and fetch the one or more log files associated with the instruction. Performance logs are then included with a dashboard response, and logged into the browser's console.

Abstract: Methods and systems are presented for auditing user feedback data corresponding to user communications received via at least one interface of a service provider. The user feedback data includes a first set of feedback categories associated with a first classification of the user communications. A first feature representation of the communications is generated from the user feedback data. The first feature representation includes a first set of textual data features extracted from the communications. A second feature representation is generated from the first feature representation using a first machine learning model. The second feature representation includes a second set of textual data features including semantic equivalents of the first set of features. A second machine learning model is trained with the second feature representation. A second classification of the user communications according to a second set of feedback categories is generated using the trained second machine learning model.

Abstract: Implementations are described herein for incorporating unstructured data into machine learning-based phenotyping. In various implementations, natural language textual snippet(s) may be obtained. Each natural language textual snippet may describe environmental or managerial features of an agricultural plot that exist during a crop cycle. A sequence-to-sequence machine learning model may be used to encode the natural language snippet(s) into embedding(s) in embedding space. The embedding(s) may semantically represent the environmental or managerial features of the agricultural plot. Using one or more phenotypic machine learning models, phenotypic prediction(s) may be generated about the agricultural plot based on the one or more semantic embeddings and additional structured data about the agricultural plot. Output may be provided at one or more computing devices that is based on one or more of the phenotypic predictions.

Abstract: A method and system for automating a process of downloading and analyzing messages from conversation rooms and chat rooms to determine topics, entities, context, and actionable items are provided. The method includes downloading a set of messages that have been communicated over a communication channel; analyzing each respective message in order to determine at least one respective topic that relates to each respective message; determining, based on a result of the analysis, metrics that relate to the set of messages; and storing historical data that relates to the downloaded set of messages and each of the metrics. The analysis may be performed by executing an artificial intelligence (AI) algorithm that is based on a Natural Language Processing (NLP) model and is trained by using the historical data.

Abstract: Embodiments provide for a temporal expression parser in a conversational data-to-text system are described herein. An example method may include receiving user query data comprising an input text string; generating, based at least in part on the input text string, a n-gram set comprising a plurality of n-gram elements; traversing each n-gram element in the n-gram set to generate a parse tree list comprising one or more parse trees based on a grammar template associated with the input text string; and generating, based at least in part on a last parse tree of the parse tree list, one or more semantic frames indicating a temporal expression associated with the input text string.

Abstract: A method for processing text data includes analyzing the text data to identify a plurality of keywords. The method also includes determining whether each of the plurality of keywords already exists in one or more databases. When a keyword in the plurality of keywords is not found in the one or more databases, the method includes tagging the keyword with a plurality of characters for storage. The plurality of characters includes at least a first character to indicate a start of the tagging, a second character to indicate a corresponding database for storing the keyword, and a third character to indicate an end of the tagging. The method also includes storing the tagged keyword in the corresponding database.

Abstract: Systems and methods for creating predictor variables from unstructured data for prediction models are provided. A variable creation application receives unstructured data and processing the unstructured data to generate processed data. Based on the processed data, the variable creation application generates an attribute pool that contains multiple predictor variables generated by applying natural language processing (NLP) procedures on the processed data. The variable creation application further executes a prediction model on at least the predictor variables in the attribute pool to generate a prediction result. Based on the prediction result, the variable creation application evaluates the predictive power of each of the predictor variables and retains predictor variables that are predictive as input predictor variables for the prediction model.

Abstract: There is included a method and apparatus comprising computer code configured to cause a processor or processors to perform generating one or more aligned inventories, wherein the one or more aligned inventories are generated using one or more word sense inventories, obtaining a word in a context sentence, determining one or more semantic equivalence scores indicating semantic similarity between the word in the context sentence and each of one or more associated glosses in the one or more aligned inventories using a semantic equivalence recognizer model, and predicting a correct sense of the word in the context sentence based on the determined one or more semantic equivalence scores.

Abstract: Database systems and methods are provided for assigning structural metadata to records and creating automations using the structural metadata. One method of assigning structural metadata to a record associated with a conversation involves obtaining a plurality of utterances associated with the conversation, identifying, from among the plurality of utterances, a representative utterance for semantic content of the conversation, assigning the conversation to a group of semantically similar conversations based on the representative utterance, and automatically updating the record associated with the conversation at a database system to include metadata identifying the group of semantically similar conversations.

Abstract: Natural language generation technology is disclosed that applies artificial intelligence to structured data to determine content for expression in natural language narratives that describe the structured data. A graph data structure is employed, where the graph data structure comprises a plurality of nodes. Each of a plurality of the nodes (1) represents a corresponding intent so that a plurality of different nodes represent different corresponding intents and (2) is associated with one or more links to one or more of the nodes to define relationships among the intents.

Abstract: Data is received that includes a passage of text generated in response to a prompt which comprises a plurality of sentences. Thereafter, the passage of text is tokenized into a plurality of tokens each corresponding to a different word in the passage of text. A first classification head of an adaptive fine-tuned transforms classifies each of the tokens into one of a plurality of classes. A second classification head of the adaptive fine-tuned transformer model classifies each of the sentences as either including or not including an argument. Data can then be provided which characterizes the first and second classifications. Related apparatus, systems, techniques and articles are also described.

Abstract: A response system includes an acquisition unit configured to acquire the question sentence of a user, a response sentence selection unit configured to select a response sentence, stored in a storage unit in advance, according to the acquired question sentence, a response sentence conversion unit configured to convert the selected response sentence according to the question sentence, and an output unit configured to output the converted response sentence. The response sentence conversion unit is configured to convert a word included in the selected response sentence to a word included in the question sentence based on the degree of similarity between the word included in the response sentence and the word included in the question sentence.

Abstract: A data processing system implements a method for training machine learning modes, including receiving a set of one or more unlabeled documents associated one or more first categories of documents to be used to train machine learning models to analyze the one or more unlabeled documents, and fine-tuning a first machine learning model and a second machine learning model based on the one or more unlabeled document to enable the first machine learning model to determine a semantic representation of the one or more first categories of document, and to enable the second machine learning model to classify the semantic representations according to the one or more first categories of documents, the first machine learning model and the second machine learning model having been trained using first unlabeled training data including a second plurality of categories of documents that do not include the one or more first categories of documents.

Abstract: This disclosure relates to extraction of tasks from documents based on a weakly supervised classification technique, wherein extraction of tasks is identification of mentions of tasks in a document. There are several prior arts addressing the problem of extraction of events, however due to crucial distinctions between events-tasks, task extraction stands as a separate problem. The disclosure explicitly defines specific characteristics of tasks, creates labelled data at a word-level based on a plurality of linguistic rules to train a word-level weakly supervised model for task extraction. The labelled data is created based on the plurality of linguistic rules for a non-negation aspect, a volitionality aspect, an expertise aspect and a plurality of generic aspects. Further the disclosure also includes a phrase expansion technique to capture the complete meaning expressed by the task instead of merely mentioning the task that may not capture the entire meaning of the sentence.

Abstract: A method to be executed on a computing device comprising (i) accessing and/or modifying a database to be automatically curated, (ii) optionally accessing additional data or information sources for further useful data or information, (iii) using one or more pre-trained large language models (LLMs) accessed via API or other connections, issuing prompts and retrieving prompt-answers and executing database curation requests that specify database curation tasks to be performed on at least one sub-structure of the database, the tasks comprising (a) a database enrichment task to compute new data records to be inserted into the database sub-structure, (b) a database verification task to verify, using the one or more LLMs, data contained in the sub-structure, and identify incorrect data, (c) a database update, and (d) a null-value or a missing value replacement task. The requested tasks are automatically performed via a computation comprising an adaptively generated prompt sequence.

Abstract: A system to identify cyber threat intelligence from a group of information is disclosed. The system includes a processing subsystem including a data sourcing module to fetch information. The processing subsystem includes a data processing module to extract textual information from the information. The processing subsystem includes a machine learning module including an entity analysis module to fragment the textual information to obtain entities. The entity analysis module is to analyze a label assigned to each of the entities to generate a first threat score. The machine learning module includes a semantic analysis module to summarize the entities to obtain a summarized text. The semantic analysis module is to evaluate sentiments pertaining to the summarized text. The semantic analysis module is configured to analyze the sentiments to generate a second threat score. The machine learning module includes a classifier module to classify the textual information categories.

Abstract: Examples provide enhanced machine learning model accuracy through post-hoc confidence score calibration. A machine learning (ML) system receives results generated by an ML model, the results comprising at least one confidence score and electronic documents. The ML system processes the results generated by the ML model comprising performing document understanding by extracting data points from the electronic documents. The ML system associates the confidence score with the extracted data points and calibrates a confidence score associated with the extracted data points using a post-hoc calibration solution set. The ML system implements confidence scoring recalibration comprising aligning the confidence score with prediction accuracy and adjusting the generated confidence score by the recalibration. Based on adjusting the confidence score, the ML system extracts an individual element of information from the electronic documents comprising an extracted value.

Abstract: A system and method for generating task-specific text by processing multimodal inputs using machine-learning models is provided. The method may include accessing first sets of tokens associated with a desired task and one or more modalities associated with a context of the desired task. The method may further include determining a second set of tokens for each of the one or more modalities using a classifier network associated with the modality. The method may further include generating a number of embedding vectors by mapping the first sets of tokens and the second set of tokens associated with each of the one or more modalities to an embedding space. The method may further include producing a sequence of words addressing the desired task by processing the number of embedding vectors with an encoder-decoder network.

Abstract: Provided is a method for querying questions. The method includes: acquiring input information of a user; acquiring intention information of the user based on the input information of the user; determining an answer generation rule; and generating, based on the input information and the intention information, a first answer in accordance with the answer generation rule, and providing the first answer to the user.

Abstract: Methods and systems are described herein for efficiently labeling user utterances, which may encompass any communication received from a user within a conversational interaction, and identifying novel user intents for large amounts of data. A machine learning model may be used, which is trained on embeddings of utterance data, and which may employ methods like prototypical networks and hierarchical local binary classification for hierarchical multi-label multi-class classification.

Abstract: A method for generating a dialogue state includes: acquiring a target dialogue state of a previous round of dialogue and dialogue information of a current round of dialogue; generating an initial dialogue state of the current round of dialogue according to the target dialogue state of the previous round of dialogue and the dialogue information of the current round of dialogue; and generating a target dialogue state of the current round of dialogue according to the initial dialogue state of the current round of dialogue and the dialogue information of the current round of dialogue.

Abstract: Techniques are disclosed for generating anomaly scores for a neuro-linguistic model of input data obtained from one or more sources. According to one embodiment, generating anomaly scores includes receiving a stream of symbols generated from an ordered stream of normalized vectors generated from input data received from one or more sensor devices during a first time period. Upon receiving the stream of symbols, generating a set of words based on an occurrence of groups of symbols from the stream of symbols, determining a number of previous occurrences of a first word of the set of words, determining a number of previous occurrences of words of a same length as the first word, and determining a first anomaly score based on the number of previous occurrences of the first word and the number of previous occurrences of words of the same length as the first word.

Abstract: Described herein is a technique for mapping the raw text of a job title of an online job posting to an entity embedding, associated with an entity or entry of a title taxonomy. The raw text of the job title is first encoded to generate a multilingual word embedding in a multilingual word embedding space. Then, the vector representation of the job title, as represented in the multilingual word embedding space is translated, using a neural network, to a vector representation of the job title in the entity embedding space. Finally, a nearest neighbor search is performed to identify an entity embedding associated with an entity or entry in the title taxonomy that has a vector representation that is closest in distance to the vector output by the neural network.

Abstract: A system and method are provided for generating a cybersecurity behavioral graph from a log files and/or other telemetry data, which can be unstructured or semi-structured data. The log files are applied to a machine learning (ML) model (e.g., a large language model (LLM)) that generates/extract from the log files entities and relationships between said entities. The entities and relationships can be constrained using a cybersecurity ontology or schema to ensure that the results are meaningful to a cybersecurity context. A graph is then generated by mapping the extracted entities to nodes in the graph and the relationships to edges connecting nodes. To more efficiently extract the entities and relationships from the data file, an LLM is used to generate regular expressions for the format of the log files. Once generated, the regular expressions can rapidly parse the log files to extract the entities and relationships.

Abstract: A method, computer program, and computer system for training a graph-to-text generation network is provided. Encoded graph information corresponding to a target sentence is received, and the encoded graph information is decoded based on a biaffine attention score. One or more loss values are determined based on the decoded information, whereby the text-to-graph generation network is trained by minimizing the one or more loss values. A first loss value is generated by reconstructing one or more triple relations based on the biaffine attention score, and a second loss value predicts the graph as a linearized sequence.

Abstract: There are provided systems and methods for named entity recognition in chat dialogues for customer relationship management systems. A service provider, such as an electronic transaction processor for digital transactions, may provide live chat service channels for assistance through live agents and chatbot services. When interacting with these channels, a user may engage in a chat dialogue with live agents. This may include lines of texts corresponding to the exchanged messages and may include named entities for particular types or categories of words that refer to a particular object or thing. To identify these named entities, a natural language processor may utilize machine learning and other engines for named entity recognition in customer relationship management systems to highlight the named entities in live service chats. Agents of the systems may view content that identify the named entities and interact with the named entities to view descriptions.

Abstract: A visual dialogue method and system is provided. The method includes obtaining original input data, where the original input data includes current image data and a new question, and the new question is related to the current image data; preprocessing text data and image data in the original input data to obtain a text feature sequence and a visual feature sequence, respectively; using a VisDial dataset to construct a text corpus; obtaining text sequence knowledge by using a potential knowledge searcher based on the visual feature sequence and the text corpus; constructing a sparse scene graph based on the visual feature sequence; performing data fusion on the text feature sequence, the visual feature sequence, the text sequence knowledge, and the sparse scene graph to obtain a data fusion result; and obtaining dialogue content of the new question by using a decoder based on the data fusion result.

Abstract: Systems and methods for automatic evaluation of the quality of NLG outputs. In some aspects of the technology, a learned evaluation model may be pretrained first using NLG model pretraining tasks, and then with further pretraining tasks using automatically generated synthetic sentence pairs. In some cases, following pretraining, the evaluation model may be further fine-tuned using a set of human-graded sentence pairs, so that it learns to approximate the grades allocated by the human evaluators.

Abstract: A question answering system includes a dangerous operation determination unit configured to determine whether question-answer pair data includes a dangerous operation, and an answer sentence replacement unit configured to replace, when it is determined that the dangerous operation is included, an answer sentence including a description of the dangerous operation included in a document by a replaced answer sentence depending on a classification of the dangerous operation, to thereby create replaced question-answer pair data.

Abstract: A method for generating control commands includes providing a document that includes at least one declaration that formulates a rule set, parsing the at least one declaration to generate a plurality of syntactical blocks, constructing terms and symbols from the syntactical blocks to generate a semantic representation, validating the semantic representation, and generating at least one control command based on the validated semantic representation, wherein the at least one control command corresponds to the rule set formulated by the at least one declaration. The at least one control command can advantageously be converted into a plurality of platform-specific instructions for driving a target platform. There is further defined a device for generating control commands and a corresponding system.

Abstract: A system for operating an autonomous agent with a remote operator includes a remote subsystem, a local subsystem, and a communication subsystem. A method for operating an autonomous agent with a remote operator includes receiving and processing a set of inputs; presenting a set of outputs to a remote subsystem; receiving an input from a remote operator; and operating the autonomous agent. Additionally or alternatively, the method can include any or all of: detecting a failure and/or triggering a failure response; detecting a particular scenario associated with the autonomous agent; triggering a remote operator request; processing remote operator input; and/or any other processes.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media for music-aware speaker diarization. In an example embodiment, one or more audio classifiers detect speech and music independently of each other, which facilitates detecting regions in an audio track that contain music but do not contain speech. These music-only regions are compared to the transcript, and any transcription and speakers that overlap in time with the music-only regions are removed from the transcript. In some embodiments, rather than having the transcript display the text from this detected music, a visual representation of the audio waveform is included in the corresponding regions of the transcript.

Abstract: A computing system is disclosed that includes a processor and memory. The memory stores instructions that, when executed by the processor, cause the processor to perform several acts. The acts include generating a prompt that is to be input to a generative language model. The prompt includes conversational input set forth by a user. The acts further comprise providing the prompt as input to the generative language model, and receiving conversational output from the generative language model, where the generative language model generated the conversational output based upon the prompt. Additionally, the acts comprise streaming the conversational output on one of a SERP or webpage to which the user has navigated from the SERP.

Abstract: Systems for use in data retrieval. A natural language processing module is used in conjunction with a classifier module to analyze and decompose a user query into its elements, and to determine a type of query. These modules are also used to determine the parameters for the query. The type of query and the parameters are then used to find a suitable function that creates a structured database query with the fields in the database query being populated using the parameters and query elements found by the NLP module. The completed structured database query is then used to retrieve relevant data records in response to the query. The retrieved records are then further processed and formatted as necessary to result in a suitable response to the query.

Abstract: A method for automatically classifying terms of a first ontology into categories of a classification scheme defined with respect to a second ontology includes generating, for each term in the first ontology and each term in the second ontology, an embedding encoding the term and a description of the term. The method further includes adding the generated embeddings to a transformer model and computing, for each pair of the embeddings consisting of a first term from the first ontology and a second term from the second ontology, a similarity metric quantifying a similarity of the first term and the second term. The method still further provides for determining a matching scheme based on the similarity metric computed with respect to each pair of the embeddings, where the matching scheme associates term of the first ontology with one or more relevant categories of the classification scheme defined with respect to the second ontology.

Abstract: The present disclosure relates to systems and methods for automatic generation of summaries for record objects using one or more large language models. The system can identify a plurality of electronic activities matched to a record object of a customer relationship management system. The system can generate one or more text strings from the plurality of electronic activities using the large language models. The system can generate a first set of topics by inputting a first set of text strings into the large language models. The system can generate a second set of topics by inputting a first subset of text strings of the first set of text strings into the large language models. The system can transmit one or more topics to a computing device for presentation.

Abstract: Embodiments are provided for generating a reasonable language model learning for text data in a knowledge graph in a computing system by a processor. One or more data sources and one or more triples may be analyzed from a knowledge graph. Training data having one or more candidate labels associated with one or more of the triples may be generated. One or more reasonable language models may be trained based on the training data.

Abstract: An information processing apparatus according to the present invention includes: an acceptance unit that accepts a process request to an operation system; a specifying unit that, based on the process request, specifies an operation task to be executed in the operation system; an extraction unit that performs text analysis on the process request and extracts an answer item corresponding to an input item required at execution of the operation task from the process request; and an execution unit that executes the operation task based on the answer item.

Abstract: A method includes obtaining a text document, generating a first mark-up of the text document by a first natural language processing (NLP) model, and displaying the text document and the first mark-up. The method also includes receiving a user interaction with a portion of the first mark-up. The portion of the first mark-up of the text document may be associated with a portion of the text document. The method additionally includes, based on the user interaction with the portion of the first mark-up, selecting a second NLP model by which to process the portion of the text document, where the second NLP model is different from the first NLP model. The method further includes, based on selecting the second NLP model, generating a second mark-up of the text document by processing at least the portion thereof by the second NLP model, and displaying the text document and the second mark-up.

Abstract: The disclosed embodiments include computer-implemented processes that generate adaptive textual explanations of output using trained artificial intelligence processes. For example, an apparatus may generate an input dataset based on elements of first interaction data associated with a first temporal interval, and based on an application of a trained artificial intelligence process to the input dataset, generate output data representative of a predicted likelihood of an occurrence of an event during a second temporal interval. Further, and based on an application of a trained explainability process to the input dataset, the apparatus may generate an element of textual content that characterizes an outcome associated with the predicted likelihood of the occurrence of the event, where the element of textual content is associated with a feature value of the input dataset. The apparatus may also transmit a portion of the output data and the element of textual content to a computing system.

Abstract: A system for conducting parallelization of tasks is disclosed. The system includes an interface for receiving messages comprising a representation of logic describing two tasks to be executed in parallel, the message further comprising a content payload for use in the tasks. The system further includes a processor comprising devices running on independent machines, each device comprising a processing manager unit and at least two processing units. The processing manager is configured to parse the received messages and to distribute the at least two tasks to the at least two processing units for independent and parallel processing relative to the content payload.

Abstract: In a method of responding based on sentence paraphrase recognition for dialog system, main keywords of a domain and patterns thereof are recognized and extracted based on a morpheme analysis result in a pre-processing process. Question domains/sub-categories/dialogue-acts are classified using the morpheme analysis result and the extracted main keywords and patterns. Learning a model is performed using classification features of the classified question domains, sub-categories, and dialogue-acts as semantic features of query sentences, and sentence semantic vectors are extracted by measuring similarity between the vectors. A language model of letters is trained and similarity in expression and structure is analyzed by extracting a sentence expression vector based on the letter.

Abstract: Systems and methods are provided that train a machine-learned language encoding model through the use of a contrastive learning task. In particular, the present disclosure describes a contrastive learning task where the encoder learns to distinguish input tokens from plausible alternatives. In some implementations, on each training example the proposed method masks out some subset (e.g., 15%) of the original input tokens, replaces the masked tokens with samples from a “generator” (e.g., which may be a small masked language model), and then trains the encoder to predict whether each token comes from the original data or is a replacement produced by the generator.

Abstract: A sentence generation device has: an estimation unit for receiving input of a first sentence and a focus point related to generation of a second sentence to be generated based on the first sentence, and estimating importance of each word constituting the first sentence using a pre-trained model; and a generation unit for generating the second sentence based on the importance, and thus makes it possible to evaluate importance of a constituent element of an input sentence in correspondence with a designated focus point.

Abstract: An entity recognition method, apparatus, electronic device, and computer-readable storage medium are provided. The method includes: determining at least one entity boundary word corresponding to a text sequence; determining at least one entity candidate region in the text sequence based on the at least one entity boundary word; and performing entity recognition on the text sequence and identifying at least one entity in the text sequence based on the at least one entity candidate region.

Abstract: A document management system analyzes document clauses using document clause clusters. The document management system uses measures of similarity between document clauses from different documents to assign clauses to clause clusters. Clause clusters may be used to perform various analyses, such as to assign clauses a classification corresponding to a relevant clause cluster. The document management system provides analyses performed using document clause clusters for user review, such as to approve clause clusters, classify clause clusters, modify clause clusters, or some combination thereof.

Abstract: Systems, methods, and devices including smart interfaces with facilitated input and mistake recovery are described. For example, a smart interface system can identify one or more portions of user input as alterable decisions, and, for each of the one or more alterable decisions, store, in a memory, information about one or more alternative options for the alterable decision. The system can also identify one of the alterable decisions as the currently alterable decision, and upon receiving an input indicative of an actuation of the alteration key, alter the currently alterable decision to another of the one or more alternative options based on the stored information.

Abstract: An information processing method includes receiving first request information entered by a user, determining a first task engine for the first request information, where a first slot is set in the first task engine, extracting key information from the first request information based on the first slot, and if the key information fails to be extracted from the first request information based on the first slot, or if the key information is extracted from the first request information based on the first slot, but the extracted key information does not meet a condition, obtaining target key information from a shared parameter list of the user.

Abstract: Provided is a computer-implemented method, system, and computer program product for process mining asynchronous support conversations using attributed directly follows graphing. A processor may collect a plurality of conversation threads from an asynchronous data stream. The processor may label each utterance of a plurality of utterances from the plurality of conversation threads with an event label. The processor may analyze the event label for each utterance of the plurality of utterances. The processor may generate, based on the analyzing of the event label for each utterance, an attributed directly follows graph (DFG).