165 results about "Template entry" patented technology

The invention relates to the technical field of knowledge graph question answering, and discloses a knowledge graph retrieval reasoning method and system based on big language model enhancement, and the method comprises the steps: carrying out the entity extraction of a user question through a big language model, and generating a candidate entity list; retrieving entity nodes corresponding to the candidate entity list in a knowledge graph database through a mixed retrieval method of fuzzy matching and vector retrieval; re-screening the entity nodes by using a large language model to generate a starting node list; performing width-first traversal on each node in the starting node list in an edge-in direction and an edge-out direction, and retrieving a reasoning path in the knowledge graph; constructing the reasoning path into a prompt template, and inputting the prompt template into a large language model to generate an answer to the user question; according to the method, the big language model is utilized to analyze questions and query the knowledge graph in the knowledge graph question-answering scene, question-answering is performed based on the knowledge graph path, and the effectiveness and accuracy of user knowledge graph question-answering are improved.

The invention discloses a CLIP-based double-prompt optimized few-sample industrial anomaly detection method. The method comprises the following steps: respectively constructing a learnable normal text prompt template and an abnormal text prompt template by utilizing learnable normal word vectors and abnormal word vectors; performing image enhancement processing and image abnormal synthesis processing on the normal sample image for model training to generate an enhanced image and an abnormal synthesis image; inputting the learnable normal text prompt template and the learnable abnormal text prompt template into a text encoder in the CLIP model to obtain text prompt embedding; respectively inputting the normal sample image, the enhanced image and the abnormal synthetic image into an image encoder in the CLIP model to obtain visual embedding; and the similarity between text prompt embedding and visual embedding is measured to carry out optimization training on learnable text prompts, so that the trained CLIP model is utilized to carry out industrial anomaly detection.

The invention provides a game strategy intelligent layout generation and dynamic optimization method and device and a storage medium. The method comprises the following steps: when a layout trigger event is detected, generating a layout generation request; in response to the layout generation request, calling a knowledge graph engine to perform correlation calculation on the relational data, and screening from a preset layout template library based on a correlation result to obtain a plurality of initial layout candidate templates; inputting the initial layout candidate template into a low-code layout editing module, and generating a corresponding layout description file according to a mapping rule of the page component and the component parameter; inputting the layout description file as a current state into a reinforcement learning decision model, and outputting target layout configuration in combination with user portrait data; and sending the target layout configuration to a front-end rendering engine, and rendering the game strategy page according to the target layout configuration. According to the method, intelligent generation and real-time dynamic optimization of the game strategy page layout can be realized, and the personalized degree of the layout and the user interaction experience are improved.

The invention discloses a multi-agent cooperative cue word engineering optimization method, system and device and a medium, and relates to the technical field of cue word engineering, the method comprises the following steps: after initially receiving a user input task, starting a cue word optimization process; an initial cue word template is created through a cue word generation agent, historical optimal templates are retrieved from a knowledge base for fusion, and candidate cue word templates are generated. And the analysis agent inputs the candidate template into the second large language model to generate a candidate answer. And the evaluation agent performs multi-dimensional evaluation on the answers and gives scores and defect analysis. And according to an analysis result, updating the cue word to generate a candidate template of the intelligent agent. And arranging the intelligent agent to cyclically execute the steps until the stop condition is met, and outputting the optimal cue word template. And finally, storing the optimal template into a knowledge base and establishing a semantic index to support subsequent retrieval reuse. According to the method, manual intervention is not needed, and a complete automatic process of generating, evaluating and optimizing the cue word is realized.

The invention relates to an intelligent bidding document compiling method, system and equipment based on a large language model. The method comprises the following steps: constructing a bid invitation knowledge graph; obtaining project demand information to determine demand fields, calling a knowledge graph association interface to match bid invitation file modules and terms for each demand field, and generating a demand field-association module mapping table; according to the mapping table and bid invitation industry specifications, a generation prompt template embedded with a dynamic nesting slot position and a knowledge graph rule is constructed, and a pre-trained bid invitation big language model is input to output a bid invitation clause draft; traversing the knowledge graph according to a demand change field to calculate a change influence radius, constructing an update prompt template in combination with a slot association rule and a constraint condition, inputting a large model to generate revised content only aiming at the demand field and an association module, replacing a corresponding position of a draft, and performing secondary verification on logic consistency through the knowledge graph, and obtaining the updated complete bid invitation file. By adopting the method, intelligence, precision and high efficiency of bid invitation file compiling can be realized.

The invention discloses a network threat knowledge automatic extraction method, electronic equipment and a storage medium, and the method comprises the following steps executed by a computer hardware system: collecting threat intelligence data related to an APT organization from a multi-source network security text, and processing the threat intelligence data to generate a standardized corpus; using the pre-training sentence vector model to generate semantic embedding for a corpus input text and a manual annotation example library text, and retrieving similar examples to construct an ICL prompt template; inputting a large language model subjected to LoRA fine tuning, and extracting structured triples of multiple types of entities and semantic relationships; generating standardized entity nodes and updated relation information by adopting semantic aggregation; and constructing an APT organization network threat intelligence knowledge graph and outputting a structured file. The method provides key technical support for APT attack tracing, threat situation awareness and automatic security policy generation.

The invention discloses a data intelligent interaction enhancement query method and system based on a large model. The method comprises the steps that S1, a user inputs a query problem through a webpage interface; s2, determining a target data table in the candidate data tables, and obtaining table structure and attribute information; s3, the query problem is rewritten; s4, acquiring a question and SQL statement pair set of which the similarity ranks top k with the query question; s5, filling a first cue word template with the rewritten query problem, the table structure and attribute of the target data table, the problem and the SQL statement, and inputting the first cue word template into a large language model to generate a target query statement; s6, executing query in the corresponding data source according to the target query statement to obtain a target query result; and S7, filling the target query result and the query question into a second cue word template, inputting the target query result and the query question into the large language model, generating a query result in a natural language form, and returning the query result to the user. According to the invention, the performance and practicability of the interactive data query system are remarkably improved.

The invention discloses a natural language database query method and system based on large-model two-channel context enhancement, and the method comprises the steps: carrying out the parallel execution of a business link and a mode link on a query statement, and respectively outputting business context information and mode link information; carrying out vectorization fusion on the two types of information to generate combined context information; generating a query intention vector based on the combined context information, and outputting a complexity label of the query statement through a complexity calculation model; and generating a prompt template according to complexity label differentiation, inputting a large model to generate an SQL statement, and executing query. Through a double-channel cooperation mechanism, the problems of semantic ambiguity, database structure complexity, poor field adaptability and the like in natural language query are solved, query accuracy and response efficiency are improved, and the method is suitable for multi-industry complex database scenes.

The invention discloses a high and steep slope deformation monitoring method combining an xLSTM-Attention time sequence prediction model and a large language model. The method comprises the following steps: 1) constructing a time sequence prediction model based on xLSTM and an attention mechanism; 2) setting a plurality of side slope monitoring points on the side slope to be detected, acquiring three-dimensional deformation time sequence data of the side slope monitoring points in the T time period, and performing feature extraction to obtain statistical features; inputting the three-dimensional deformation time sequence data in the T time period into a slope deformation monitoring model to obtain a deformation prediction result in a future T'time period; 3) extracting an important hysteresis feature Top-KLags of the slope deformation monitoring model to form a standardized natural language prompt template; 4) calling a pre-training language model with engineering understanding ability, and performing field fine tuning; and 5) inputting the standardized natural language prompt template into the pre-training language model after field fine tuning to obtain a slope deformation safety risk assessment report. According to the invention, efficient deformation prediction and risk analysis can be realized.

The invention relates to the technical field of research and development management, can be applied to business scenes of financial science and technology, medical health and the like, and discloses a multi-page application construction method and device, equipment and a medium, and the method comprises the steps: configuring a mapping relation between a template path and a script path; traversing a page module under the source file directory, and extracting a module identifier and a script path parameter; generating an entry script reference and injecting a placeholder of the template engine, and generating a page entry file; setting an entry configuration item of the construction tool and loading a module script file; executing a code compiling process, a dependency analysis process and a static resource packaging process; extracting a public dependency library of the page module and combining the public dependency library with the module script file into a public code block; and packaging the public code block and the compiled product into a deployable application package. According to the method, the mapping relation between the template entry and the script path is uniformly managed, the page entry file is automatically generated, and the construction tool entry is configured, so that the page entry configuration complexity is reduced.

The invention provides a literature extraction method, system and device based on a large language model and a storage medium, and relates to the technical field of medical research, and the method comprises the steps: constructing a literature extraction template according to a literature extraction demand; wherein the literature extraction requirements comprise extraction rules, format requirements and pre-extraction fields, and the pre-extraction fields comprise research basic information, research design types, research groups, participant characteristics, intervention measures of each research group and background medication of each research group; obtaining a to-be-extracted literature, and converting the to-be-extracted literature into a processable text format to form a first text; and inputting the first text and the literature extraction template into a large language model, performing reading analysis on the first text by applying the large language model, and extracting the first text according to the literature extraction template to obtain a literature extraction result corresponding to the to-be-extracted literature. According to the scheme, the accuracy and efficiency of document extraction are improved, and the requirements of system review can be better met.

The invention discloses a steel ladle hoisting safety monitoring and early warning method, system and equipment and a medium. The method comprises the following steps: acquiring double-side images in a steel ladle hoisting process; inputting the bilateral images into a multi-target detection model, and outputting bounding boxes, category labels and confidence coefficients of the plurality of key targets; obtaining a local cutting image of each key target according to the bilateral images and the bounding box, the category label and the confidence of each key target, and generating a structured text in combination with the bounding box position, the relative pixel distance and the category label of each key target; inputting the local cutting image, the structured text and the prompt word template of each key target into a multi-modal big language model, and outputting a behavior label, a risk level, a reason, a recommendation action and an instruction state of the current steel ladle hoisting scene through multi-modal reasoning; early warning is conducted according to the recommended action, and the steel ladle hoisting equipment is controlled according to the instruction state. According to the invention, intelligent monitoring and linkage early warning in the steel ladle hoisting process can be realized.

The invention discloses a public opinion home address standardization method and system, and the method comprises the following steps: S1, obtaining and processing multi-source address original data to obtain an address data block, processing the data block based on a pre-trained text embedding model to obtain a corresponding semantic vector, and storing the address data block and the corresponding semantic vector, establishing a multi-source address knowledge base; s2, processing public opinion text data through the text embedding model to obtain a query vector of the public opinion text data; s3, querying the multi-source address knowledge base based on the query vector to obtain a plurality of address data blocks related to the query vector; and S4, constructing a cue word template based on the public opinion text data and the address data blocks, and inputting the cue word template into a large language model to perform public opinion attribution address standardization analysis.

The invention provides an information recommendation method and device, electronic equipment, a storage medium and a program product, and is applied to the technical field of artificial intelligence. The method comprises the steps that first feature data of a first user and first multi-source behavior data of the first user are acquired; inputting the first feature data, the first multi-source behavior data and a pre-constructed thinking chain prompt template into a pre-trained reasoning model, and reasoning the first feature data and the first multi-source behavior data according to the thinking chain prompt template through the reasoning model, and predicting a target information type in which the first user is interested. According to the method, through the multi-source behavior data and the thinking chain prompt template, the prediction accuracy of the reasoning model on the information type interested by the user is improved.

The invention provides a knowledge editing method based on multilingual semantic retrieval, and belongs to the technical field of natural language process.The method achieves cross-lingual knowledge updating through the two stages of multilingual knowledge retrieval and context editing and comprises the steps that firstly, a multilingual retrieval model based on XLM-R is used for mapping queries and knowledge base entries to a shared semantic space, and then the shared semantic space is used for conducting context editing; semantic correlation is judged through a classifier, and target language knowledge is retrieved; and splicing the retrieval result and the query in a zero sample or small sample mode to generate a prompt template, and inputting the prompt template into a large language model to complete editing. The method is characterized in that monolingual limitation is broken through, collaborative updating of 12 languages is supported, a retrieval-editing decoupling architecture is adopted to achieve model irrelevant adaptation, and retrieval and context learning are fused to improve editing accuracy. The method can efficiently correct multi-language large model fact errors, is suitable for scenes such as search engines and intelligent customer services, avoids the cost of full-model retraining, and remarkably improves the efficiency and accuracy of cross-language knowledge updating.

The invention discloses a model robustness automatic evaluation method based on large model driving and related equipment. The method comprises the following steps: inputting a task template of a tested model into a dynamic Prompt engine module to obtain a sample generation instruction; generating an adversarial sample according to the sample generation instruction through a sample generation and verification module, and performing model evaluation on the adversarial sample; if the confrontation sample passes the model verification, extracting a high-dimensional multi-modal feedback vector of the tested model through a multi-modal feedback analysis module; a reusable semantic instruction is generated through a strategy control module according to the high-dimensional multi-modal feedback vector so as to optimize a generation strategy; and if the confrontation sample does not pass the model verification, performing effect evaluation on the confrontation sample through an evaluation and interpretation generation module and generating an interpretability evaluation report. According to the method, progressive generation of the adversarial sample and exponential-level improvement of the model robustness evaluation efficiency can be realized, a clear model improvement direction is provided for developers, and the method can be widely applied to the technical field of computers.

The invention relates to a text interaction method and device for assisting social software interaction, equipment and a medium. The method comprises the steps of obtaining a screen image of current social software in response to an obtained auxiliary interaction instruction; the auxiliary interaction instruction is used for instructing the intelligent terminal to capture a screen image according to a corresponding screenshot mode and feed back the screen image in a standard format form; preprocessing the screen image to obtain a to-be-recognized image; performing text recognition on the to-be-recognized image to obtain structured text data; the structured text data comprises text content and role affiliation information; performing semantic analysis on the structured text data to obtain a formatted prompt template; and inputting the formatted prompt template into a pre-trained auxiliary social text interaction large model to obtain an auxiliary interaction text. By adopting the method, timely and accurate interaction support can be provided for the user when the user chats on the social software, and the time for thinking, language organization and searching of the user is saved.

The invention discloses an OCR (Optical Character Recognition) result correction method and device, equipment and a computer storage medium. The correction method comprises the following steps: acquiring a to-be-corrected text after OCR (Optical Character Recognition), wherein the to-be-corrected text comprises one of messy code characters, character confusion errors or grammar errors caused by OCR errors; constructing a semantic correction instruction template, wherein the semantic correction instruction template comprises a task definition, a rule definition and a format constraint target; the to-be-corrected text and the semantic correction instruction template are input into a large language model, a corrected text result is output through the large language model, and the large language model is a pre-training language model. According to the method and the device, messy code characters, character confusion errors or grammar errors caused by OCR recognition errors can be accurately corrected, so that the accuracy and availability of texts are effectively improved, and high-quality basic data are provided for subsequent text processing; the method can adapt to various documents in different formats and multi-language mixed text scenes, the maintenance cost is saved, and meanwhile the correction efficiency is improved.

The invention provides a text subject matching method and device based on a large model, and the method comprises the steps: determining a standard subject list template and a to-be-matched text, inputting the text and the standard subject list template into a subject matching model, and obtaining a subject matching result of the text outputted by the subject matching model in the standard subject list template. The subject matching model is obtained based on training of a training data set, the training data set comprises multiple groups of training data, and each group of training data comprises guidance information, a standard subject list template, a non-standard subject text and a matching result corresponding to the non-standard subject text. The subject matching model is trained by adopting the training data set, the subject matching model can perform error correction and mapping on the to-be-matched text based on the standard subject list template, accurate error correction and semantic matching on the to-be-matched text are realized, and the robustness and accuracy of text subject matching are remarkably improved.

The invention relates to the technical field of knowledge maps, and provides a screening type dynamic classification self-adaptive relation extraction method, device and equipment, and the method comprises the following steps: defining an entity type list, inputting a text by a user according to a cue word template in a labeling format, judging the length of the text, and extracting the entity type list according to the length of the text; selecting a corresponding strategy according to the text length to generate a candidate entity pair; the association strength of each candidate entity pair is calculated by combining static corpus statistics and a dynamic semantic hybrid scoring model, and the candidate entity pairs with low association are filtered; constructing a dynamic prompt request LLM to generate a natural language abstract describing the subject entity and the object entity, and classifying the abstract into a predefined relationship type to obtain a structured triple; and importing the triple into a graph database to construct the knowledge graph. According to the method, the knowledge extraction efficiency is improved, and the problems of calculation redundancy and high cost are solved.

The invention relates to the technical field of artificial intelligence, and discloses an AI dialogue system and method based on multi-modal data fusion, and the system comprises a processing module which is used for obtaining multi-modal user dialogue data, carrying out the preprocessing of the multi-modal user dialogue data, and obtaining a user emotion feature set; the type module is used for determining a user emotion type according to the user emotion feature set and setting a prompt word template according to the user emotion type; the model module is used for filling the multimodal user dialogue data into a cue word template, and inputting the cue word template filled with the data into a large language model to obtain a model answer result; and the correction module is used for acquiring the preset number of multi-modal user dialogue data, determining a user emotion trend according to the preset number of multi-modal user dialogue data, inputting the user emotion trend into the cue word template, and generating a model correction answer result. According to the method, the multi-modal dialogue data of the user can be accurately recognized, and the accuracy and efficiency of multi-modal intention understanding of the user are effectively improved.

The invention discloses a network threat behavior reasoning method and system based on large model retrieval enhancement, and the method comprises the following steps: a computer system executes the following steps: determining a query target and an associated scene or organization by receiving a network threat analysis query request of a user; retrieving related nodes and relationships from the threat intelligence knowledge graph, and generating an initial reasoning sub-graph; calling a GraphRAG framework, and carrying out semantic matching retrieval on a subgraph node context and an external threat intelligence knowledge base to obtain supplementary evidence; constructing a comprehensive prompt template, inputting the large language model subjected to LoRA fine tuning, and generating a conformity ATTamp; reasoning a conclusion of the CK framework; and finally, supplementing the newly added nodes and the relationship to the knowledge graph, and outputting a structured file or a visual result. According to the method, retrieval enhancement and large model reasoning technologies are combined, the problems of cross-document attack chain restoration and complex threat behavior logical reasoning are effectively solved, and the method has remarkable application value in the aspects of threat intelligence tracing, behavior prediction and defense decision assistance.

The invention discloses an audit report generation method and device, a storage medium and electronic equipment. The method comprises the steps that first auditing data, second auditing data and an auditing prompt template are obtained, the data format of the first auditing data is different from that of the second auditing data, the type of the first auditing data is the same as that of the second auditing data, and the auditing prompt template comprises a target prompt word; determining a plurality of groups of initial character strings according to the first audit data and the second audit data, and executing a target merging operation on the plurality of groups of initial character strings based on the target prompt word to obtain a target character string; and inputting the target character string and the audit prompt template into the target business model to generate a target audit report, and generating different parts in the target audit report by setting temperature parameters of different target business models. The technical problem that the efficiency of manually filling in the audit report is low is solved.

The invention provides a data processing method and device, equipment, a storage medium and a program product, and relates to the technical field of computers. The method comprises the steps that first question information input by a user is subjected to word segmentation processing and then subjected to semantic matching with a pre-configured lexicon, and second question information is determined; the second question information is normalized and vectorized and then subjected to similarity calculation with a pre-configured table library, an entity meeting a similarity threshold value and a standardized query template are determined, the pre-configured table library comprises a plurality of tables, and the standardized query template comprises a structure query language and a skeleton question; inputting the first question information, the entity and the standardized query template into a first large language model, and determining multi-table query information; and inputting the first question information and the multi-table query information into a second large language model, and determining response information. Based on the large language model, the cross-multi-table question can be effectively answered by deeply analyzing the user question and combining with the table library generated by multi-table preprocessing, and the problem that cross-multi-table answers cannot be effectively processed due to the fact that table question answering driven by the large language model mainly aims at a single table is solved.

The present invention discloses a frame title recognition method, storage medium and device based on target detection. By classifying supported resources and sampling a part of each type of resource, a corresponding recognition model is obtained through training. The subsequent frame title can be automatically realized, which not only greatly reduces the workload of early template entry, but also reduces the workload of subsequent frame title, and improves the degree of automation of intelligent correction.

1. The name of the design product: electronic device research and development experiment management graphical user interface. 2. The use of the design product: an electronic device. 3. The design points of the design product: the interface content in the graphical user interface. 4. The picture or photo that best shows the design points: front view. 5. The product carrier is the conventional design, and the rear view, left view, right view, top view, and bottom view are omitted. 6. The use of the graphical user interface: graphical user interface for research and development experiment management system. 7. Other circumstances that need to be explained: the front view is the task list interface of the research and development experiment management system. Clicking "+" experiment design" in the interface enters the change state diagram 1 task addition interface. The interface pops up the task information setting pop-up window. Fill in the relevant information in the pop-up window. Click the "experiment template" addition icon (the icon with a magnifying glass on the right side of the experiment template input box) to enter the change state diagram 2 task template addition interface. The interface pops up the experiment template selection pop-up window. The left side of the pop-up window is the template display area, and the middle is the process node display area. After selecting the required template in the template display area, the corresponding task node is generated in the process node display area. Click any node in the process node display area to enter the change state diagram 3 node information interface. The right side of the interface is the node editing area. Click "add group" in the area to enter the change state diagram 4 group setting interface. The interface pops up the "edit group" pop-up window. After setting the relevant parameters in the pop-up window, click "OK" to enter the change state diagram 5 complete group setting. Click "group design" in the change state diagram 5 under the three small dot icons on the right side of the group design bar to enter the change state diagram 6. The interface pops up the "material calculation" pop-up window. After setting the relevant parameters in the pop-up window, click "generate data" to enter the change state diagram 7. Click "preference settings" in the change state diagram 7 to enter the change state diagram 8. The interface pops up the execution mode setting pop-up window. Click "OK" after selecting "detailed view" in the pop-up window to enter the change state diagram 9 detailed mode experiment execution.

The present invention discloses a method and device for obtaining a custom instruction template. The method includes: obtaining a custom instruction template entry request sent by a target user; based on the custom instruction template entry request, collecting a static period target signal and an instruction entry period target signal; performing quality detection on the static period target signal and the instruction entry period target signal to obtain a signal quality detection result; if the quality detection result meets a preset quality condition, performing feature extraction on the instruction entry period target signal to obtain a signal feature; performing similarity matching between the signal feature and the corresponding signal feature of a pre-stored custom instruction template; if the similarity meets the preset similarity condition, the custom instruction template entry is successful; otherwise, the custom instruction template entry fails and a prompt message is output. The method can effectively screen out low-quality signals due to environmental factors, improve the quality of custom instruction template entry, optimize the personalized interactive experience, and enhance the device response accuracy and execution effect.

The invention provides a leading-edge technology maturity level interval discriminant analysis method and device, and the method comprises the steps: obtaining a technology text which is pre-collected for a to-be-judged technology, and constructing an original data set; preprocessing the original data set to obtain an updated data set; inputting each piece of data in the updated data set into a preset data extraction model, wherein the data extraction model extracts triple data consisting of a technical direction, a technical dynamic state and a technical index in each piece of data in the updated data set; determining a transition level of the to-be-judged technology based on matching of a preset transition keyword library and the updated data set; technical information including the triple data and the transition level is supplemented into a preset structured JSON cue word template, the filled structured JSON cue word template is input into a preset large language model, and the large language model outputs a maturity judgment level.

The invention provides an interaction method, electronic equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring an input interaction context; the interaction context and a preset system cue word template are input into a large language model, at least one candidate PEFT module is obtained, and the system cue word template is used for indicating the large language model to select and output the candidate PEFT module from a plurality of preset PEFT modules according to the interaction context; determining a target PEFT module according to the candidate PEFT module, and loading the target PEFT module to the large language model to obtain a large language model after the module is loaded; and inputting the interaction context into the large language model loaded with the module to obtain an interaction response. According to the method and the device, automatic selection and loading of the PEFT module can be realized, and the interaction performance and expression of a large language model for the current interaction context can be improved.

The invention discloses a dynamic test report generation method, system and equipment based on a template engine and an automatic tool and a medium, belongs to the technical field of software testing, and aims to solve the technical problem of how to efficiently, flexibly and accurately generate a test report, reduce the labor cost and improve the report quality at the same time. According to the technical scheme, a management template is used for creating, editing and storing report templates; inputting and processing data: receiving a project name and test data input by a user and an uploaded picture file; analyzing the data: analyzing the test data into structured dictionary or list data, and storing the picture to a specified directory; filling the template: dynamically filling the data into the template; generating a report: converting the filled template into a final Word document and PDF document report file; and outputting a report: storing the generated report file to a server or directly returning the report file to a user for downloading.