92 results about "Context processing" patented technology

The invention is suitable for the technical field of artificial intelligence, and provides a multi-modal large language model dialogue generation method based on natural language understanding, comprising the following steps: receiving multi-modal information input by a user through a multi-modal large language model, the multi-modal information comprising a plurality of modal data; preprocessing modal data in the multi-modal information, and extracting to obtain multi-modal features; performing unified fusion feature representation on the extracted multi-modal features based on a dynamic weighted fusion strategy of a multi-head attention mechanism; determining the dialogue state of the long context based on a long context processing algorithm, and ensuring that the generated reply content is associated with the dialogue history; and according to the unified fusion feature representation and the dialogue state, a natural language is generated through an RAG retrieval enhancement generation technology for replying. According to the method, input of multiple modes can be processed and understood, the ability of understanding the intention of the user is improved, the continuity of the context is maintained, and the generated reply is more natural.

A system for improving employee engagement and automating the integration of business processes. The system includes: a user interaction module configured to receive multimodal inputs, including natural language text and voice requests from employees via enterprise portals, mobile applications, voice-activated devices, and collaboration platforms, and to generate real-time responses via conversational output channels; a context processor that is operationally coupled with the user interaction module, wherein the context processor stores the interaction history, employee preferences, organizational role metadata, and task results in both short-term and long-term memory and dynamically derives context for controlling responses and initiating workflows; A natural language and intent processor that is communicatively linked to the context processor. The intent engine consists of large language models trained on company-specific lexicons to perform intent detection, entity extraction, ambiguity resolution, and sentiment or urgency classification from employee queries; a workflow orchestration layer in communication with the intent engine and the context processor, wherein the workflow orchestration layer includes a rule-based and AI-powered process execution engine configured to automatically initiate, route, and complete cross-functional business tasks, including approvals, escalations, and compliance checks, with workflows defined using modular templates that include conditional logic and time-based triggers; Webhooks and authentication protocols provide secure interoperability between the workflow orchestration layer and external enterprise platforms; a feedback and learning module that is operationally coupled with the workflow orchestration layer and the natural language understanding engine, wherein the feedback and learning module is configured to analyze task completion rates, response accuracy, latency metrics, and user feedback signals, and to retrain underlying language and workflow models for adaptive improvement in real time; and an administration console with role-based access controls, compliance dashboards, audit trails and interfaces for workflow configuration, whereby the administration console enables authorized personnel to monitor system operations, adjust interaction rules and enforce data protection restrictions across departments.

One aspect of the instant disclosure may provide a system and method for processing scatter-gather direct memory access (S-G DMA) instructions. During operation, the system may receive an S-G DMA instruction associated with a message and gather instruction context for the S-G DMA instruction. An S-G DMA processor may process the S-G DMA instruction based on the gathered instruction context and determine whether there exists a pending S-G DMA instruction associated with the message. In response to the presence of the pending S-G DMA instruction, the system stores the instruction context in a hot context cache at an address corresponding to the pending S-G DMA instruction. In response to the absence of the pending S-G DMA instruction, the system stores the instruction context in a cold context cache.

The invention discloses an intelligent data question answering and visualization system and method for leader decision making, and relates to the technical field of artificial intelligence, and the system comprises a knowledge base unit, an intelligent question answering module, a dynamic visualization unit, a quality control unit and a performance optimization unit. According to the method, the customized knowledge base is constructed through the knowledge base unit, cross-modal alignment and layered long context processing are weakly supervised during data fusion, the defects that cross-modal data labeling cost is high and long context processing is redundant can be overcome, and an interaction platform of the knowledge base and a data engine in the knowledge base unit is constructed through the intelligent question and answer module; the model mixing module mixes an NLP model, constructs a combined architecture of an LLM + field fine tuning model, ensures understanding of natural languages and intention analysis precision, introduces an online expert error correction mechanism, and reduces misjudgment during question answering, so that the processing capacity of long texts or complex logic can be ensured, missing of key information in question answering is avoided, and then the answering quality is ensured.

Various systems and methods for contextual capture and processing of screen capture data are disclosed. An example method for screen capture data processing in a computing device may include: determining an active screen of the computing device based on a user interaction event; identifying a graphics rendering event associated with a software application presented in the active screen; identifying screen capture data in a buffer (e.g., of graphics processing circuitry such as a GPU) that corresponds to the graphics rendering event; and communicating a contextual screen update event via an application programming interface (e.g., a an API operated by a GPU driver). The receipt of this contextual screen update event can be used by an AI engine to control whether to perform contextual processing on particular frames of the screen capture data.

The invention provides a large model context processing method and device, computer equipment, a storage medium and a computer program product. The method comprises the steps of obtaining a new action step request, updating a context view tree and a source interaction data tree based on the new action step request, and determining a pressure coefficient according to the updated context view tree; according to the pressure coefficient, determining whether a to-be-compressed node exists in the updated context view tree or not; if the to-be-compressed node exists, combining the updated source interaction data tree and the updated context view tree to perform back-to-source compression on the to-be-compressed node so as to update the updated context view tree and the updated source interaction data tree, and returning to the step of determining the pressure coefficient according to the updated context view tree until the to-be-compressed node does not exist; and based on the updated context view tree, expanding the context view tree into context content and inputting the context content into the large language model. By adopting the method, the large language model can accurately understand and respond to the new action step request.

The invention provides a retrieval recall method and system fusing document fragmentation and authority control, and relates to the technical field of artificial intelligence and information retrieve.The method comprises the steps that a role of a user is obtained according to the identity of the user, and a corresponding authority strategy is determined in combination with a preset multi-level authority mapping relation; the permission strategy comprises a label allowed to be accessed, a data set and a document set; the query request is converted into a semantic vector, and initial candidate document fragments are retrieved from a vector database in combination with an associated data set or tag identifier; according to the method, the candidate results are subjected to access control filtering on the basis of the authority strategy, and the final retrieval recall result is generated, so that role-based fine-grained access control is realized while high retrieval precision is ensured, the problems of rough authority management and weak context processing capability in traditional retrieval are solved, and the security and efficiency of retrieval are remarkably improved.

The invention discloses a multi-round dialogue system and method based on conversion from a natural language to an SQL, and relates to the technical field of artificial intelligence, and the system comprises a user interaction module which is used for supporting a user to input a query problem and displaying a corresponding SQL query result; the context processing module is used for screening the effective dialogue history of the current question and re-integrating the effective dialogue history into a complete dialogue; the ambiguity processing module is used for identifying an entity from a dialogue history to define a user intention, processing a fuzzy keyword in a current question and guiding the user to complement a query condition; the re-splicing module is used for reconstructing a complete dialogue according to a time sequence, eliminating ambiguity, generating a summary through a large model, and combining the summary with a current problem to construct a complete problem; the SQL generation module is used for performing semantic analysis on the complete problem and outputting an SQL query instruction with the highest matching degree; and the execution and rendering module is used for executing the instruction on the target database and returning a query result. The conversation processing capability can be improved.

The present invention provides a system and method for multi AI agent driven data processing in enterprise application developed by codeless platform. The invention includes one or more AI agents configured for processing one or more input received on conversational assistant interface. The invention includes triggering contextual processing of received input by orchestration agent interacting with multi-AI agents for executing a task identified from the received input.

The invention provides an automatic long text fine-tuning instruction set construction method based on a large language model, which comprises the following steps: acquiring an input text, and segmenting the input text by adopting a recursive character segmentation method to generate a paragraph set; aiming at the generated paragraph set, the large language model generates a question set and an answer set by adopting a self-guidance learning method through a preset question type set and a prompt template according to the task type; and generating an instruction set based on the generated question set and answer set, evaluating the quality of the instruction set in multiple dimensions, and optimizing the instruction set according to an evaluation result to obtain an optimized instruction set. According to the method, the high-quality long text fine tuning instruction set can be automatically generated, so that the performance of a large language model is improved, and meanwhile, the challenge of long context processing is solved. By automatically constructing the long text fine-tuning instruction set, the requirement of manual annotation is reduced, the cost is reduced, and meanwhile, the efficiency of the fine-tuning process and the long text processing capacity of the model are improved.

The invention relates to the technical field of artificial intelligence, in particular to an adaptive context processing method and system for a large language model. The method comprises the following steps: collecting multi-source candidate context segments related to current user query; performing utility evaluation on each candidate fragment, outputting a quantized utility score, and performing multi-granularity compression on each fragment based on the utility score and a preset quality constraint to obtain a compressed fragment set; taking the maximization of the total utility score as a target, taking the condition that the total token cost is less than or equal to the global token budget as a constraint, and carrying out optimization selection to obtain an optimal context combination; and inputting the optimal context combination into a large language model to generate an output result for the current query. By applying the method, the context window management of the large language model can be optimized, the adaptive context processing capability of the information utilization efficiency is improved, and the problems of context window limitation, value judgment and the like are solved.

The invention relates to the technical field of psychological intervention adjustment, in particular to a dynamic structured self-adaptive training method which is applied to a training framework comprising an acquisition terminal, a processing terminal and an execution terminal. The acquisition terminal is used for acquiring a physiological signal, a behavior signal and an environment context of an individual, the processing terminal fuses the information and calculates an anxiety index and an uncertainty probability value, and when the uncertainty probability value is lower than a preset threshold value, an initial task graph is constructed according to a target task and the environment context. The initial task graph is subjected to structure and parameter updating through a preset rule generator in combination with a reward function and an anxiety index, an individualized task graph is generated, cross-age migration is further carried out based on physiological and behavior signals, an execution graph bound with an execution terminal is obtained, and the execution terminal carries out intervention according to the execution graph. The problem that an existing training method is difficult to give consideration to behavior complexity and psychological bearing capacity is solved, and training individuation and stability are improved.

A method for dynamic execution of sub-graphs within executable graph-based models is provided. Processing circuitry obtains an executable graph-based model comprising a plurality of sub-graphs and an overlay structure comprising processing logic associated with the plurality of sub-graphs. Each sub-graph defines a hierarchical structure of related nodes. The processing circuitry receives a stimulus and a context associated with the stimulus. In response to the stimulus being received and based on the context, the processing circuitry maps the stimulus to a first sub-graph of the executable graph-based model. The processing circuitry causes execution of processing logic within the overlay structure based on the mapping. The processing logic is associated with one or more nodes of the first sub-graph.

The invention discloses a coarse-to-fine multi-document question and answer method based on a search head, which comprises the following steps: S1, coarse-grained filtering: quickly rejecting a large number of background documents, and reserving a candidate document set; and S2, fine-grained guidance: further suppressing the interference document from the candidate set, and enhancing the attention on the golden evidence document. According to the method, background documents are filtered step by step, and the model is guided to pay attention to key documents, so that the retrieval and reasoning capabilities of the model under long context input are remarkably improved; according to the method, a dynamic document rearrangement strategy based on a retrieval head is introduced, and a key document is placed at the tail of an input sequence, so that the locality characteristic of a model is better utilized, and the efficiency and accuracy of long context processing are further improved.

The invention relates to the technical field of nuclear power station monitoring, and relates to a method and system for rapidly identifying small foreign matters in a reactor pool, and the method comprises the steps: firstly, processing a reference image and a matched image based on an RAFT optical flow method combined with an attention mechanism to obtain an optical flow graph, and then carrying out the segmentation of the optical flow graph to extract a motion region; and the motion tends to be mapped to the matched image to obtain a foreign matter recognition result. According to the invention, by adding an attention mechanism, the context processing capability of the RAFT method is enhanced, and the optical flow of the water surface ripples can be calculated, so that the precision of optical flow estimation is improved.

The present disclosure relates to techniques for improving inference efficiency and memory utilization in transformer-based neural networks. A modified architectural design is introduced that decouples key and value matrix generation from inter-layer dependencies, enabling statically computed or parallelizable projections across layers. The disclosed approach may eliminate the need for layer-wise prefilling, support linear-time inference, and substantially reduce the memory footprint associated with key-value (KV) caching. The disclosed architecture may use shared or layer-specific projections, with a single KV-cache serving all or subsets of layers. In some embodiments, a non-linear transformation (e.g., implemented via a feed-forward network), may preprocess input embeddings prior to query, key, and value generation. A lookup table of transformed embeddings may be precomputed to further accelerate inference. The disclosed system can enhance scalability and may allow deployment of large models on resource-constrained hardware, offering practical benefits for latency-sensitive applications and long-context processing in transformer-based models.

A processor may receive a request to display a user interface (UI) for a user account. The processor may determine a context for the UI from information specific to the user account. The processor may select one of a plurality of UI elements based on the determining. In some embodiments, the selecting may include evaluating a value function taking the information specific to the user account and information about the context as inputs. The processor may cause the UI including the selected one of the plurality of UI elements to be displayed in response to the request.

A system includes an image signal processor (ISP). The ISP is a discrete hardware unit and includes a line buffer, a memory controller, and one or more image processors. The memory controller is configured to assign blocks within the line buffer to a plurality of contexts, and for each of the contexts: receive one or more lines of image data associated with the respective context, and storing the one or more lines of the image data associated with the context in a respective block assigned to the respective context. The one or more image processors of the ISP are configured to process, for each respective context of the plurality of contexts, the one or more lines of the image data stored in the line buffer associated with the respective context.

The application relates to the technical field of recruitment, and discloses a real-time position intelligent recommendation system based on an instant communication context, which comprises an instant communication access module, a real-time context processing engine, a multi-dimensional semantic understanding module, a dynamic position matching engine, a real-time recommendation pushing module, a data storage module, a permission management module, a business opportunity conversion module and a compliance processing module, and each module cooperatively works; the application solves the context split problem, realizes seamless experience of "communication is data", through efficient communication architecture of the instant communication access module and the back end, real-time capture of bidirectional chat messages of HR and candidates, customers, breaking of the split state of the communication scene and the recommendation system in the prior art, inclusion of each interaction into the recommendation consideration, complete capture of dynamic demands and implicit demands in the communication process, and avoidance of demand information omission.

According to the EGANet polyp image segmentation method provided by the invention, the omission ratio and the error detection ratio can be reduced, and the colonoscopy accuracy is improved. The method is designed for three challenges: a ResNet50-SPD encoder is combined with space-to-depth layer downsampling, and a deconvolution decoder is matched, so that the problem of missing detection due to too small polyp size is solved; designing an edge attention module, supervising an advanced feature layer by using edge information, and coping with missing detection and under-segmentation caused by morphological differences; through data set reflection removal and color shift preprocessing, a global attention module is combined to capture long-distance dependence and channel context, and error detection caused by polyp and background similarity is processed. And finally fusing edge and global attention features to relieve over-segmentation. Experiments show that the segmentation precision of the scheme is superior to that of an existing model.

The present application relates to the technical field of data processing, and more particularly to an environment context processing and model optimization method and system based on edge computing, which stores initial data in each time period together with historical data in a target storage, guarantees the continuity and integrity of data in the time dimension, and fuses data of different sources and times to mine the association between current data and past environment states, more comprehensively grasps the state of the target physical place and each target object at the end of each time period, so that the target storage stores continuous historical data over time, facilitates understanding the dynamic changes of the target physical place and the target object in combination with context data, enables the edge world model to better adapt to the characteristics and changes of the target physical place, improves the adaptability between the edge world model and the target physical place, and improves the completion efficiency and completion effect of the target application task.

The application discloses a long context processing robust enhancement method and system based on neuron dynamic adjustment, and belongs to the technical field of text sequence processing, and comprises the following steps: embedding a nerve regulation unit in each Transform layer of a large language model used for a long context processing task; and the nerve regulation unit sequentially performs the following steps: dynamically activating abnormal perception, that is, adaptively calculating an abnormal detection threshold of a current layer according to activation value statistical characteristics of the current layer; activating a hard-gate truncation, that is, generating a binary hard mask according to the abnormal detection threshold, and the binary hard mask realizes deterministic isolation of abnormal components in the activation value while keeping normal components unchanged; and performing hierarchical response compression, that is, performing learnable dimension-level linear compression on the isolated abnormal components, re-integrating the compressed abnormal components and the normal components through gate fusion, and outputting an activation tensor after suppressing the abnormality. The application can prevent rank collapse and gradient oscillation, and enhance the robustness of long context task execution.

This application relates to the field of natural language processing, and more particularly to a method for ultra-long context tensor parallel incremental attention and differential compression transmission. It employs a multi-level precision sliding window attention mechanism, dynamically allocating precision levels such as fp16 / int8 / int4 based on the distance between the window and the current computation position, and the information entropy of the window content. A tensor parallel incremental attention engine enables cross-GPU / node incremental attention computation and asynchronous All-Reduce synchronization, improving computational efficiency. Semantic differential compression transmission technology is used, transmitting only the differential data based on similarity detection, significantly reducing cross-node transmission overhead. Combined with an extreme performance optimization and security control strategy, a balance between performance and security is achieved through hardware monitoring and dynamic power consumption / frequency adjustment. This invention can significantly reduce memory usage and transmission overhead in ultra-long context processing, improve computational efficiency and hardware utilization, enhance the understanding ability of large models for long texts, adapt to heterogeneous hardware environments, and is suitable for large model inference and training scenarios.

The invention discloses a context information management method and device, equipment, a medium and a program product, and relates to the technical field of data processing, and the method comprises the steps that a first component command is acquired, and the first component command carries a collection range of context information; the execution of the context component is triggered based on the first component command to obtain a first vector library, the first vector library is in one-to-one correspondence with the context type of the context information, and the context component is configured to be provided with a context acquisition module, a context processing module, a context storage module and a context retrieval module; the context storage module is used for storing the processed context information based on the context type of the processed context information, the context retrieval module is used for providing a retrieval tool aiming at the first vector library, and the retrieval tool is configured to be provided for the context processing module to call or provide a first object call of a first component command. According to the method, the context information in the first vector library can be efficiently and accurately obtained.

A computer-implemented method for training a machine learning model for managing prompt. A processor set determines patterns of data in a sample dataset to identify representative data from the sample dataset. The processor set combines the representative data with context for a number of tasks to generate a number of simple prompts. Each simple prompt comprises a portion of the representative data and context for a task from the number of tasks. The processor set trains the machine learning model using a training dataset comprises the number of simple prompts. The machine learning model is trained to identify priorities of words in the number of simple prompts.

Various examples of embodiments described herein relate to methods, apparatuses, and computer programs for UE context handling, and more specifically (but not exclusively) UE context handling for enhanced (e.g.,simplified) baseline mobility. One such example of an embodiment relates to a method that includes providing, to a network management entity or function, an endpoint terminal identifier related to an endpoint terminal configured with a first radio resource control, RRC, configuration and having completed a handover associated with an access network entity or function, the access network entity or function being at a target side of the handover; receiving endpoint terminal specific information corresponding to the provided endpoint terminal identifier, the endpoint terminal specific information indicating the endpoint terminal's capability to be configured with a second RRC configuration different from the first RRC configuration; and providing the second RRC configuration to the endpoint terminal based on the endpoint terminal specific information.

Surgical planning based on a three-dimensional (3D) model includes obtaining a 3D model of an anatomical region, wherein the anatomical region includes the patient's anatomical features; identifying at least one deformity, wherein the at least one deformity is identified with respect to a target anatomical value of the patient's anatomical features, wherein the target anatomical value includes a desired range into which the anatomical measurement should fall; and using the anatomical measurement, determining at least one planar correction to be made to at least one anatomical structure of the patient. Generating a 3D model using two-dimensional (2D) imaging data may include performing anatomical context processing, which includes annotating the model, contours, edges, or surfaces of the anatomical structure, as well as performing 2D-to-3D reconstruction to optimize the orientation, placement, and scale of the 3D digital volume modeling the anatomical structure, resulting in a 3D anatomical representation of the anatomical region as a 3D model.

A dialog understanding device includes a sound collection module configured to collect a sound signal, a contextual processing module, and a dialog system configured to perform a dialog with a human. The contextual processing module includes a plurality of layers for processing information obtained from the sound collection module. A fallback processing module of a case where a predetermined process has not succeeded for the collected sound signal is provided on each layer of the plurality of layers. A process of the next layer is performed after a fallback corresponding to a fallback process performed by the fallback processing module is performed. A sound signal obtained when the contextual processing module completes a process is input to the dialog system.

The present disclosure provides techniques for seamless roaming with uplink / downlink context transfer. A serving AP transmits a sequence of downlink data units for a TID to a client device, the sequence of downlink data units having sequence numbers falling within a transmit window for the TID. The serving AP receives, from the client device, a roaming request identifying a target AP. In response, the serving AP sends a roaming context message to the target AP, comprising at least one of a SSN of the transmit window corresponding to the TID, or a NSN for the TID, where the NSN is a first sequence number to be assigned for downlink data units of the TID transmitted by the target AP to the client device.

The invention discloses a KV cache intelligent scheduling and semantic structure compression method and system based on a knowledge graph, and the method comprises the steps: firstly, extracting entities and relationships from an input text to construct the knowledge graph, and then employing a graph embedding and structure compression algorithm to aggregate nodes with similar semantics into a high-density semantic KV item, so as to eliminate redundancy, and improve the KV cache intelligent scheduling and semantic structure compression efficiency. And then constructing a semantic scheduling graph, and dynamically loading the most relevant semantic sub-graph cache through a semantic migration mechanism. According to the method, the problems of semantic redundancy, lack of structural understanding, poor dynamic adaptability and the like in a traditional KV cache are effectively solved, the cache utilization rate, semantic consistency and long-range context processing capacity are remarkably improved, and the method is suitable for scenes such as large language model reasoning acceleration and intelligent dialogue.