Techniques are disclosed for extracting and applying structured debugging knowledge to resolve software bugs. A computing system processes electronic representations of software bug reports, including comments describing reasoning processes for debugging. A first neural language model extracts structured reasoning paths from the reports, which are used to fine-tune a second neural language model. When a new bug is encountered, the second model generates suggested actions or deploys a fix. Alternatively, structured reasoning paths are embedded into a vector database, enabling similarity searches against new bug descriptions. Retrieved reasoning paths are submitted to a neural language model to generate debugging recommendations or automated fixes. A hierarchical structure preserves relationships between reasoning paths, facilitating reasoning path retrieval and refinement. The system supports tree reconstruction, relational database storage, and hierarchical summarization of debugging knowledge to improve software maintenance and reduce resolution time.
As discussed herein, a generative AI (GenAI) tool is improved by using multiple LLMs, at least one of which is allowed to access tools that provide additional information to improve the quality of results. The output from a first LLM is provided as part of a prompt to a second LLM for review. If the second LLM rejects the results provided by the first LLM, a revised prompt for the first LLM is generated. This process is repeated until the second LLM approves a response by the first LLM. The functionality of the GenAI tool may also be expanded by allowing it to respond to a user query with a sequence of scenarios instead of a single scenario.
In an example embodiment, hallucinations in LLMs are reduced by incorporating specific training data that includes question/answer pairs where the answer in the training data is some variation of "I cannot answer this question." This technique involves constructing questions that cannot be answered due to unknown factual knowledge or logical questions that cannot be answered due to missing information. By training the LLM with such data, the model learns to recognize when it lacks the necessary information to provide a correct answer, thereby reducing the likelihood of generating plausible-sounding but incorrect responses. The described technique enhances user trust in LLMs by minimizing the risk of decisions being made based on incorrect information.
The described examples offer a solution by providing a design-time environment and a run-time environment. The design-time environment helps developers create and maintain configuration and other information that can be formed into chatbot runtime capabilities more easily. Developers can define what data can be accessed and how the data is to be filtered or displayed. The run-time environment is where the actual interaction with the user happens. When a user inputs a query, the system translates the query into a structured format that the ERP system can understand. This involves several steps, including identifying a capability that matches the query, generating a prompt for a large language model (using the capability), receiving a query, executing the query, and then processing the results to present them to the user.
In an example embodiment, a solution is provided that allows users to interact directly with underlying object data lake storages in hyperscalers when accessing user data. This feature is known as direct access. Use of this solution increases performance of the user systems and the cloud systems, without impacting the functionality nor increasing the complexity. This is true even for Spark users, where the driver abstracts the communication path between the Spark application and HDL files. This results in an opt-in solution that allows users to reduce their cost by increasing complexity.
In an example embodiment, additional metadata is attached to a delta tables, via delta shares, in the form of a common schema notation (CSN) entity. When a request is received from a client for a share of a range of delta tables where the range comprises the delta table to which the additional metadata is attached, the most recent CSN entity for each of the delta tables in the range are aggregated and returned to the client.
Methods, systems, and computer-readable storage media for receiving a first job with a first time-series of a first type of historic resource utilization and a second time-series of a second type of historic utilization, receiving a second job with a third time-series of the first type of historic resource utilization and a fourth time-series of the second type of historic utilization, determining a first correlation coefficient between the first time-series and the third time-series, determining a second correlation coefficient between the second time-series and the fourth time-series, combining the first correlation coefficient with the second correlation coefficient to generate a first total correlation coefficient, and in response to the first total correlation coefficient being below a threshold, transmitting the first job and the second job as a first job pair to a first executor of the plurality of job executors to be executed concurrently by the first executor.
A computer implemented method can receive n training samples including sample values corresponding to m attributes and respective target values (n and m are positive integers), duplicate the sample values corresponding to the m attributes, pack the sample values into ciphertexts based on a batching option, and train a regression tree using the ciphertexts. The training is configured to encrypt the regression tree through homomorphic operations on the ciphertexts.
H04L 9/00 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité
G06N 5/01 - Techniques de recherche dynamiqueHeuristiquesArbres dynamiquesSéparation et évaluation
H04L 9/06 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité l'appareil de chiffrement utilisant des registres à décalage ou des mémoires pour le codage par blocs, p. ex. système DES
9.
Preserving tabular data integrity for query processing system
System, method, and various embodiments for a tabular data integrity and query processing system are described herein. An embodiment operates by receiving a query to be executed against a knowledgebase. One or more keywords are identified from the query, and a vector search is performed against the knowledgebase based on the one or more keywords, the knowledgebase including documents that have been divided into a plurality of chunks. A subset of chunks related to generating an answer for the query are identified based on the vector search, the subset including a first chunk with a table ID. A table image corresponding to the table ID is identified. A prompt is generated instructing a language model to generate the answer to the query based on the subset of chunks, including the first chunk and the table image. The answer is provided.
Disclosed herein are system, method, and computer program product embodiments for implementing database queries using a major commit time stamp (CTS). An embodiment operates by receiving a data query indicating a predetermined delay from a user equipment (UE) and retrieving a major CTS from a memory. The data query corresponds to data in the memory. The major CTS indicates an age of the data. The embodiment determines that the major CTS is within the predetermined delay. In response to determining that the major CTS is within the predetermined delay, the embodiment transmits the data corresponding to the major CTS to the UE.
Generative AI (GenAI) applications make use of a prompt to guide the output of the Large Language Model (LLM). The prompt is generally composed of a static template with placeholders for input data, which are populated when the LLM is invoked. A solution is presented herein for generating synthetic data directly from the use case prompt template, reducing the dependency on actual data collection processes across a wide range of task domains. In some example embodiments, two main steps are performed. First, the LLM is asked to generate personas. Using the personas, the LLM is asked to generate input data for the placeholders of a prompt template. The responses for multiple personas are aggregated and de-duplicated, resulting in a synthetic set of values for the placeholder of the prompt template.
Methods, systems, and computer-readable storage media for querying database systems using queries generated by prompting of LLMs to generate query statements from natural language, where a LLM system provides a set of query statements in response to user input that is provided in natural language, the set of query statements includes a first query statement provided using an in-context prompt, a second query statement provided using a chain-of-thought (CoT) prompt, a third query statement provided using a contrastive CoT prompt, and a fourth query statement provided using a mixture-of-experts (MoE) prompt, and a query statement is selected from the set of query statements and is used to query a database system, which returns a query result responsive to the query statement.
During machine learning training, encoding and decoding are decoupled into two different processes. In a first process, an untrained version of an encoder is trained by optimizing encoder parameters in order to increase an amount of stochastic dependence between a vector embedding set and ground truth data, where the vector embedding set is generated by the untrained version of the encoder. In a second process, an untrained version of a decoder is trained with the vector embedding set generated by the trained version of the encoder. Output data generated by a trained version of the decoder is provided to one or more software applications to enable the one or more software application to perform one or more tasks.
Embodiments of the present disclosure include techniques for improving dynamic log level management. In one embodiment, a plurality of logging systems in a plurality of applications stores a plurality of log topics in a data storage system, each log topic having a first log level. Different log levels produce different amounts of information for log messages. A first logging system in a first application receives a first instruction to change the first log level of a first log topic to a second log level, having greater level of detail, changes the first log topic to the second log level in the application and the data storage system. The plurality of applications periodically detects modification of the first log topic data and change the first log level for the first log topic to the second log level in the plurality of applications excluding the first application.
A computer-implemented method can receive a submitted vulnerability specifying a vulnerability type and steps to reproduce a suspected bug of a software, determine a first base model score based on keywords appeared in the vulnerability form, determine a second base model score based on vulnerability type-specific contextual information extracted from the vulnerability form, determine a first probability modifier based on features identified from the vulnerability form, determine a second probability modifier based on mapping any file attachment associated with the vulnerability form to a corresponding step, determine a third probability modifier based on pairing and grouping selected steps, determine a vulnerability score based on the first and second base model scores, and the first, second, and third probability modifiers, and classify the vulnerability form as valid or invalid based on the vulnerability score. Related systems and software for implementing the method are also disclosed.
G06F 21/57 - Certification ou préservation de plates-formes informatiques fiables, p. ex. démarrages ou arrêts sécurisés, suivis de version, contrôles de logiciel système, mises à jour sécurisées ou évaluation de vulnérabilité
17.
UPGRADE OF NON-IMPORTED CONFIGURATION DATA IN A VERSION REPOSITORY DURING SOFTWARE CHANGES
For a first table storing configuration data for an enterprise resource planning software application, the first table is renamed with a second name. Next, a second table is created as a separate version of the first table. Then, a first view is created of the first table, where the first view is a union of the first table and the second table, where the union is implemented with a first select clause for the first table for all clients other than a first client, and where the union is implemented with a second select clause for the second table for only the first client. Finally, one or more queries are executed by accessing first data via the first view.
In an example embodiment, an COMPRESSED schema representation is used for input to LLMs to reduce the token count of such input, thus decreasing inference time and cost in contrast to previous LLM-based solutions. This COMPRESSED schema representation is then used by the LLM when presented with a prompt to generate an intermediate representation of computer code, leading to more COMPRESSED LLM generation of the intermediate representation.
In some implementations, the techniques may include receiving a first input indicating a request for modifying a displayed translation. In addition, the techniques may include receiving a second input indication a portion of the displayed translation. The techniques may include determining domain information for the portion of the displayed translation. The techniques may include accessing a data repository for a modified translation based at least in part on the domain information. When the modified translation is not stored in the data repository techniques may include receiving the modified translation from an expert user and storing the modified translation and the domain information in the data repository. The techniques may include modifying the displayed translation using the modified translation. The techniques may be performed on one or more computing devices. The techniques may be performed by a system or stored as a series of instructions on a computer-readable tangible medium.
G06F 40/166 - Édition, p. ex. insertion ou suppression
G06F 3/0481 - Techniques d’interaction fondées sur les interfaces utilisateur graphiques [GUI] fondées sur des propriétés spécifiques de l’objet d’interaction affiché ou sur un environnement basé sur les métaphores, p. ex. interaction avec des éléments du bureau telles les fenêtres ou les icônes, ou avec l’aide d’un curseur changeant de comportement ou d’aspect
G06F 40/58 - Utilisation de traduction automatisée, p. ex. pour recherches multilingues, pour fournir aux dispositifs clients une traduction effectuée par le serveur ou pour la traduction en temps réel
20.
COMBINING STRUCTURED AND UNSTRUCTURED DATA FOR RAG
A system and method include identification of a plurality of stored multi-dimensional numerical vectors similar to a first multi-dimensional numerical vector representing the received text, identification of a first plurality of documents associated with respective ones of the identified plurality of stored multi-dimensional numerical vectors, determination of a second plurality of the first plurality of documents which are associated with a respective metadata value of the first metadata field, and prompting of a text generation model to determine relevancies of each metadata value to the received text based on the second plurality of documents.
A computing system receives a machine learning model that has been trained using historical event data to predict deviations between scheduled and actual event outcomes. Example training include temporal attributes, contextual attributes, or observed deviation patterns. The system processes attributes of a planned event using the trained model to generate a predicted deviation. If the predicted deviation exceeds a configurable threshold, the system initiates at least one of: generating an alert, blocking execution of related operations, adjusting scheduling parameters, modifying execution conditions, receiving user input to override an alert or adjust scheduling data, or dynamically updating threshold parameters based on historical deviation trends or real-time conditions. The technology can be integrated into workflow automation, scheduling optimization, and decision-support applications to improve planning accuracy, operational efficiency, and predictive reliability across various domains.
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
A computer-implemented method can receive a log message comprising a sequence of L characters, wherein L is a positive integer, generate embeddings for the sequence of L characters, and generate feature vectors for N tokens based on the embeddings. A token represents M consecutive characters in the log message, M is an integer greater than one, and N is a smallest integer greater than or equal to L divided by M. The method can predict N binary coded values based on the feature vectors for N tokens, and generate a sequence of L parameter masks based on the N binary coded values. A parameter mask indicates that a corresponding character in the log message is a static character or a variable character. Related systems and software for implementing the method are also disclosed.
Methods, systems, and computer-readable storage media for receiving, by a first LLM-based agent, a data schema description, a task description, and a set of examples, prompting, by the first LLM-based agent, a first LLM using a first prompt that is provided based on the data schema description, the task description, and the set of examples, the first LLM returning a set of metrics responsive to the first prompt, filtering data from a query document and data from a target document using the set of metrics, prompting, by a second LLM-based agent, a second LLM using a second prompt that is provided based on unstructured data of the query document, the second LLM returning a set of property-value pairs responsive to the second prompt, providing a merged query document, and processing the merged query document and the target document using a global ML model to generate a set of results.
System, method, and various embodiments for an application modification system, are described herein. An embodiment operates by receiving a user instruction to modify a data object, and identifying a plurality of applications to which the user has access. The data object is compared to the specification for at least a subset of the plurality of applications, and a first specification that includes the data object is identified. One or more requirements for performing the modification to the data object in accordance with first specification are identified. Feedback corresponding to the one or more requirements is received from the user. The API call to the first application is generated, and the data object of the first application is modified in accordance with the generated API call.
A system and method including specifying, for a defined integration flow, a first entity type of at least one data object associated with at least one producer system and a second entity type of at least one data object associated with at least one consumer system are logically equivalent; receiving, from the at least one producer system and the at least one consumer system, metadata including key fields of the at least one data object of the first entity type from the at least one producer system and a source identifier of the at least one data object of the second entity type from the at least one consumer system that references the at least one producer system; and determining, based on the metadata, a same value for an object identifier for each of the at least one first entity type and the second entity type, respectively.
Using a data analysis activity (DAA) definition, a DAA associated with a software application is triggered. An instance selector query is executed to generate a set of instance values as input for a data query. A data query to generate a data set is executed using instance values of the set of instance values. Using the data set, an instruction for an artificial intelligence (AI) engine is computed. A result based on the instruction for an AI engine is received from the AI engine. The result based on the instruction for an AI engine is stored into an AI Result History Store. Prior results from earlier DAA executions is read from the AI Result History Store. A notification to a defined target audience is sent using the software application.
Systems and methods include reception of a request from an application for text generation including a scenario identifier and a payload, determination of a stored scenario definition associated with the scenario identifier, determination of a prompt template definition from the scenario definition, determination of a model deployment from the scenario definition, generation of a prompt based on the prompt template definition and the payload, transmission of the prompt to the model deployment, reception of a response to the prompt from the model deployment, and return of the response to the application.
Systems and methods described herein relate to data expansion techniques for machine learning. A generative model trained on an initial dataset generates candidate images based on input images and corresponding class labels. A mask is applied to each candidate image to obtain masked images that are then reconstructed using the generative model. Candidate images are filtered by removing those that are too dissimilar when compared to their reconstructed versions. The remaining candidate images are combined with the initial dataset to form an expanded dataset.
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
G06T 5/77 - RetoucheRestaurationSuppression des rayures
G06V 10/72 - Préparation de données, p. ex. prétraitement statistique des caractéristiques d’images ou de vidéos
G06V 10/74 - Appariement de motifs d’image ou de vidéoMesures de proximité dans les espaces de caractéristiques
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
A system includes monitoring of performance metrics associated with one or more predefined system performance requirements to obtain historic performance metric data, predicting from the historic performance metric data a quantitative compliance metric indicating compliance with the one or more predefined system performance requirements for a plurality of future time frames, estimating an impact of a proposed maintenance activity on the quantitative compliance metrics, calculating, for each future time frame, a risk factor indicative of a likelihood of violation of each predefined system performance requirement, the risk factor being based on the predicted quantitative compliance metric and the estimated impact of the proposed maintenance activity, computing, for each future time frame, a combined risk factor from the risk factors, obtaining a selected time frame for performing the maintenance activity based on the combined risk factors, and initiating the maintenance activity in the selected time frame.
A computing system and methods for aligning process graph representations and processing multi-representational datasets are disclosed. A first process graph representation is aligned with a second reference process graph representation using a matcher implemented in a matcher code module. A similarity evaluation is performed at one or more levels of process abstraction, and process-wide metrics are generated to assess overall alignment quality. A user interface renders the metric results and allows user modification of alignment parameters or element correspondences. Additionally, a dataset with at least two representations is processed to generate embedding vectors using different embedding techniques. The embedding vectors are combined into a hybrid vector representation, which is analyzed to determine a similarity measure relative to an input query. Search results are rendered based on the similarity measure. The disclosed techniques improve alignment accuracy, computational efficiency, and the relevance of search results in multi-representational datasets.
A system and method include receiving a text description of a service, generating a prompt based on the text description to request object properties of the service, prompting a text generation model with the prompt to generate the object properties of the service, receiving the generated object properties from the text generation model, converting the generated object properties from a first format to a second format, and instructing an object generator to generate artifacts of the service based on the converted object properties.
G06F 40/35 - Représentation du discours ou du dialogue
G06F 40/143 - Balisage, p. ex. utilisation du langage SGML ou de définitions de type de document
H04L 51/02 - Messagerie d'utilisateur à utilisateur dans des réseaux à commutation de paquets, transmise selon des protocoles de stockage et de retransmission ou en temps réel, p. ex. courriel en utilisant des réactions automatiques ou la délégation par l’utilisateur, p. ex. des réponses automatiques ou des messages générés par un agent conversationnel
A system and method include reception of a prompt, determination that the prompt is not semantically similar to any of a plurality of text generation model prompts, in response to the determination, prompt a first text generation model to determine whether the prompt is malicious, in response to a determination that the prompt is not malicious, prompt a second text generation model with the prompt to determine a first prompt output, prompt a third text generation model to determine whether the first prompt output is malicious, and, in response a the determination that the third text generation model determined that the first prompt output is not malicious, return the prompt output in response to the prompt.
In an event-based introspection protocol for landscape and event exchange monitoring, an event client subscribes as an event client with a messaging infrastructure. The event client publishes a client connected message to the messaging infrastructure. The event client publishes a client alive ping message to the messaging infrastructure. The event client consumes event data from one or more other event clients subscribed to the messaging infrastructure from the messaging infrastructure. The event client processes status information based on the consumed event data from one or more other event clients. The event client adds the processed status information to a status table. The event client sends a client disconnected message to the messaging infrastructure to disconnect from the messaging infrastructure.
An enterprise Retrieval-Augmented Generation (“RAG”) pipeline data store may contain electronic records that represent RAG pipelines, each record including a pipeline identifier and at least one tuning parameter. A Generative Artificial Intelligence (“GenAI”) launchpad platform, associated with at least one Large Language Model (“LLM”), includes an enterprise RAG pipeline engine that accesses information associated with a first RAG pipeline that was created by a first tenant. The enterprise RAG pipeline engine receives from a user an adjustment to a tuning parameter of the first RAG pipeline (e.g., for document ingestion, chunk tuning, embed tuning, similarity search, and/or context tuning) and automatically calculates an overall pipeline credibility score for the adjusted first RAG pipeline. The pipeline engine may then display the overall pipeline credibility score to the user and store information about the adjusted first RAG pipeline into the RAG pipeline data store.
G06F 16/215 - Amélioration de la qualité des donnéesNettoyage des données, p. ex. déduplication, suppression des entrées non valides ou correction des erreurs typographiques
G06F 16/21 - Conception, administration ou maintenance des bases de données
In an example embodiment, a filtering system is introduced to optimize the organization and processing of LLM request. By carefully categorizing queries based on their input prompt lengths and anticipated output token numbers, the workflow is streamlined prior to submission to an LLM inference server. Each bucket corresponds to a different range of number of tokens (both input and output combined). Each bucket also has a size, indicating the maximum number of requests that can be placed in the bucket. Each request is assigned to a bucket when it is received, and when a bucket is filled the requests in the bucket are batched together and send to the LLM for processing.
The present disclosure involves systems, software, and computer implemented methods for system provisioning. One example method includes receiving a request for provisioning of a software solution. A knowledge graph is accessed that comprises a graph of object types of sample data and dependency information for the object types. The knowledge graph is traversed to identify object types and object dependencies included in the software solution. An interface is invoked to determine whether a data repository includes, for each identified object type, data for the identified object type. In response to determining that the data repository includes data for each identified object type, the interface is invoked to iteratively retrieve, in a dependency order determined based on the object dependencies of the identified object types, data from the data repository of each identified object type. Retrieved data is provided, in the dependency order, for deployment during provisioning of the software solution.
In an example embodiment, a mechanism is provided to allow an LLM to generate multiple text blocks in which different grammars are strictly enforced with a single invocation. This mechanism defines a set of trigger tokens and end tokens. When a trigger token is encountered by the LLM, a strict grammar referenced by the trigger token is begun to be enforced and this enforcement ends when an end token is encountered. Between the time an end token is encountered, and another trigger token is encountered, no grammar is strictly enforced. By including multiple types of such trigger token/end token pairs, it becomes possible for the LLM to generate texts having different strictly enforced grammars in a single invocation.
A computer-implemented method can receive a vulnerability report for a software. The vulnerability report specifies a vulnerable library used by the software and a path of the vulnerable library within the software. The method can generate a summary of the vulnerability report using a generative artificial intelligence (AI) model, retrieve, from a bug report database, a set of relevant bug reports specifying the vulnerable library, generate a synopsis for the set of relevant bug reports using the generative AI model, generate multiple preliminary decisions on validity of the vulnerability report using the generative AI model based on the summary of the vulnerability report and the synopsis for the set of relevant bug reports, and generate a final decision on validity of the vulnerability report based on the multiple preliminary decisions. Related systems and software for implementing the method are also disclosed.
G06F 21/00 - Dispositions de sécurité pour protéger les calculateurs, leurs composants, les programmes ou les données contre une activité non autorisée
G06F 21/57 - Certification ou préservation de plates-formes informatiques fiables, p. ex. démarrages ou arrêts sécurisés, suivis de version, contrôles de logiciel système, mises à jour sécurisées ou évaluation de vulnérabilité
A system associated with a cloud computing environment may include an UPSERT batch optimization engine that identifies an UPSERT VALUES (?, ?, ?, …) WITH PRIMARY KEY case. The optimization engine may then receive batch parameters for the UPSERT VALUES (?, ?, ?, …) WITH PRIMARY KEY case and create a temporary table with the received batch parameters. According to some embodiments, the optimization engine can then execute a single left outer join with the temporary table and an UPSERT target table. The optimization engine may then fetch a result of the single left outer join. A Data Manipulation Language (“DML”) execution engine dispatches a set of rows to be inserted and a set of rows to be updated. A partition-wise insert and update run can then be executed in accordance with the set of rows to be inserted and the set of rows to be updated.
Techniques and solutions are provided for configuring a process mining template to include events for particular process mining enhancements. For example, users can select to add events relevant to specific industries or for value drivers. The events are associated with database queries that can be executed to determine occurrences of events. Events for process mining enhancements can be determined by clustering events from one or more existing process mining templates. A sample of an entity's data, in a database, can be processed prior to deploying a process mining template to determine overlap between currently defined events of the client and the events for the process mining enhancements, or to compare an entities metrics to reference values.
Some embodiments provide a non-transitory machine-readable medium that stores a program. The program may receive a set of data from a data source. The program May generate a plurality of time series data based on the set of data. The program may determine a subset of the plurality of time series data as anomalies. The program may provide notifications indicating that the subset of the plurality of time series data are anomalies.
G06F 18/22 - Critères d'appariement, p. ex. mesures de proximité
G06F 18/2113 - Sélection du sous-ensemble de caractéristiques le plus significatif en classant ou en filtrant l'ensemble des caractéristiques, p. ex. en utilisant une mesure de la variance ou de la corrélation croisée des caractéristiques
G06F 123/02 - Types de données dans le domaine temporel, p. ex. des données de séries temporelles
A central framework unifies the data protection and privacy domain with the business domain using application objects and business scenarios. Application objects are linked to a particular data category, a name of a data object, and primary key attributes of the data object. Each data category is linked to a purpose. And business scenarios include a set of application objects and a sequence for the set. Worklist entries are generated in response to a particular instances of application objects being created or modified. For each worklist entry, purposes are determined for the particular application object using purpose assignment rules for the particular application object. Each of the determined purposes for the particular application object are stored in a purpose assignment table. Then each determined purpose are proposed for master data to be used across the plurality of different application.
Systems and methods described herein relate to hybrid machine learning techniques for forecasting master data, such as scrap data. A hybrid machine learning model includes a first machine learning model (e.g., balanced random forest classifier) that generates a prediction with respect to whether or not a manufacturing process will result in scrap generation. If the first machine learning model predicts that no scrap will be generated in the manufacturing process, then an output prediction is that no scrap will be generated in the manufacturing process. If the first machine learning model predicts that scrap will be generated in the manufacturing process, then a second machine learning model (e.g., gradient boost regressor) predicts a value of scrap percentage. The value of scrap percentage is provided as an output prediction of a percentage of scrap that will be generated in the manufacturing process.
G06F 16/215 - Amélioration de la qualité des donnéesNettoyage des données, p. ex. déduplication, suppression des entrées non valides ou correction des erreurs typographiques
45.
SOFTWARE DEVELOPMENT OBJECT ARTIFICIAL INTELLIGENCE TRAINING DATA GENERATION
The present disclosure involves systems, software, and computer implemented methods for artificial intelligence (AI) training data generation. A method includes identifying a request to generate training data based on software development objects of a software development system. A plurality of exporters each configured for a given object type are invoked. Each exporter invokes a respective interface of the software development system to iterate, in a shared object repository, over objects of an object type to retrieve object data and object metadata for instances of the object type. Object data and object metadata are received from each exporter. Received object data and metadata are stored in a first format in an exported data repository. AI training data is generated by transforming the object data and metadata in the first format to a second format suitable for training AI models. The AI training data is provided to at least one AI system.
A computer-implemented system includes: a component of a local development environment generating a request to complete a source code and a code suggestion system. The request includes the context of the local development environment. The code suggestion system parses the context to define dependencies of code snippets of the source code under development. The code suggestion system determines corresponding databases including additional code objects associated to code objects of the source code under development identified by the dependencies of the source code. The code suggestion system retrieves the additional code objects and generates a prompt using a portion of the additional code objects. The prompt is formatted for minimizing a prompt size and provided to the prompt to a large language model to generate a response including a suggested code for completing the source code under development.
A system and method include determination of input fields of automation scripts, acquisition of metadata of database object fields, prompting of a text generation model using a chain-of-thoughts prompt, the input fields and the acquired metadata to determine mappings between the input fields and the database object fields, prompting of an embedding model to generate embeddings based on each input field, associating each embedding with each mapping that includes the input field on which the embedding was generated, identification of a first input field of an automation script, prompting of a second embedding model to generate a first embedding based on the first input field, searching for embeddings similar to the first embedding, identification of a candidate mapping associated with each of the embeddings, and determination of a first database object field to bind to the first input field based on the identified candidate mappings.
A computer-implemented system, includes: an application programming interface (API) generating a request to complete code documentation for a source code including code entities and a software code suggestion system. The software code suggestion system processes the request to identify a code type of the source code by performing a syntax analysis and a semantics analysis of a structure of the code entities. The software code suggestion system retrieves a matching code for the source code corresponding to the code type of the source code. The software code suggestion system retrieves a context of the matching code for the source code. The software code suggestion system generates a prompt using the matching code and the context. The software code suggestion system provides the prompt to a large language model of the large language model type to generate a response including a code documentation.
In an example embodiment, constrained generation during LLM inference is performed using a multi-layered approach to incremental parsing. This constrained generation process is able to more effectively generate computer code for sentence-based programming languages. Specifically, it is designed to wait until a full statement is generated prior to analyzing the syntactical correctness of a statement, because the meaning of a token in a sentence-based programming language can change its meaning based on later tokens in the same statement. Furthermore, the external (block) structure of statements can be analyzed by looking only at the first word of statements without looking into the statements more closely.
In an example embodiment, a large language model (LLM) is used to capture semantic relationships between entities so that machine learning techniques can be used to analyze patterns and relationships of entities and dependencies. This includes analyzing version requirements and dependencies among software products. The system is trained to recognize compatibility patterns and understand the impact of product upgrades on dependencies.
Methods, systems, and computer-readable storage media for receiving an image and a reference color scheme, the image being generated by an AI model, the reference color scheme having populated a prompt that the image was generated in response to, for each pixel in the image, generating a color space vector representative of a color represented by the pixel to provide a set of vectors, processing the set of vectors to define a set of clusters, each cluster representative of colors of a sub-set of vectors, for each cluster in the set of clusters and for each color in the reference color scheme, determining a similarity score that is included in a set of similarity scores, calculating a color alignment metric based on the set of similarity scores, determining a color alignment result based on the color alignment metric, and executing one or more tasks responsive to the color alignment result.
G06V 10/56 - Extraction de caractéristiques d’images ou de vidéos relative à la couleur
G06V 10/74 - Appariement de motifs d’image ou de vidéoMesures de proximité dans les espaces de caractéristiques
G06V 10/762 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant le regroupement, p. ex. de visages similaires sur les réseaux sociaux
52.
EVALUATING LARGE LANGUAGE MODEL GENERATED CODE USING APPLICATION SERVER CONTEXT
The present disclosure involves systems, software, and computer implemented methods for evaluating code generation. A method includes identifying a code artifact for a benchmark task. A portion of code for the benchmark task is determined in a first copy of the artifact. A second copy of the artifact is automatically generated by replacing, in the first copy of the artifact, the portion of code with a fill-in marker. A prompt and at least a portion of the second copy of the artifact are provided to a model. The prompt instructs the model to generate code to replace the fill-in marker. A third copy of the artifact is automatically generated by replacing, in the second copy of the artifact, the fill-in marker with model-generated code. The model-generated code is evaluated by executing an executable version of the third copy of the artifact.
A prompt optimization interface retrieves semantic information characterizing a source code under development that includes code entities. A code suggestion system coupled to the prompt optimization interface, generates, from the semantic information, a dependency graph exposing relations and dependencies between the code entities. The code suggestion system generates a ranked list of code entities indicative of a relevance of each code entity of the code entities in the dependency graph based on a relevance to a query code entity. The code suggestion system minimizes the dependency graph using the ranked list of code entities and a context of the query code entity for generating a minimized dependency graph to be within a set window. The code suggestion system converts the minimized dependency graph into a prompt format processable by a large language model to generate matching code for the source code under development.
A non-user-interface (non-UI) test automate associated with a web application test case includes a sequence of state-change requests. The state-change requests and responses thereto are captured during recording of a UI test automate for the test case or adapted from an existing UI test automate for the test case, whereas non-state-change requests of the corresponding UI test automate are excluded from the non-UI test automate. Responsive to a prompt to perform a non-UI automated test for the test case, mappings between properties of the state-change requests, responses, and test data are determined. At runtime of the non-UI test automate, the mappings are referenced to preserve sequence and data dependencies among the state-change requests. Pop-up window content received in network responses during execution of the non-UI test automate is classified by type using a classification machine learning model and handled based on the determined type.
Systems and methods include reception of a request to modify an instance of a reuse component and, in response to the request, determine an interface entity of the reuse component and a property of the reuse component which stores an identifier of a host object instance of the reuse component instance, call a first predefined operation with the type of the host object and an identifier of the host object instance to set a lock on the host object instance, create a data container of key field values of the host object instance based on data types of the key fields, and call a second predefined operation with the type of the host object and the data container to check an authorization to modify the host object instance.
In an example embodiment, a novel data structure called a “mind graph” is introduced, which defines all of the different possible combinations of tasks within a software project, as well as the potential flows among the tasks. This mind graph is then passed as input, along with a prompt generated by natural language input from a user, to an LLM. The LLM uses the mind graph and the prompt to generate an execution plan which defines which tasks from the mind graph will be generated and the path the flow takes though those tasks. This execution plan is then presented to the user in a graphical user interface that allows the user to edit the execution plan. The edited execution plan is then submitted to the LLM to generate the actual code for each of the tasks. This code is again presented to the user to accept, or modify, the code for each of these tasks.
Methods, systems, and computer-readable storage media for retrieving data from a data store, the data being in a storage format and including multiple dimensions in a hierarchy, at least one dimension having a sub-hierarchy, converting the data from the storage format to an analytics format including a set of nodes, each node representing a dimension of the multiple dimensions, a set of hierarchy characters, one or more hierarchy characters separating two or more nodes to represent a hierarchical relationship between nodes, and an attribute delimiter to separate attribute values of a node, generating a prompt that references the data in the analytics format, transmitting the prompt to a LLM system, and receiving a response to the prompt from the LLM system.
Systems and methods described herein relate to automated control of software development pipeline progression. Examples herein provide for automated validation and progression through multiple sequential stages without manual intervention between states. A user interface of a continuous integration tool receives a user selection of a target state in a software development pipeline. Based on the selected target state, the system automatically progresses an increment through a sequence of states by executing automated tests associated with a current state, detecting when a current state precedes the target state, and automatically transitioning to a next state after successful completion of the automated tests if the current state precedes the target state. In some examples, the automated progression repeats until the current state reaches the target state, with results data presented via the user interface indicating at least the current state and the successful completion of the automated tests.
A blackboard data store contains records representing a plurality of AI agent operation results associated with the system (including an operation identifier). A decision support platform, coupled to the blackboard data store and being associated with at least one LLM, may receive a decision support request from a user associated with the system and determine a series of operations associated with the decision support request. A coordination agent may arrange for the series of operations to be performed by a plurality of AI agents, with operation results being recorded in the blackboard data store. Decision support information can then be presented to the user in response to the decision support request. According to some embodiments, the plurality of AI agents include a question planning and analysis agent, a research agent a decision option suggestion agent, a decision option evaluation agent, a critique agent, and/or a decision presentation agent.
Systems and methods select a candidate scenario of a plurality of scenarios in a scenario group associated with a hard level, based on a user query, generate a candidate selection rationale using a first machine learning model and analyze, using a second machine learning model, the user query, the candidate scenario, and the candidate selection rationale to generate a selection decision and selection decision feedback. The systems and methods further, until a selection decision is positive or a maximum number of iterations has been reached, select a new candidate scenario of the plurality of scenarios in the scenario group and generate a new candidate selection rational for based on the user query and the selection decision feedback, using the first machine learning model, and generate a new selection decision and new selection decision feedback based on the user query and new candidate selection rationale, using the second machine learning model.
Arrangements for holistic end-to-end process data automation operations are provided. Process data including entity-level data and group-level data may be identified. The entity-level data may include individual tasks, and the group-level data may include sets of tasks to be executed in a predefined sequence. The group-level data may be synchronized with the entity-level data by mapping respective content and configurations. A logical tree structure for the entity-level data and the group-level data may be generated. The generating may include dividing, via a topmost sub-hierarchy layer, the logical tree structure into a first subtree of nodes representing the entity-level data and a second subtree of nodes representing the group-level data. An instance may be generated from the logical tree structure based on one of the first subtree or the second subtree. A consolidated report unifying the entity-level data and the group-level data may be output based on the generated instance.
A database of text associated with different domains is maintained. Large language models (LLMs) are prepared for use in the different domains by providing the associated text to an instance of an LLM. Thus, using multiple instances of the same pre-trained LLM, domain-specific LLMs are generated. The text provided to the LLM instance may be selected based on an account identifier of the user accessing the LLM, the tenant accessing the LLM, a user selection of a domain, or any suitable combination thereof. A pool of prepared LLM instances may be generated before the access request is received. If a response provided by an LLM instance in a domain to a prompt was rejected by a user and additional information was received during the session to improve the response of the LLM instance, the additional information may be to the text used to prepare future LLM instances for the domain.
Methods, systems, and computer-readable storage media for executing semi-supervised training of a network backbone using unlabeled training data, the semi-supervised training including two or more iterations including selecting a batch of unlabeled training data, generating first predictions using a first ML model and second predictions using a second ML model, determining a contrastive loss based on the first predictions and the second predictions, the contrastive loss being determined based on a global normalized confusion matrix (NCM) and a global cluster matrix (CM), adjusting second parameters of the second ML model in response to the contrastive loss, and adjusting first parameters of the first ML model using the second parameters of the second ML model.
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
64.
CLOUD COMPUTING INTEGRATION SUITE MESSAGE RECORDER
A system associated with a cloud computing environment includes a message interceptor that intercepts details of messages of a process executed by an enterprise. The message interceptor then determines an occurrence of a process failure, and, responsive to the determination, automatically records details of the intercepted messages that may be relevant to the process failure. A collected message data store contains details of the intercepted messages that may be relevant to the process failure. A message recorder can then access the collected message data store and arrange for information about the recorded messages to be transmitted to a support team. In some embodiments, the message recorder also provides at least some of the information about the relevant messages to a LLM (e.g., via a prompt that includes some of a process code base), and a response includes possible troubleshooting information to fix the code base.
The present disclosure involves systems, software, and computer implemented methods for data privacy. One example method includes determining to perform a test of a first multiple-application landscape and a first data privacy integration service instance that manages data privacy integration of multiple applications in the first multiple-application landscape. A test work package is created in response to determining to perform the test of the first multiple-application landscape and the first data privacy integration service instance. The test work package is provided to applications of the first multiple-application landscape and test work package responses are received from applications of the first multiple-application landscape. The test work package responses are evaluated to determine a correctness of the first multiple-application landscape and the first data privacy integration service instance.
G06F 21/57 - Certification ou préservation de plates-formes informatiques fiables, p. ex. démarrages ou arrêts sécurisés, suivis de version, contrôles de logiciel système, mises à jour sécurisées ou évaluation de vulnérabilité
G06F 21/62 - Protection de l’accès à des données via une plate-forme, p. ex. par clés ou règles de contrôle de l’accès
66.
REDUCING PARTICIPANTS IN DATA PRIVACY INTEGRATION PROTOCOLS
The present disclosure involves systems, software, and computer implemented methods for data privacy. One example method includes receiving a first request to start a data privacy integration protocol for a first object instance. At least one second request is sent to at least one other service for information regarding which subset of applications of a multiple-application landscape have received a copy of the first object instance. A work package is created for the first object instance and the data privacy integration protocol and is sent to applications in the subset and not sent to landscape applications not in the subset. Work package responses are received from applications in the subset of applications. A data privacy integration protocol result is determined based on the work package responses and is sent in response to the first request to start the data privacy integration protocol.
The present disclosure involves systems, software, and computer implemented methods for data privacy. One example method includes identifying, at a data privacy integration service that manages data privacy integration, a data privacy integration protocol request. Application availability of applications in a multiple-application landscape is determined, including a determination that at least one application is not currently available for data privacy integration requests. A determination is made to proceed with data privacy integration protocol processing for the data privacy integration protocol request even though the at least one application is unavailable. A data privacy integration work package is sent to applications of the multiple-application landscape and data privacy integration work package responses are received. The data privacy integration work package responses are evaluated to determine a data privacy integration protocol result and the data privacy integration protocol result is provided in response to the data privacy integration protocol request.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
68.
SCHEDULING OF DATA PRIVACY INTEGRATION PROTOCOL PROCESSING BASED ON APPLICATION AVAILABILITY
The present disclosure involves systems, software, and computer implemented methods for data privacy. One example method includes logging data privacy integration protocol activity information. A data privacy integration protocol request having a context is received. Logged data privacy integration protocol activity information is determined based on the context, along with first work package timing information for a first application subset that differs from second work package timing information for a second application subset. A first work package is sent to applications in the first and second application subsets based on the first and second timing information, respectively. A data privacy integration protocol result is determined based on work package responses to the first and second work packages. The data privacy integration protocol result is provided in response to the first data privacy integration protocol request.
In some implementations, there is provided a computer-implemented method comprising receiving, by a processor, at least a first text-based input, determining, by the processor and using a first machine learning model, at least a first predicted result corresponding to at least the first text-based input, training a second machine learning model on at least the first text-based input and at least the first corresponding predicted result, determining, by the trained second machine learning model, an explanation for at least the first predicted result determined by the first machine learning model, and outputting, to a user interface of user equipment, the explanation for at least the first predicted result.
An enterprise documentation data store contains records representing a plurality of documents in an enterprise corpus (including a document identifier). A semantic search pipeline data store contains records representing semantic search pipelines (including a pipeline identifier and at least one tuning parameter). A GenAI validation platform can then identify at least one document in the enterprise documentation data store to be validated. The GenAI validation platform accesses information in the semantic search pipeline data store associated with a semantic search pipeline. It automatically performs a fact-based validation of the identified document using the semantic search pipeline to generate a validation report suggesting changes to the identified document. Embodiments may also arrange to automatically implement the suggested changes.
The present disclosure provides techniques for managing cookies in interactions involving multiple network domains. Upon receiving a request to set cookies for a second network domain while processing a web page of a first network domain, both partitioned and unpartitioned cookies are issued and stored. Having both types of cookies available allows for seamless operation regardless of the browser, first-party domain, or third-party domain support for different cookie handling mechanisms, including when traditional third-party cookies are not available. Policies can be set to determine when both cookie types should be set, or when only one type of cookie is needed. If a third-party domain does not support the issuance of both types of cookies, a web proxy can be used to perform the relevant operations.
This disclosure describes systems, software, and computer implemented methods for initiating a large language model (LLM) with a base prompt, the base prompt providing the LLM with a set of possible output actions, and constraining the LLM to respond with a thought response, output action, or an answer response; receiving, a natural language query; providing the natural language query to the LLM; receiving a first response from the LLM comprising a first token and a structured query; passing the structured query including the arguments to a database on behalf of the user; receiving a return from the database; generating a return token and providing the return token and the return from the database to the LLM; receiving a second response from the LLM, the second response comprising a second token indicating that the second response is an answer response; and providing the second response to the user.
In an example embodiment, a mechanism is provided to allow an LLM to generate multiple text blocks in which different grammars are strictly enforced with a single invocation. This mechanism defines a set of trigger tokens and end tokens. When a trigger token is encountered by the LLM, a strict grammar referenced by the trigger token is begun to be enforced and this enforcement ends when an end token is encountered. Between the time an end token is encountered, and another trigger token is encountered, no grammar is strictly enforced. By including multiple types of such trigger token/end token pairs, it becomes possible for the LLM to generate texts having different strictly enforced grammars in a single invocation.
Methods, systems, and computer-readable storage media for converting a first set of values of first digital content to a second set of values, dividing a sub-set of values of the second set of values into a set of blocks, applying a transform to each block to provide, for each block, a frequency domain representation, for each block in the set of blocks, embedding a sub-set of watermark bits based on the parameters, executing an inverse transformation of the set of blocks to provide a modified second set of values representative of an image with a digital watermark in the second color space, converting the modified second set of values to a modified first set of values representative of the image with the digital watermark in the first color space, and providing second digital content including the image with the digital watermark.
In an example embodiment, rather than use large language model (LLM) to directly generate desired computer code, an intermediate representation is generated by the LLM. The LLM is used to generate the portion of the computer code that cannot be computed programmatically (which may be called the “creative” part for purposes of the present disclosure). The intermediate representation can then be fed into a separate programmatic component that compiles the intermediate representation into compilable computer code. This fine-tuning may involve, for example, sanitizing the intermediate representation, enhancing the intermediate representation, and formatting the intermediate file, as well as modifying the intermediate representation based on a feature set.
G06F 16/3329 - Formulation de requêtes en langage naturel
G06F 16/383 - Recherche caractérisée par l’utilisation de métadonnées, p. ex. de métadonnées ne provenant pas du contenu ou de métadonnées générées manuellement utilisant des métadonnées provenant automatiquement du contenu
78.
GENERATIVE ARTIFICIAL INTELLIGENCE (GAI) REUSE SERVICE FOR INTEGRATION OF GAI INTO APPLICATION SCENARIOS
Methods, systems, and computer-readable storage media for receiving a scenario service request for a scenario of an application, determining a scenario flowchain represented in the scenario service request, retrieving a scenario flowchain configuration of the scenario flowchain from a database, the scenario flowchain configuration including a data object that defines a set of steps that are to be executed in an order, each step being associated with a step type and a set of parameters, executing steps in the set of steps, where at least one step is executed to prompt a large language model (LLM), and returning a result to the application, the result comprising a response from the LLM that is responsive to the prompt.
Methods, systems, and computer-readable storage media for receiving a first request parameter for each of the plurality of tenants, receiving a second request parameter for each of the plurality of tenants, assigning the plurality of tenants to an N plurality of tenant groups based on the first request parameter for each of the plurality of tenants, assigning each tenant in the N plurality of tenant groups to a server group in an M plurality of server groups based on the second request parameter for each of the plurality of tenants, and directing, by a load balancer, tenant requests of tenants in the plurality of tenants to servers based on the M plurality of server groups.
A database management system (DBMS) receives an input query and parses the received input query to generate an abstract parse tree. Next, the DBMS traverses the abstract parse tree to detect any parameterized structured query language (SQL) views. If a first parameterized SQL view is detected in the abstract parse tree, the DBMS generates a first view parse tree if a first search of a first cache for the first parameterized SQL view results in a miss. Otherwise, the DBMS retrieves, from the first cache, a previously generated view parse tree if the first search of the first cache results in a hit. Then, the DBMS generates a first query compile tree if a second search of a second cache for the first parameterized SQL view results in a miss. Finally, the DBMS generates and executes a query execution plan based on the first query compile tree.
Techniques and solutions are provided for obtaining a suggested configuration for a configurable object. Typically, a particular object and object configuration are recommended based on technical characteristics of the object. However, a user or process wishing to obtain a recommendation may be more familiar with their operational requirements. Disclosed techniques can include an overall solutions category containing solutions of different solutions category subtypes. Sets of requirements attributes and configuration (technical) attributes can be defined for the solutions category. In some cases, a first machine learning model is trained using input values for the requirements attributes and the configuration attributes, and is used to recommend a particular solution in response to a set of input requirement attribute values. Different machine learning models can be trained for the various solutions, including using the configuration attributes for a particular solution, and can be used to recommend a configuration of a selected/recommend solution.
Techniques and solutions are disclosed for managing subprocess versions and their associated components in a computing system. These techniques enable dynamic adaptation and traceability through version control. A subprocess definition is received and updated to reflect modifications to its components, configurations, or execution sequence. Changes are identified and propagated using unique version identifiers and events. Iterative refinement is supported through actions such as validating updated subprocesses, reprocessing data, or maintaining provenance chains. The disclosed solutions provide efficient and structured management of subprocesses.
Techniques are provided for determining elements of a routing. A set of inputs is obtained, where respective inputs are associated with sets of one or more characteristics. Values for the characteristics are associated with a set of labels defining operational routing attributes and are used to train a machine learning model. Inference data including characteristic values for inputs is analyzed using the machine learning model to produce an inference result identifying predicted labels. The predicted labels are used to execute at least a portion of a routing operation including assignments of work centers, execution sequences, or resources. Updated inference data may result in different predicted labels and different routing assignments. Using characteristic values enables improved inference accuracy and allows a greater portion of available data to be used for training.
Methods, systems, and computer-readable storage media for receiving an inference request including inference data, and determining, from the inference request, that generic line-item matching (GLIM)-based inference is to be executed, and in response, transmitting a GLIM inference request including at least a portion of the inference data and a model identifier, retrieving a GLIM model from a model repository using the model identifier, processing the at least a portion of the inference data through the GLIM model to generate inference results, and returning the inference results to an application.
In an example embodiment, a large language model (LLM) is utilized to generate a semantic vector for each given time series. These semantic vectors represent additional information generated based on descriptions of the type of the time series (e.g., a description of the material, whose demand over time comprises the time series). The semantic vectors can then be used to stabilize the assignment of clusters in a cluster-based machine learning model, especially for short time series, to improve reliability of predictions.
If the sizing for an application is incorrect, either too many resources are allocated for the application or too few. Traditionally, sizing is based on answers to a questionnaire, assumptions, data gathered after the application is deployed, or various combinations thereof. As discussed herein, directed testing is used to gather information about the hardware requirements of the application. The gathered information is used along with quality of service information to accurately size the application. A load unit component performs load testing of the application to find a linear dependence of the application on hardware resources for one or more dimensions of demand on the application. Based on the data gathered by the load unit component, a multi-variable linear regression is performed to determine a resource unit for the application and a number of instances of the resource unit to be allocated to the application.
Disclosed herein are a system, method, and computer program product embodiments for enabling document collection creation in accordance with a semi-structured data type, retrieving and filtering data from both relational databases and a document collection, and determining a data type in which the data is retrieved. For example, a statement configured to generate a collection of semi-structured documents in a document store based on a schema is processed. The statement specifies a semi-structured data type in which a plurality of entities from the collection are to be returned from the document store. A determination is made that an entity of the plurality of entities is defined by the schema as being a particular data type different from the semi-structured data type. A query for the entity is provided to the document store. The entity is received, based on the query, in accordance with the particular data type.
Embodiments are described for a graphic-based programing system comprising a display, a memory, and at least one processor coupled to the display and the memory. The at least one processor is configured to receive a programing layout that comprises one or more function components and one or more connections between the one or more function components and store locations of the one or more function components in the memory. The at least one processor is further configured to determine that at least one function component out of the one or more function components corresponds to a first characteristic data and determine a working area in the programing layout. The at least one processor is further configured to determine that the at least one function component is outside the working area and display at least one indicator based on the location of the at least one function component.
The disclosure generally describes methods, software, and systems for generation of negative commits. An object representation of a dataset is received, the object representation exposes readily accessible attributes of modified files in the dataset. Positive commits corresponding to source code issue tracking tickets referencing the modified files in the dataset are determined, by processing the readily accessible attributes of modified files in the dataset. Candidate negative commits are determined for each of the positive commits, by processing the source code issue tracking tickets. A matching score between the positive commits and the candidate negative commits is determined. A sorted set of commits including in each set a positive commit and one or more negative commits for the modified files in the dataset is generated, using the matching score. A machine learning model for detection of source code security issues is trained, using the sorted set of negative commits.
The disclosure generally describes methods, software, and systems for generation of attack vectors and threat modeling using artificial intelligence models. A request to perform threat modeling for a software system including nodes performing operations vulnerable to threats is received. Tokens that include pairs of nodes and one or more edges are generated. A set of paths corresponding to a known threat applicable to a portion of the nodes of a respective path is generated, using the tokens. A similarity search for the set of paths is performed, to determine similar paths corresponding to known threats and known mitigations. A prompt for a prediction model is generated, using the similar path, the known threats, and the known mitigations. A list of threats and a mitigation plan to modify the software system to mitigate the threats is received, from the prediction model.
A system and method include reception of a user query from a user in a chatbot session, issuance of a function call to a data source based on the user query, reception of data from the data source in response to the function call, determination of one or more suggested function calls based on the function call and records associating function calls with subsequently-issued function calls, and returning of a response to the user query to the user in the chatbot session, the response based on the received data and including suggested user input associated with each of the one or more suggested function calls.
G06F 16/3329 - Formulation de requêtes en langage naturel
H04L 51/02 - Messagerie d'utilisateur à utilisateur dans des réseaux à commutation de paquets, transmise selon des protocoles de stockage et de retransmission ou en temps réel, p. ex. courriel en utilisant des réactions automatiques ou la délégation par l’utilisateur, p. ex. des réponses automatiques ou des messages générés par un agent conversationnel
H04L 51/046 - Interopérabilité avec d'autres applications ou services réseau
92.
Systems and methods for data aggregation and suggestion generation
Embodiments of the present disclosure include techniques for data aggregation and suggestion generation. In one embodiment, profile information from remote systems is cached in a local database. As searches are issued in the remote systems, event messages comprising search artifacts are generated and sent to the database. The search artifacts are aggregated and stored in the local database. A profile may be analyzed based on the stored data and suggestions are provided regarding attribute values that may be added to the profile to improve search performance.
In an example embodiment, resource utilization and resource assignment among StatefulSet instances is monitored and usage metrics are maintained. Multiple different StatefulSets are established, with each set having a different level of resource allocation. Individual instances can then be dynamically assigned/reassigned to the different StatefulSets based on resource utilization. An auto-scaler is provided to scale each StatefulSet to the needed number of instances.
A core data services (CDS) explain agent receives a CDS explain request to generate an explanation with respect to a CDS data model. A CDS explain process is orchestrated for the CDS data model using a CDS explain handler of the CDS explain agent. A validator of the CDS explain agent and advanced business application programming (ABAP) dictionary metadata is used to validate if a CDS entity exists and if the CDS entity can be explained. Using a prompt assembler of the CDS explain agent, a large language model (LLM) prompt is assembled by combining multiple prompt snippets that satisfy the explanation with respect to a CDS data model. Using a CDS explain request processor of the CDS explain agent, the LLM prompt is transmitted to an LLM. Using a post-processor of the CDS explain agent, relevant information of a response from the LLM is extracted.
Embodiments of the present disclosure include techniques for analyzing and reducing dependent code. In one embodiment, application source code is linked to dependency code, some of which is used and some of which is not used. The application source code and dependency code are analyzed to produce a usage graph comprising nodes corresponding to functional code elements and edges between the nodes corresponding to calls between the code elements. Edges may be associated with weights that may be used to determine an amount of used in subtrees of the dependency code. Dependency code that is little used may be replaced by dependency code without indirect dependencies. Replacement dependency code may be autogenerated by extracting existing dependency code elements to build a prompt for a generative AI.
A system and method include receipt of a query from a user, determination, from a plurality of stored text portions, of first text portions which are semantically similar to the query, determination of a first score associated with each of the first text portions, generation of a first prompt based on the first scores, the first prompt including the query and the first text portions, transmission of the first prompt to a text generation model, receipt of a response to the first prompt from the text generation model, presentation of the response and the first text portions, receipt, from the user, of a rating of one of the presented first text portions, and updating of the first score associated with the one of the first text portions based on the rating.
A system and method include training a classification model to classify data based on first data associated with a first usage scenario, receiving second data associated with a second usage scenario inputting the second data to the classification model and receiving a likelihood of a first classification from the classification model, determining a similarity between the second data and a plurality of data associated with the second usage scenario, modifying the likelihood based on the determined similarity, determining a second classification of the second data based on the modified likelihood, and processing the second data according to the second classification of the second data.
A computer-implemented method includes finding, using a plugin, a set of reachable dependencies in a production environment. Using the plugin, a set of reachable dependencies in a test environment is found. Using the plugin, a set of candidate dependencies is found. Using the plugin, a determination is made that an empty set does not exist. Using the plugin and as incorrect dependencies, a set of incorrectly scoped dependencies is found. The found incorrect dependencies are reported.
A system and method include execution of a chatbot session consisting of user queries and chatbot feedbacks, insertion of one of the chatbot feedbacks into an analytics document as analytical insight data, reception of a user selection of one or more of the chatbot feedbacks, generation of a data lineage graph from the selected one or more of the chatbot feedbacks, and storage of the data lineage graph in association with the analytical insight data.
G06F 16/901 - IndexationStructures de données à cet effetStructures de stockage
H04L 51/02 - Messagerie d'utilisateur à utilisateur dans des réseaux à commutation de paquets, transmise selon des protocoles de stockage et de retransmission ou en temps réel, p. ex. courriel en utilisant des réactions automatiques ou la délégation par l’utilisateur, p. ex. des réponses automatiques ou des messages générés par un agent conversationnel
The disclosure generally describes methods, software, and systems for identification of code embeddings. A source code fragment is received. Embedding models corresponding to the source code fragment are determined. Each of the embedding models converts the source code fragment into numerical vectors that capture a semantic meaning and a functionality of the source code fragment. Probes corresponding to properties of the source code fragment are determined. Each of the probes perform an analysis of the properties of the source code fragment. Performance metrics indicative of encapsulations of the probes in the embedding models are generated, using a machine learning model trained to process the embedding models and the probes. A ranked representation of the embedding models is generated using the performance metrics.
