3627results about "Natural language analysis" patented technology

An example operation may include one or more of storing a database of payment card data, conversing with a user via a chatbot within a chat window of a software application, wherein the conversing comprises receiving a query from the user about a payment card during a chat session between the user and the chatbot, executing a large language model (LLM) on the query about the payment card and the database of payment card data to generate a chatbot response, and displaying the generated chatbot response via the chatbot within the chat window of the software application during the chat session.

Systems and methods are provided for implementing automatic prompt optimization using textual gradients. In various embodiments, a feedback prompt, input into a large language model (“LLM”), is used to generate textual gradients that criticize a current prompt. The feedback prompt includes the current prompt and predictions that are incorrect compared with corresponding labels associated with minibatch data processed by the LLM using the current prompt. The textual gradients and current prompt are used in an editing prompt to the LLM to obtain a set of optimized prompts, which may be expanded using a paraphrasing prompt that is input into the LLM to generate a set of paraphrased prompts. A selection algorithm is used to select one or more optimized prompts from the set of optimized prompts and / or the set of paraphrased prompts, and the process is repeated with the selected one or more optimized prompts replacing the current prompt.

Techniques for a knowledge-graph system to use large language models (LLMs) to build knowledge graphs to answer queries submitted to a chatbot by users. The knowledge-graph system builds the knowledge graph using answers produced by an LLM for novel queries. The chatbot will continue to use the LLM to answer novel queries, but the chatbot may harness the knowledge graph to answer repeat questions to gain various efficiencies over LLM-backed chatbots. For example, the knowledge-graph system may easily debug or otherwise improve the answers in knowledge graphs, store provenance information in knowledge graphs, and augment the knowledge graphs using other data sources. Thus, the reliability and correctness of chatbots will be improved as the bugs and inaccuracies in answers provided by the LLM will be corrected in the knowledge graphs, but the chatbots can still harness the abilities of LLMs to provide answers across various subject-matter domains.

An example operation may include one or more of storing a data store of credit card documentation, conversing with a user via a chatbot within a chat window of a software application, wherein the conversing comprises receiving natural language inputs with requests for information about a payment card, identifying a benefit obtained by the user based on the conversation, executing a large language model (LLM) on the identified benefit and the data store of the credit card documentation to generate a letter of confirmation, and generating an electronic message with the letter of confirmation attached, and transmitting the electronic message to a user device.

Guiding query creation for a generative artificial intelligence (AI)-enabled assistant, including: receiving, via a natural language interface for a security framework monitoring a cloud deployment, a natural language input comprising one or more entity identifiers; gathering information based on the one or more entity identifiers; providing, to a generative artificial intelligence (AI) model, a prompt based on the natural language input and comprising the gathered information; and receiving, from the generative AI model, a response to the prompt.

Systems / techniques that facilitate data drift detection for unstructured texts via deep learning autoencoders are provided. In various embodiments, a system can access a pre-trained natural language model and a set of unstructured texts on which the model is to be executed. In various aspects, the system can determine, via execution of a deep learning autoencoder, how different a first set of reconstruction errors associated with the set of unstructured texts are from a second set of reconstruction errors associated with a set of training unstructured texts on which the pre-trained natural language model was trained. In various instances, the system can generate, in response to a determination that the first and second sets of reconstruction errors differ by more than a threshold margin, a first alert indicating that data drift has occurred and that the model is thereby unable to confidently analyze the set of unstructured texts.

Architecture for a generative artificial intelligence (AI)-enabled assistant, including: receiving, via a natural language interface for a security framework monitoring a cloud deployment, a natural language input; selecting, by a delegator generative artificial intelligence (AI) model and based on the natural language input, a particular generative AI model from a plurality of selectable generative AI models; prompting the particular generative AI model based on the natural language input; and providing, via the natural language interface and in response to the natural language input, a response from the particular generative AI model.

An Applicant Assistant System may receive, from a user device, a user input indicating a query or task associated with information displayed in a user interface. The device may determine one or more components associated with the user interface and access component metadata of the determined one or more components, the component metadata for each of the one or more components indicating one or more of a fact, a data link, or an action. A prompt comprising at least some of the component metadata, at least some context information, and an indication of one or more available response elements may be provided to a large language model (LLM). The LLM may response with one or more response elements that may be processed by the System, such as to cause updates to the user interface on the user device.

Technology is disclosed herein for generating a visualization of data based on an AI-generated data object. In an implementation, an application, such as a data analytics application, receives a natural language input from a user which relates to a table of data in the application. The table includes data organized according to table columns. The application generates a prompt for a large language model (LLM) service which includes the names of the table columns. The prompt tasks the LLM service with selecting columns for the visualization based on the natural language input and the names of the table columns. The prompt tasks the LLM service with generating a response in a JSON format. The application populates the JSON object, which describes the visualization, according to the response. The application then creates visualization based on the JSON object.

Providing a customizable generative artificial intelligence (‘AI’) assistant, including: identifying one or more customizations for the generative AI assistant, the generative AI assistant configured to receive information describing a monitored deployment and a natural language input, the generative AI assistant further configured to generate a response to the natural language input; and modifying, based on the one or more customizations, the generative AI assistant.

Verifying large language model responses involves obtaining a query and its corresponding answer from a large language model. This conversational text is then fed into a second large language model, which translates the answer into first-order logic. The verification process uses an automated theorem prover. It checks the validity of this logic translation by determining the unsatisfiability of two scenarios: one where the negation of the logic translation and domain-specific logic formulas are combined, and another where the logic translation itself is combined with these formulas. Based on this analysis, the theorem prover ascertains whether the translated answer is valid, invalid, or neither. The final step is communicating this verification status through an appropriate output medium, such as a graphical user interface, a database, or a report, providing a structured and methodical approach to assessing the accuracy and reliability of language model responses.

The present invention relates to the technical field of knowledge graph completion tasks, and particularly relates to a knowledge graph link prediction method. The method comprises: using a precoding model to obtain an embedding layer vector, and constructing a corresponding masked triple; adding a corresponding position code to each element in the masked triple, so as to obtain a corresponding input sequence, inputting the input sequence into a trained main masking model, and outputting an entity classification probability; and on the basis of the entity classification probability, predicting potential candidate entities. The method further comprises: concatenating semantic information corresponding to the embedding layer vector and structural information obtained by an embedding model, so as to obtain fused head entity and relation representations, and constructing a corresponding fused masked triple; and adding a corresponding position code to each element in the fused masked triple, so as to obtain a corresponding fused input sequence. The present invention uses a precoding method, thereby effectively reducing the training burden on a model, and improving the inference speed of a model; and a fusion module is used before inputs are fed into a main masked model, thereby ensuring the integrity of textual description information and improving prediction accuracy.

A system and method for extracting data from large language model (LLM) outputs, including: training a model using labeled data items to assign rankings to LLMs; selecting, by the trained model, one or more of the LLMs based on the rankings; sending an LLM prompt to selected models; and outputting, by the model, a refined response to the prompt based on responses to the prompt by the LLMs. Some LLM prompts according to some embodiments may include different sets input parameters of different types—such as, e.g., a set of block parameters and a set of editorial parameters. In some embodiments, a model or LLM may be updated or retrained using a reinforcement learning approach and based output items or refined responses generated by that model or LLM—which may for example be scored or ranked and used in combination with reward or cost functions to update model parameters.

A computer-implemented method for jailbreaking Large Language Models (LLMs) requiring only black-box access-Branch-and-Prune. LLMs are powerful tools displaying many capabilities. However, despite safety training, LLMs can generate harmful, biased, and toxic content-demonstrated by the prevalence of human-designed “jailbreaks” that override LLM's safety “guardrails.” At the core, the method engages an “attacker” LLM into conversations where it generates variations of the original prompt that may jailbreak the target LLM. Compared to prior methods, the disclosed Branch-and-Prune adaptively decides which conversations to engage in (pursuing conversations with a high-likelihood of success while abandoning less-promising ones), and “mixes” the different conversations (increasing the success rate of the method). This enables jailbreaks for state-of-the-art LLMs for over 80% of the existing harmful prompts while requiring fewer queries to the target LLM.

Embodiments described herein relate to systems and methods for automatically generating content, generating API requests and / or request bodies, structuring user-generated content, and / or generating structured content in collaboration platforms, such as documentation systems, issue tracking systems, project management platforms, and other platforms. The systems and methods described use a network architecture that includes generative interface panel used to access a prompt generation service and a set of one or more purpose-configured large language model instances (LLMs) and / or other trained classifiers or natural language processors used to provide generative responses for content collaboration platforms.

Methods and systems for a document review process are provided. The method includes receiving an input digital model representation comprising at least one externally-accessible model endpoint for generating a digital artifact. Then, generating a document splice comprising access to multiple document subunits, with at least one document subunit written in a natural language and comprising the digital artifact; the access to each document subunit is provided through an externally-accessible document endpoint for the subunit. Then, generating a document by combining the document subunits, and generating a view associated with the document, based on an user authorization result including selective access rights to the document subunits. The view comprises access to the digital model representation, the digital artifact, each document subunit, and the document. Finally, receiving a user input and updating, via one of the externally-accessible document endpoints, the document splice based on the user input.

A computer program comprising instructions which, when executed by a computer, cause the computer to perform the following operations: Receiving an input document (104), wherein the input document (104) contains content elements including at least one text element; Performing a lossless hierarchical text processing process (106) in which the input document (104) is reproduced as a hierarchical representation (108) in which all text elements from the input document (104) are preserved verbatim and organized according to a structure of the input document (104), the lossless hierarchical text processing process (106) comprising: Prompting a generative language model to generate the hierarchical representation (108), wherein the prompt instructs the generative language model to generate the hierarchical representation (108) without summarizing or omitting any of the text elements; and Performing a retrieval unit generation process (110) in which retrieval units (112) are generated based on the hierarchical representation (108) and stored in a memory accessible by a generative client language model for retrieval-enhanced generation.

A computer-implemented method of updating a set of clusters representative of a classification of a text-based dataset into a plurality of different text types for use in a cyber security system is described as part of a classification pipeline. The method comprises receiving text data associated with an entity. The method further comprises generating one or more vector embeddings representative of the text data. The method further comprises using incremental learning to update the set of clusters based on the one or more vector embeddings.

The invention discloses an industrial network risk perception and collaborative early warning method based on a dynamic risk map, and relates to the technical field of industrial internet security, and the method comprises the steps: S1, multi-source perception deployment; s2, heterogeneous data fusion acquisition; s3, constructing a knowledge graph engine; s4, analyzing depth data; s5, performing dynamic risk assessment; and S6, intelligent early warning decision making. According to the industrial network risk perception and collaborative early warning method based on the dynamic risk map, through fusion perception of OT layer data such as equipment states and process parameters, the problems of single perception dimension, evaluation lagging and disjunction in the prior art are solved, the false alarm rate is extremely low, and the method is suitable for popularization and application. Particularly, a dynamic adjustment mechanism of a time-varying risk weight matrix is improved, novel attacks can be dynamically responded, meanwhile, cross-domain risk conduction analysis is achieved, the accuracy and response speed of industrial network security early warning are improved, and meanwhile a closed-loop mechanism of attack path prediction and disposal suggestions is constructed.

Aspects of the subject disclosure may include systems and methods, for example, including receiving a user input in natural language, retrieving a first answer including a list of a first number of documents relevant to the user input by searching indexed documents in a knowledge base, applying the first answer to a large language model reader, resulting in a second answer, re-ranking the first answer by using the second answer, resulting in a third answer including a re-ranked list of the first number of documents, and generating a final response including the second answer and one or more documents among the third answer.

Described systems and techniques determine an event graph of a causal chain of events representing a situation within a network, the event graph including event text characterizing at least one event of the causal chain of events. The event graph may then be processed using a large language model that includes at least one topological context adapter that includes a graph adapter and a text adapter, including processing the event graph with the graph adapter and the event text with the text adapter. The at least one topological context adapter may be trained using existing narratives describing past situations, and / or may be trained using worklog data describing past situations and corresponding actions taken to remedy the past situations. Outputs of the graph adapter and the text adapter may be combined to generate a narrative of the situation that explains the causal chain of events and / or instructions to remedy the situation.

The present disclosure provides a computer-implemented system for matching at least one client with at least one service provider. The system includes a terminal for receiving input data comprising a processor, a memory, a database module, and a network module. An engine for data processing comprises a machine learning process trained to match the at least one client and the at least one service provider. An explainable model minimizes bias of the machine learning process. An explanation interface shows output data comprising results of the machine learning process after application of the explainable model, offering a recommendation for matching the at least one client with the at least one provider.

A cloud platform may include a model interface that receives from a client and at an interface for accessing a large language model, a prompt for a response from the large language model, and the client is associated with a set of configuration parameters via a cloud platform that supports the interface. The cloud platform may modify, in accordance with the set of configuration parameters, the prompt that results in a modified prompt and transmit, to the large language model, the modified prompt. The cloud platform may receive the response generated by the large language model and provide the response to a model that determines one or more probabilities that the response contains content from one or more content categories. The cloud platform may transmit the response or the one or more probabilities to the client.

Systems and methods for detecting and mitigating prompt injection attacks on a generative LLM are disclosed. A deployment scenario is considered, in which the generative LLM supports a task automation function. Prompts are received and interpreted by the generative LLM, and outputs from the generative LLM are used to trigger automation actions. The prompts are constructed based on a combination of user input and external data and are, therefore, vulnerable to prompt injection attacks though manipulation of the external data. To mitigate this risk, a separate discriminative classification, decoupled from the generative LLM, engine is configured to identify malicious prompts, and filter out any malicious prompts before they reach the generative LLM.

A cyber security system is adapted to contextualize and visualize cloud architectures featuring ephemeral cloud assets with generation of a cloud asset remediation plan to group alerts for handling based on security team responsibilities.

A data processing system implements obtaining build logs that include information associated with a software build problem; analyzing the logs to generate a knowledge graph identifying the relationship between various entities in the logs; extracting a signature of a candidate root cause of the build problem from the knowledge graph representing a subset of nodes and edges of the knowledge graph; providing the signature of the candidate root cause to a graphical language model to obtain a prediction of a category of root cause failure selected from among a plurality of root cause failures; constructing a prompt for a language model to generate a root cause failure analysis that describes the root cause of the build problem, the prompt including the category of root cause; receiving the root cause failure analysis from the language model; and performing one or more actions in response to receiving the root cause failure analysis.