Abstract: A digital device is provided with circuitry configured to select object elements stored in a storage of the digital device, and assemble the selected object elements, a plurality of sensors configured to measure at least one parameter of structural parameter, operational parameter, and environmental status parameter, the at least one parameter being associated with a physical object represented by the assembled object elements, wherein the assembled object elements include data structure representing states of subsystems of the physical object, and the data structure holds a value of the at least one parameter.

Abstract: A computer-implemented method of generating a digital twin. To improve the efficiency of handling a digital twin the method includes the steps of: providing a digital twin modular system, the digital twin modular system being characterized by digital twin modules being adapted or being adaptable to digital twin features, the digital twin features relating to at least one of calculation properties, scope of simulation, model architecture, communication safety, accessibility, data storage, encryption functions, resource allocation, hardware requirements, providing a tuning module, wherein the tuning module includes a tuning module parameter set, wherein the tuning module parameter-set includes tuning module parameters respectively corresponding to at least one digital twin feature generating a digital twin by using the tuning module performing the steps of selecting, configuring, and combining digital twin modules of the digital twin modular system according to the tuning module parameters.

Abstract: A method for simulating a product in a computing environment includes analyzing data related to at least constructional representation, determining a requirement of prefabricating at least one custom construction artefact for a construction project based on the analysis of the data, determining at least one of: an intent or at least one objective associated with the at least one custom construction artefact, and enabling simulation of the at least custom construction artefact based on the determination of the intent or the at least one objective to render a digital representation of the product in a graphical user interface.

Abstract: The technology disclosed relates to simplifying updating of a predictive model using clustering observed points. In particular, it relates to observing a set of points in 3D sensory space, determining surface normal directions from the points, clustering the points by their surface normal directions and adjacency, accessing a predictive model of a hand, refining positions of segments of the predictive model, matching the clusters of the points to the segments, and using the matched clusters to refine the positions of the matched segments. It also relates to distinguishing between alternative motions between two observed locations of a control object in a 3D sensory space by accessing first and second positions of a segment of a predictive model of a control object such that motion between the first position and the second position was at least partially occluded from observation in a 3D sensory space.

Abstract: A method of optimizing golf club head and golf ball design is disclosed herein. The method includes inputting a RBF for stress, a RBF for CT and a RBF for ball speed into an Adaptive Simulated Annealing algorithm to generate an optimized structure for the club head for ball speed.

Abstract: Systems and methods for setting up a physics-based model are provided. One system includes one or more components that are executed by one or more computer subsystems and that include a physics-based model describing a semiconductor fabrication-related process and a set up component configured for setting up the physics-based model in multiple phases in each of which only a subset of all of the parameters of the physics-based model are set up. A configuration of the set up component is changed between at least two of the multiple phases based on the subset of all of the parameters of the physics-based model set up in the at least two of the multiple phases. The set up component may perform a Bayesian optimization technique for cascaded model set up or calibration using multiple information sources and objective functions.

Abstract: A method for analyzing disaster prevention and mitigation effectiveness of an ecological seawall is provided, including: performing seawall ecologicalization on a target seawall; establishing three-dimensional space hydrodynamic force for the target ecological seawall; simulating wave climbing on a dike body and a wave overtopping on a dike top of the target ecological seawall to obtain a wave overtopping index; calculating wave-flow bottom shear stress of the target ecological seawall, establishing a sediment movement model, and calculating suspended load and bed load sediment transportation volumes; calculating the change index of coastal bed surface according to the suspended load and bed load sediment transportation volumes, and determining a development index of tidal flats in front of dike of the target ecological seawall according to the change index; and calculating the disaster prevention and mitigation effectiveness grade of the target ecological seawall according to the wave overtopping index and t

Abstract: An integrated circuit including a standard cell includes: a plurality of first wells extending in a first horizontal direction with a first width and of a first conductivity type; and a plurality of second wells extending in the first horizontal direction with a second width and having a second conductivity type, wherein the plurality of first wells and the plurality of second wells are alternately arranged in a second horizontal direction that is orthogonal to the first horizontal direction, and when m and n are integers greater than or equal to 3, the standard cell has a length in the second horizontal direction, the length being equal to a sum of m times a half of the first width and n times a half of the second width.

Abstract: Address generators for use in verifying an integrated circuit hardware design for an n-way set associative cache. The address generator is configured to generate, from a reverse hashing algorithm matching the hashing algorithm used by the n-way set associative cache, a list of cache set addresses that comprises one or more addresses of the main memory corresponding to each of one or more target sets of the n-way set associative cache. The address generator receives requests for addresses of main memory from a driver to be used to generate stimuli for testing an instantiation of the integrated circuit hardware design for the n-way set associative cache. In response to receiving a request the address generator provides an address from the list of cache set addresses.

Abstract: This application discloses a computing system implementing a design verification tool to perform functional verification on a circuit design describing an electronic device and collect samples of performance data during the functional verification. The computing system can also include a performance visualization tool to generate a profile presentation based on the samples of performance data. The profile presentation, when displayed, can annunciate portions of the circuit design corresponding to at least one performance hotspot. The performance visualization tool can receive a data reduction request based on the performance hotspot annunciated by the profile presentation. The data reduction request can identify a subset of the performance data in the profile presentation. The performance visualization tool can generate a refined profile presentation based, at least in part, on the samples of performance data and the subset of the performance data identified in the data reduction request.

Abstract: A system is disclosed that includes a memory, and a processor configured to perform operations stored in the memory. The processor performs the operations to analyze each of a first set of sequential elements of a plurality of sequential elements to determine an edge of a clock signal pattern of a clock associated with each of the first set of sequential elements causing an output change at corresponding one or more sequential elements of the first set of sequential elements. The processor further performs the operations to discard one or more cycles of the clock signal pattern of the clock from emulation that do not include the edge of the clock signal pattern that causes at least one sequential element of the first set of sequential elements to change the output and emulate remaining cycles of the clock signal pattern of the clock.

Abstract: Aspects of the present disclosure address systems and methods for driver resizing using a transition-based capacitance increase margin. An integrated circuit (IC) design stored in a database in memory is accessed. The IC design comprises a net comprising a set of driver cells. A capacitance increase margin for resizing an initial driver cell is determined based on a total capacitance of the net and transition time target associated with the initial driver cell. An alternative driver cell is selected from a library to resize the initial driver cell and is used to replace the initial driver cell in the net. The alternative driver is selected such that a pin capacitance of the alternative driver cell exceeds an initial pin capacitance corresponding to the initial driver cell by no more than the capacitance increase margin.

Abstract: A method for designing a circuit includes adding, to a circuit design, a power switch configured to produce only one output over an acknowledgement port. The power switch does not include input and output supply ports. The method also includes adding, to the circuit design, an isolation circuit in which only one select pin is used to produce an output. The isolation circuit does not include isolation power and retention circuitry. The method also includes adding, to the circuit design, a retention circuit. The retention circuit includes a clock gating enabled register, a first AND gate connected to a clear pin of the register, and a second AND gate connected to a chip enable pin of the register. The method further includes compiling, by a processing device, the circuit design.

Abstract: Systems, methods and devices are provided, which can include an engineering change order (ECO) base. A base layout cell includes metal layer regions, conductive gate patterns arranged above metal layer regions; oxide definition (OD) patterns, metal-zero layer over oxide-definition (metal-zero) patterns, at least one cut metal layer (CMD) pattern; and at least one via region. The base layout cell can be implemented in at least two non-identical functional cells. A first functional cell of the at least two non-identical functional cells includes first interconnection conductive patterns arranged connecting metal-zero structures corresponding to at least two metal-zero patterns in a first layout, and a second functional cell of the at least two non-identical functional cells includes second interconnection conductive patterns arranged connecting metal-zero structures corresponding to at least two metal-zero patterns in a second layout.

Abstract: Various embodiments provide for context-aware circuit design layout construct, which may be part of electronic design automation (EDA). In particular, some embodiments enable use of a circuit design layout construct with a layout of a circuit design (hereafter, a circuit design layout), where a programmable pattern of layout shapes of the circuit design layout construct can be inserted into a circuit design layout and can be adapted based on context information associated with the location of its placement within the circuit design layout.

Abstract: An integrated circuit includes a first and second active region, a first insulating region, and a first and second contact. The first and second active regions extend in a first direction, are in a substrate, and are located on a first level. The first active region includes a first drain/source region and a second drain/source region. The second active region includes a third drain/source region. The first insulating region is over the first drain/source region. The first contact extends in a second direction, overlaps the third drain/source region, is electrically coupled to the third drain/source region and is located on a second level. The second contact extends in at least the second direction, overlaps the first insulating region and the first contact. The second contact is electrically insulated from the first drain/source region, is electrically coupled to the third drain/source region, and is located on a third level.

Abstract: Embodiments described herein relate to a system, software, and a method of using the system to edit a design to be printed by a lithography system. The system and methods utilize a server of a maskless lithography device. The server includes a memory. The memory includes a virtual mask file. The virtual mask file includes cells and the cells include sub-cells that form one or more polygons. The server further includes a controller coupled to the memory. The controller is configured to receive a replacement table. The replacement table includes instructions to replace the cells of the virtual mask file. The controller is further configured to replace the cells with replacement cells according to the replacement table to create an edited virtual mask file.

Abstract: A system for generating and using templates is described. In an example embodiment, a system may receive a template list and previews of templates in the template list and display the previews to a user via a graphical interface. In some instances, the system may receive an input selecting a certain template from among the templates from the user via the graphical interface, retrieve a template file describing the certain template, and display a content item of the user integrated into the certain template based on the template file. In some embodiments, the system may facilitate unlocking or providing access to templates, the template file may be associated with an asset package including files used when applying the template file to content, or the template may include multiple template pages into which content elements may be inserted to create an arrangement or sequence of content.

Abstract: A computer-implemented method and corresponding systems and apparatuses for configuring the visual appearance of a webpage. The method involves providing access to color schemes through a user interface of a software application and assigning color schemes to corresponding sections of the webpage based on user input. Each color scheme includes at least one foreground color paired with a corresponding background color that contrasts the foreground color. The method further involves generating source code representing the webpage. In some examples, the source code includes a first container corresponding to a first section and a second container corresponding to a second section. The first container includes properties corresponding to a first color scheme. Similarly, the second container includes properties corresponding to a second color scheme. Because the second section is a subsection of the first section, the source code is generated with the second container being nested within the first container.

Abstract: A method for dynamic linking between graphic documents and comment databases is described and includes entering a base graphic representation of a project in a computer system. The base graphic representation may be the conversion of a user supplied format graphic to a web support format graphic. A coordinate overlay grid of a coordinate overlay program may be linked to the base graphic representation for overlay positioning and display on the base graphic representation. The coordinate overlay program may have a notation subprogram linked to the coordinate overlay grid for a user to designate notation with an identifier relative to a distinct location of the base graphic representation on the coordinate overlay grid. A text or graphic comment data record may be entered in a comment data record referenced to an identifier and displayed using a computer cursor device and display to activate an identifier.

Abstract: Systems, methods, and devices for encoding and decoding a packet buffer for data exchange over a communications network, including parsing a data container definition file to determine a packet format; generating a code file based on the data container definition file; inserting the code file into a source tree of an application; compiling the source tree to generate an executable file; and executing the executable on a computing device in communication with the communications network, wherein the executable file encodes and decodes the packet buffer for data exchange over the communications network.

Abstract: An information processing apparatus includes a processor configured to: associate, with an edited part of a first document displayed on a display, information indicative of a second document different from the first document in a case where the second document is open on a document display application when the first document is edited; and display the information indicative of the second document on the display when the edited part is selected.

Abstract: The present disclosure describes a system, method, and computer program for syncing content across workspace pages. The system creates a first synced block on a first workspace page. Content on workspace pages rendered by the system is stored in blocks and each of the workspace pages has a hierarchy of blocks. The system adds one or more child blocks to the first synced block as content to be synced. The system creates a reference synced block on a second workspace page and adds a pointer to the reference synced block that points to the first synced block. This is done in order to sync the one or more child blocks of the first synced block across the first and second workspace pages, where editing the first synced block or the reference synced block includes editing the one or more child blocks of the first synced block.

Abstract: An information interaction method and apparatus, a server, a system, and a storage medium are provided. The method includes: receiving first text information sent by a second server, and generating second text information according to the first text information; updating the second text information in response to receiving an editing instruction sent by a client of a user; and sending updated content of the second text information to the client of each user, such that the client of each user updates the first text information according to the updated content.

Abstract: Systems and methods of the present technology generally provide computer implemented assistance for data summary, including organizing and generating a summary of data selected form source documents. A data set of user-identified information from one or more source documents and/or custom information may be received by a system server of a data summary system. The system server assigns a unique identifier to the user-identified information and/or the custom information. The system server may then analyze and prioritize the user-identified information and any custom information, and generate a formatted summary that includes the user-identified information and the custom information. The system server may then provide the summary via the communication link to a client device.

Abstract: Systems and methods for populating the fields of an electronic form are disclosed. A method includes receiving, by a processing circuit, a speech input for filling out a field of an electronic form; determining, by the processing circuit, an irregularity in the speech input; refining, by the processing circuit, the speech input based on the determined irregularity; converting, by the processing circuit, the refined speech input into text; and providing, by the processing circuit, the text to a user device to populate the field with the text.

Abstract: A method is applied to an electronic device, and includes: obtaining content copied by a user on a first page, and determining a category to which the content belongs; obtaining a category of a second page; matching the category of the content against the category of the second page; and if a matching result is not empty, displaying at least one operation option on the second page, where the at least one operation option has a function of performing an operation for the content on the second page, so that the electronic device directly performs the function after the user selects the operation option.

Abstract: A method for using a user-fillable form in a host container includes receiving, at a host container, a user-fillable form bound to dynamic data from an underlying data source where the user-fillable form has a data structure generated by prepopulated coding. The method further includes translating the user-fillable form into a hostable format for the host container. The method also includes rendering, using the hostable format for the host container, the user-fillable form in a user interface. The method further includes receiving, at the user interface of the host container, from a user of the host container, a data entry for input to the user-fillable form and updating, by the host container, the dynamic data from the underlying data source by persisting data from the data entry in a data store associated with the underlying data source.

Abstract: A method for partially publishing edits to a document is described. A first document including a first cell and a second cell is displayed on first user account on a first computing device and on a second user account. A second document including a linked version of the first cell and a linked version of the second cell is displayed. An edit of the first cell is received from the first user account and an edit of the second cell is received. In response to receiving an indication from the first user account that the edit of the first cell is to be published, the edit of the first cell is made to the linked version of the first cell, while the linked version of the second cell is unchanged so that the content of the linked version of the second cell no longer matches that of the second cell.

Abstract: A method comprises retrieving a file comprising a parent worksheet comprising a first row comprising a first statement, a first data type identifier, and a first logic; in response to receiving a first rendering request from a client computing device, generating a child worksheet in the spreadsheet comprising a second row, wherein the second row inherits the first row; receiving a second request to modify at least one of the first statement in the second statement cell, the first data type identifier in the second data type cell, or the first logic in the second logic cell; and rendering a graphical user interface based on the modified child worksheet.

Abstract: Methods and systems are provided for facilitating generation of fillable document templates. In embodiments, a document having a plurality of tokens is obtained. Using a machine learned model, a token state is identified for each token of the plurality of tokens. Each token state indicates whether a corresponding token is a static token that is to be included in a fillable document template or a dynamic token that is to be excluded in the fillable document template. Thereafter, a fillable document template corresponding with the document is generated, wherein for each dynamic token of the document, the fillable document template includes a fillable field corresponding to the respective dynamic token.

Abstract: A system processes unstructured data to identify a plurality of subsets of text in a set of text in the unstructured data and determines, for a subset from the plurality of subsets, probabilities based on a position of the subset in the set of text, a part of speech (POS) of each word in the subset, and POSs of one or more words on left and right hand sides of the subset, a number of the one or more words being selected based on a length of the set of text. The system generates a feature vector for the subset, the feature vector including the probabilities and additional features of the subset; and classifies, using a classifier, the subset into one of a plurality of classes based on the feature vector for the subset, the plurality of classes representing an ontology of a domain of knowledge.

Abstract: One or more computer processors parse a received natural language question into an abstract meaning representation (AMR) graph. The one or more computer processors enrich the AMR graph into an extended AMR graph. The one or more computer processors transform the extended AMR graph into a query graph utilizing a path-based approach, wherein the query graph is a directed edge-labeled graph. The one or more computer processors generate one or more answers to the natural language question through one or more queries created utilizing the query graph.

Abstract: Techniques to capture document data are disclosed. It is determined that a sequence of pages in a stream of document page images comprise a single multi-page document. Data is extracted from two or more different pages included in the sequence. The data extracted from two or more different pages included in the sequence of pages is used to populate a data entry form associated with the multi-page document.

Abstract: Information indicating what technology and service are necessary to solve issues is provided by a device including dictionaries recording words having similar meanings and concepts concept co-occurrences in which two of the concepts are combined, a classification of each of technology names and service names, and a score assigned to each of the classifications, a sentence input unit receiving a sentence representing an issue, a morpheme analysis unit performing morpheme analysis on the sentence and output morphemes, a concept conversion unit extracting a word from the morphemes and convert the extracted word to a concept, a classification identifying unit configured to sum scores of respective classifications in which a set of the extracted concepts and the concept co-occurrence match and identify the classification having a highest summed score, and a recommendation unit configured to obtain, from the name list, the technology names and service names belonging to the identified classification.

Abstract: Implementations set forth herein relate to providing selectable autofill suggestions, which correspond to application actions that are at least partially fulfilled using server command data—prior to a user selecting a particular selectable autofill suggestion. Proactively fulfilling command data in this way mitigates latency between user selection of a suggestion and fulfillment of a particular action. Initially, a partial input can be processed to generate autofill suggestions, which can be communicated to a server device for further processing. The autofill suggestions can also be rendered for selection at a touch display interface, thereby allowing a user to select one of the autofill suggestions. As command fulfillment data is provided by the server, the command fulfillment data can be available to a corresponding application(s) in order that any corresponding actions can be at least partially fulfilled prior to user selection.

Abstract: Techniques are described for training and/or utilizing sub-agent machine learning models to generate candidate dialog responses. In various implementations, a user-facing dialog agent (202, 302), or another component on its behalf, selects one of the candidate responses which is closest to user defined global priority objectives (318). Global priority objectives can include values (306) for a variety of dialog features such as emotion, confusion, objective-relatedness, personality, verbosity, etc. In various implementations, each machine learning model includes an encoder portion and a decoder portion. Each encoder portion and decoder portion can be a recurrent neural network (RNN) model, such as a RNN model that includes at least one memory layer, such as a long short-term memory (LSTM) layer.

Abstract: A computer-implemented method includes receiving user input including codified knowledge management information and/or engine data; and processing the input using one or more trained machine learning models to generate one or more living documents. A computing system includes one or more processors; and a memory having stored thereon instructions that, when executed, cause the computing system to receive user input including codified knowledge management information and/or engine data; and process the user input using one or more trained machine learning models to generate one or more living documents. A non-transitory computer-readable storage medium includes executable instructions that, when executed by a processor, cause a computer to receive user input including codified knowledge management information and/or engine data; and process the user input using one or more trained machine learning models to generate one or more living documents.

Abstract: A processor-implemented method for detecting an information operations campaign includes retrieving a first neural network language model including a natural language model trained on a first dataset. The first neural network language model is modified, via transfer learning and based on a second dataset, to produce a second neural network language model. Social media post data associated with a social media post is received, and features are extracted from the social media post data. The features are tokenized to produce at least one token including a value. A prediction score for the at least one token is generated using the trained neural network language model. If the prediction score exceeds a threshold value, a threat warning including a representation associated with at least one of the social media post or an account associated with the social media post is generated.

Abstract: The disclosure herein describes a system for interpreting text-based similarity between a seed item and a recommended item selected by a pre-trained language model from a plurality of candidate items based on semantic similarities between the seed item and the recommended item. The system analyzes similarity scores and contextual paragraph representations representing text-based descriptions of the seed item and recommended item to generate gradient maps and word scores representing the text-based descriptions. A model for interpreting text-based similarity utilizes the calculated gradients and word scores to match words from the seed item description with words in the recommended item description having similar semantic meaning. The word-pairs having the highest weight are identified by the system as the word-pairs having the greatest influence over the selection of the recommended item from the candidate items by the original pre-trained language model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating author vectors. One of the methods includes obtaining a set of sequences of words, the set of sequences of words comprising a plurality of first sequences of words and, for each first sequence of words, a respective second sequence of words that follows the first sequence of words, wherein each first sequence of words and each second sequence of words has been classified as being authored by a first author; and training a neural network system on the first sequences and the second sequences to determine an author vector for the first author, wherein the author vector characterizes the first author.

Abstract: Provided are a server, a client device, and operation methods thereof for training a language model. The server, the client device, and the operation methods thereof identify a word or phrase including a named entity that is incorrectly pronounced by a user or is difficult for the user to accurately pronounce from an input text for use in training a natural language understanding (NLU) model, generate text candidates for use in training the NLU model by replacing the identified word or phrase with a word or phrase predicted to be uttered by the user and having high phonetic similarity to the identified word or phrase, and train the NLU model by using the generated text candidates.

Abstract: In general, according to one embodiment, a named-entity extraction apparatus according to an embodiment includes: an unit extracting, using an extraction dictionary, named entities and relations from extraction document data; a unit designating character strings corresponding to the named entities extracted among character strings in learning document data; a generator generating, by applying a relation extraction rule, a learning document in which relations between named entities belonging to the categories of the relation extraction rule among the named entities designated; and a unit learning the extraction dictionary based on the learning document.

Abstract: Techniques are provided for using context tags in named-entity recognition (NER) models. In one particular aspect, a method is provided that includes receiving an utterance, generating embeddings for words of the utterance, generating a regular expression and gazetteer feature vector for the utterance, generating a context tag distribution feature vector for the utterance, concatenating or interpolating the embeddings with the regular expression and gazetteer feature vector and the context tag distribution feature vector to generate a set of feature vectors, generating an encoded form of the utterance based on the set of feature vectors, generating log-probabilities based on the encoded form of the utterance, and identifying one or more constraints for the utterance.

Abstract: Techniques for providing and implementing a single skill associated with custom functionality and system-provided functionality are described. The skill may be used to invoke functionality in response to a user input without requiring a user remember exact formulations to cause the functionality to be performed. The skill may be associated with more than one domain. For example, the skill may be associated with custom sample user inputs (corresponding to the custom functionality) that correspond to a custom domain while the skill may also be associated with system-provided sample user inputs (corresponding to the system-provided functionality) associated with a non-custom domain.

Abstract: A system and method for analyzing and explaining the temporal evolution of policies. The analysis of the temporal evolution of policies includes determining semantically related changes in the text of the corresponding policies over time and across the different versions of the (two or more) policy documents. An explanation of the temporal evolution of policies is provided consisting of human interpretable information relating each change to one or more events and/or contextual data. A next steps policy change is also suggested that includes a set of policy conditions based on potential future changes of similar policies and contextual data and event data. The suggestions are related to the policy relevant to the target cohort or individual. The system adapts machine learned models by receiving user feedback provided on an analysis of the correctness of the temporal evolution of policies, corresponding explanations and suggestions of the next actions.

Abstract: System and method for aspect-level sentiment classification. The system includes a computing device, the computing device has a processer and a storage device storing computer executable code. The computer executable code is configured to: receive a sentence having a labeled aspect term and context; convert the sentence into a dependency tree graph; calculate an attention matrix of the dependency tree graph based on one-hop attention between any two nodes of the graph; calculate multi-head attention diffusion for any two nodes from the attention matrix; obtain updated embedding of the graph using the multi-head diffusion attention; classify the aspect term based on the updated embedding of the graph to obtain predicted classification of the aspect term; calculate loss function based on the predicted classification and the ground truth label of the aspect term; and adjust parameters of models in the computer executable code based on the loss function.

Abstract: According to an aspect of an embodiment, operations include receiving a set of NL descriptors and a corresponding set of PL codes. The operations further include determining a first vector associated with each NL descriptor and a second vector associated with each PL code, using language models. The operations further include determining a number of a set of semantic code classes to cluster the set of PL codes into the set of semantic code classes, based on the number, the first vector, and the second vector. The operations further include training a multi-class classifier model to predict a semantic code class, from the set of semantic code classes, corresponding to an input NL descriptor. The operations further include selecting an intra-class predictor model based on the predicted semantic code class. The operations further include training the intra-class predictor model to predict a PL code corresponding to the input NL descriptor.

Abstract: This disclosure describes techniques and architectures for evaluating conversations. In some instances, conversations with users, virtual assistants, and others may be analyzed to identify potential risks within a language model that is employed by the virtual assistants and other entities. The potential risks may be evaluated by administrators, users, systems, and others to identify potential issues with the language model that need to be addressed. This may allow the language model to be improved and enhance user experience with the virtual assistants and others that employ the language model.

Abstract: In some embodiments, a knowledge graph generation system extracts noun-phrases from sentences of a knowledge corpora and determines the relations between the noun-phrases based on a relation classifier that is configured to predict a relation between a pair of entities without restricting the entities to a set of named entities. The knowledge graph generation system further generates a sub-graph for each of the sentences based on the noun-phrases and the determined relations. Nodes or entities of the sub-graph represent the non-phrases in the sentence and edges represent the relations between the noun-phrases connected by the respective edges. The knowledge graph generation system merges the sub-graphs to generate the knowledge graph for the knowledge corpora.