18 results about "Seed entity" patented technology

The invention belongs to the crossing field of artificial intelligence and a power system, and discloses a power grid fault handling plan knowledge retrieval method and a related device. The power grid fault handling plan knowledge retrieval method comprises the following steps: on the basis of an obtained expression, generating a query by utilizing a large language model, and performing retrieval in a constructed graph database to obtain a power grid fault handling plan knowledge retrieval result; when the graph database is constructed, firstly, all themes in a free text of a selected power grid fault field are recognized, and an entity which can best represent the theme in each theme serves as a seed entity; then, entity relation triads in the free text are extracted through a large language model and serve as candidate triads, integration is carried out to obtain a fused knowledge graph, the fused knowledge graph is stored, and a constructed graph database is obtained. The problems that an existing retrieval method is low in query efficiency and poor in accuracy are solved.

The invention belongs to the technical field of large language models, and provides a prediction-guided knowledge retrieval enhanced question-answering system, which comprises a knowledge graph vectorization module for constructing a vector representation of an entity relationship in a knowledge graph; the two-way matching module is used for extracting a related entity set from the knowledge graph according to a query text input by a user, and extracting a target entity set from the knowledge graph according to collected user information; the retrieval enhancement module is used for carrying out weighted fusion on the related entity set and the target entity set to obtain a seed entity set; pruning the redundant structure in the seed entity set based on a preset semantic understanding model to obtain a simplified sub-graph; and generating and outputting a corresponding answer according to a query text and the simplified subgraph input by the user. According to the technical scheme, the user intention can be better reflected while the reliability of the query result is improved.

The invention provides a policy intelligent question-answering method and system based on a knowledge graph and a storage medium. The method comprises the following steps: performing intention recognition and key information extraction on user query content to obtain a query intention and a key entity corresponding to the user query content; determining a plurality of seed entities in a pre-constructed knowledge graph according to the key entities; based on a multi-hop retrieval algorithm, according to the seed entities, determining an importance score of each entity in the knowledge graph; according to the importance score, determining a target entity in the knowledge graph and a target relationship directly associated with the target entity; and generating a query result corresponding to the user query content according to the target entity, the target relationship and the user query content by applying a pre-trained large language model, and outputting the query result. The method is used for achieving the effect of efficiently and accurately outputting the query result.

The invention provides a language model-based teaching subject knowledge graph construction method, which comprises the following steps of: obtaining a teaching subject field seed entity through a language model, after forming a two-tuple, retrieving a related text for relation extraction in the following steps, and extracting relation words by utilizing an unsupervised method according to the two-tuple and the corresponding text to form a triple; the semantics of the relational words are defined through the sentences corresponding to the relational words instead of simply mapping through the semantics of the relational words, the mapping is more accurate, the mapping range is wider, entities and relations of the knowledge graph are automatically complemented by utilizing the capability of a language model, the process is unsupervised, additional annotation is not needed, and the mapping efficiency is improved. And the construction efficiency of the knowledge graph is greatly improved.

A system is adapted to identify suspected mule accounts. It includes a processor configured to select seed entities, and identify a network of accounts associated with each seed entity. For networks that includes at least one known mule account, the processor computes a similarity score between each pair of accounts and, based on the similarity scores, clusters the accounts, labels the clusters as to whether they are high-mule-rate clusters, and uses the clusters to train a link prediction model. The processor then, in real time, receives a transaction for an entity, identifies a second network of accounts associated with the entity and, with the link prediction model, for each pair of accounts in the second network, computes a mule risk score and, if the score exceeds a second threshold value, adds the accounts in the pair of accounts to a suspected mules list.

A malicious campaign detection system (“detection system”) detects malicious campaigns according to a graph structure of network-related entities and uses the graph structure to generate descriptions of the malicious campaigns and to monitor the malicious campaigns for changes. As the detection system ingests seed entities of interest, the detection system retrieves seed entity graphs and then prunes / merges the graphs. The detection system applies rules to the pruned merged / graphs to determine whether each pruned / merged graph is malicious. Finally, the detection system propagates verdicts in the detect malicious graphs and identifies entities having malicious propagated verdicts as being important / releasable for the corresponding malicious campaign.

The present invention is applicable to the field of knowledge graph technology and provides an entity alignment method based on noise optimization and attribute fusion, comprising the following steps: performing noise initialization on an entity seed set to obtain accurate noise entity pairs; injecting a noise-aware optimization generator to obtain more noise entity pairs around the relevant distribution of true seed entity pairs; after generating noise entity pairs, feeding them into a noise-aware optimization discriminator to distinguish true seed entity pairs from noise entity pairs; iterating the noise-aware optimization generator and the noise-aware optimization discriminator to obtain a trust score, which is compared with a threshold θ to obtain clean labeled data; structurally embedding the obtained true seed entity pairs through GCN to ultimately obtain a structural alignment model. This method utilizes a combination of noise optimization and attributes to not only reduce complexity but also improve the effect of entity alignment, achieving excellent entity alignment results while reducing experimental overhead.

The invention discloses a psychological health assessment field knowledge graph construction method and a storage medium. The method comprises the following steps: S1, data preprocessing and structuring: processing an original document of a mental health assessment scale, and converting the original document into scale descriptive text and structured data; s2, entity extraction and seed entity set construction: extracting measurement factors and dimension entities in a single scale range, and constructing a standardized seed entity set; s3, relation triad extraction and deduplication: extracting structural and semantic triads under the constraint of a single scale, and performing semantic discrimination and deduplication by using LLM; and S4, multi-level map fusion and dynamic mapping are carried out, a global three-level semantic skeleton is constructed, entities are normalized and mapped to a standard concept, and problem access and map completion are realized. According to the method, the problems of cross-scale semantic confusion, inconsistent entity granularity and lack of hierarchical logic in the prior art are solved, and the specialty, accuracy and traceability of the knowledge graph are improved.

The present invention discloses a large model-enhanced semi-open knowledge extraction method and system, which can accurately screen high-quality corpora related to the issues of concern, and realize efficient and accurate knowledge extraction. It includes: first, based on given seed entities, relevant corpora containing these seed entities are retrieved; then, based on the purpose of knowledge extraction and the issues of concern, the degree of association between the retrieved corpora and the focus is calculated, and the corpora with high association are screened out as the source corpora for knowledge extraction; then, the large model is used to extract triples with the seed entity as the head entity from the source corpus; the large model is used to judge the confidence of the extracted triples, and the triples with confidence exceeding the threshold are screened out; the grammar, part of speech, etc. of the obtained triples are further checked, and the triples that meet the specifications are screened out; finally, through the entity, relationship standardization alignment, synonym extraction fusion and other technologies, the knowledge is deduplicated and standardized, and the final required triples are obtained.

A large model question and answer data generation method and device, medium and electronic equipment, relating to the technical field of artificial intelligence, the method comprises: acquiring a seed entity, and searching the seed entity to obtain summary content, the summary content being summary content of multiple search results related to the seed entity; determining a set of associated entities of the seed entity based on the summary content, the set of associated entities including multiple associated entities; generating a target question from the set of associated entities, and generating the large model question and answer data from the target question and the seed entity. The disclosure generates a target question based on the set of associated entities generated from the summary content, which not only ensures the timeliness of the large model question and answer data generation, but also ensures the uniqueness of the answer.

The present invention proposes a knowledge graph fact-checking method enhanced with hint subgraphs, comprising: S1, locating entities of the factual statements to be checked and identifying seed entities; S2, using the seed entities as seed nodes, performing a depth- and width-restricted heuristic bundle search on the knowledge graph to construct a local hint subgraph for the factual statements to be checked; S3, inputting the hint subgraph as a structured context into a large language model to generate a query program composed of three types of restricted functions: SEARCH, MATCH, and VERIFY; S4, executing the query program on the knowledge graph through an interpreter and outputting whether the factual statements to be checked are supported by the current knowledge graph. This invention, by introducing a hint subgraph constraint program generation process, solves the problems of easily fabricating relation names, omitting intermediate entities, and query path mismatches in existing methods, thereby improving the matching degree between the generated program and the real structure of the knowledge graph.

The application discloses an entity alignment method and system based on information entropy fusion of multi-view features, equipment and medium, the method comprises: semantic initialization representation is carried out to the entity in the entity set of the source knowledge graph and the target knowledge graph to be aligned, and a fixed-length entity embedding vector is obtained; the fixed-length entity embedding vector is subjected to attention map convolution based on semantic similarity, and a structure-aware entity embedding is generated; for the entity pair in the entity set of the source knowledge graph and the target knowledge graph to be aligned, multi-view feature extraction is carried out based on the structure-aware entity embedding, then quantitative evaluation is carried out, and a fusion similarity is obtained after information entropy fusion; based on the fusion similarity, the similarity is calculated after forcibly pushing away the interference nodes in the feature space, and matching is carried out through a stable matching algorithm. The application can realize adaptive fusion of multi-view features, can effectively expand high-quality supervision signals under the condition that seed entities are extremely scarce, and can eliminate feature confusion and one-to-many matching ambiguity caused by implicit mutually exclusive entities.

This application relates to the field of artificial intelligence technology, and in particular to a knowledge graph question answering method, system, and storage medium based on structural induction and differentiable connectivity constraints. It decomposes natural language questions to obtain seed entities, employs personalized random walks around these seed entities to obtain their structural importance distribution, and integrates this distribution with semantic similarity as a retrieval criterion to filter candidate entity subgraphs. A gated graph neural network with geometric features such as hop count and node degree is introduced into the candidate subgraphs for structural inductive reasoning. Furthermore, differentiable connectivity constraints are constructed to softly model entity rewards and edge costs, ensuring the compact connectivity of the reasoning subgraph. End-to-end optimization is performed by combining question answering main loss, ranking loss, and connectivity regularization, thereby achieving accurate reasoning and prediction of answer entities in the knowledge graph.

The invention belongs to the technical field of artificial intelligence, and relates to a multi-agent-based knowledge graph automatic construction and query method and system. The method comprises the following steps: performing entity identification on an input document to obtain an entity set; determining a theme of the input document, and obtaining a seed entity according to the association strength between each entity in the entity set and the theme; extracting a relationship between seed entities based on a layered strategy label system; the seed entities and the extracted relations form candidate knowledge triples, and verification of different dimensions is carried out on the candidate knowledge triples through multi-agent cooperation; storing the knowledge triples passing the verification into a knowledge base to form a knowledge graph; and querying by using the knowledge graph to obtain answers to the questions. According to the method, the knowledge can be organized into a traceable logic chain, the interpretability and reasoning capability of the knowledge graph are improved, the consistency and reliability of the stored knowledge can be improved, and the timeliness maintenance capability and query accuracy of the knowledge base can be improved.

This invention belongs to the field of privacy computing technology and provides a method for constructing a multi-departmental government knowledge graph and making collaborative decisions with intelligent agents based on asynchronous federated learning. The method includes: CLIP joint embedding feature extraction, cross-departmental collaborative updating, relational reasoning and cross-departmental entity matching, comprehensive judgment and decision-making execution path matching, construction of a domain-level large graph and exception condition library, construction of an RPA rule base, and generation of automated processes. This invention achieves semantic-level fusion of heterogeneous data through seedless entity alignment and multimodal embedding technology, breaking down data silos; improves the rationality of human-machine collaboration through a two-factor threshold judgment mechanism; and enhances the automation level of processes by automatically mapping stable business rules to RPA action sequences.

The embodiment of the present application provides an entity recommendation method, device, storage medium and equipment. After obtaining the input text, the input text is first semantically understood to determine the search intent and seed entity words. When the search intent matches the seed entity words and the search intent belongs to the precise recommendation intent, the entity relationship is determined based on the search intent, and the seed entity words are mapped to the standard entity words in the medical knowledge graph; the entity words in the medical knowledge graph that have an entity relationship with the standard entity words are recommended as recommended entities. In other words, in the embodiment of the present application, entity recommendations are made based on the medical knowledge graph only when the search intent matches the seed entity words and the search intent belongs to the precise recommendation intent, thereby ensuring the accuracy of medical entity recommendations.

The invention is applicable to the technical field of multi-modal knowledge graph entity alignment, and provides a multi-modal knowledge graph entity alignment method based on mutual information enhancement and iteration, which comprises the following steps of: obtaining triple text information; text description information and structure information of entity relations or attributes are learned through a cross attention mechanism multilayer perceptron, and relation embedding vector representation is obtained; embedding other features; learning from existing alignment seeds by using a bidirectional iteration strategy, expanding training data by generating new alignment seeds, and progressively optimizing an entity alignment task through dynamic data enhancement and a semi-supervised learning mechanism; and displaying a classification result of final entity alignment by adopting a cross-view contrast learning method. According to the method, semi-supervised entity alignment is realized by utilizing an alignment method of triple text information and structure information, the dependence on entity name text information is reduced, and the dependence on a seed entity set is reduced.

The invention discloses a seed detection device intelligent management and control system and method based on an unreal engine, and the construction method comprises the steps: constructing a three-dimensional model of each seed entity detection device, and importing the three-dimensional model into an unreal engine platform; a material editor is used for adding materials and textures to the three-dimensional model, rendering is carried out, and a virtual model is obtained; the virtual model and the corresponding seed detection entity equipment are interacted through blueprint compiling, data transmission, remote control and state feedback between a virtual digital space and real equipment are achieved, and the seed detection entity equipment is monitored, controlled and simulated through the system; a control interface is perfected, an information interface, a menu bar and a toolbar module are designed, an intelligent management and control system is formed, and remote management and interaction functions are provided.