316 results about "Triplestore" patented technology

The TABLE function mechanism available in a RDBMS is used to integrate RDF models into SQL queries. The table function invocation takes parameters including an RDF pattern, an RDF model, and an RDF rule base and returns result rows to the SQL query that contain RDF triples resulting from the application of the pattern to the triples of the model and the triples inferred by applying the rule base to the model. The RDBMS includes relational representations of the triples and the rules. Optimizations include indexes and materialized views of the representations of the triples, precomputed inferred triples, and a method associated with the TABLE function that rewrites the part of the SQL query that contains the TABLE function invocation as an equivalent SQL string. The latter technique is generally applicable to TABLE functions.

The TABLE function mechanism available in a RDBMS is used to integrate RDF models into SQL queries. The table function invocation takes parameters including an RDF pattern, an RDF model, and an RDF rule base and returns result rows to the SQL query that contain RDF triples resulting from the application of the pattern to the triples of the model and the triples inferred by applying the rule base to the model. The RDBMS includes relational representations of the triples and the rules. Optimizations include indexes and materialized views of the representations of the triples, precomputed inferred triples, and a method associated with the TABLE function that rewrites the part of the SQL query that contains the TABLE function invocation as an equivalent SQL string. The latter technique is generally applicable to TABLE functions.

The invention discloses a multi-triplets extraction method based on the entity relationship joint extraction model, comprises: performing segmentation processing on the target text, and tagging position, type and whether is involved with any relation or not of each word in the sentence; the joint extraction model of the entity relationship is established; the joint extraction model of the entity relationship is trained; the triple extraction is performed according to the joint extraction model of the entity relationship; the tri-part tagging scheme designed by the present invention is in the process of joint extraction of the entity relationship an entity that is not related to the target relationship can be excluded; the multi-triplets extraction method based on the entity relationship joint extraction model can be used to extract multiple triplets, and based on the model of the triplet extraction method of the present invention other models have stronger multi-triplets extraction capabilities.

A method for enterprise application integration that uses “connectors” that can be instantiated via downloading (e.g., using Java® or other such technologies) to provide interfaces to respective disparate database systems. The databases systems may comprise any variety of now or heretofore known systems, e.g. SAP, Oracle, and so forth. The connectors can, for example, translate between a native language (or API) of the respective database systems and an internal language / protocol of the enterprise application integration system. To this end, the connectors can utilize a scripting language to access the respective database systems. Data retrieved from the database systems can be stored in a central data store in the form of RDF triplets, from which directed graphs can be generated for to generate presentations consolidated from the multiple database systems.

A service registry and repository based on a triplestore comprises: receiving a request to aggregate a service document; shredding elements of the service document to create logical objects within the triplestore; for each logical object, searching for all policy attachments logical objects that have a relationship with the logical object; for each located policy attachment, retrieving details of the policy and building a list of policies and associated logical objects in the repository; and returning an indication of the list of polices and associated logical objects. The list of logical objects and associated polices is used to compile a service document containing details of policies that have relationships with the selected service document logical objects. Objects that have associated polices are rendered with a hypertext policy icon next to the object and selection of the hypertext policy icon opens a new window with the policy details.

The invention discloses a medical text-oriented entity relationship joint extraction method. The method comprises the following steps of: recombining a medical text entity relationship data set according to a triple mode; vectorization representation is carried out on the medical text statement; a parameter sharing layer is constructed by adopting a bidirectional long-short-term memory network anda self-attention mechanism; predicting a head entity label by adopting a softmax function, performing relation-tail entity joint decoding by adopting CNN-softmax, enhancing joint decoding representation by combining a head entity prediction result, and training a parameter sharing layer and a joint decoding layer by adopting a joint loss function optimization mode. According to the method, the problems that entities and relationship categories and positions in the medical text are distributed unevenly, and multiple pairs of relationships appear in the same sentence at the same time are solved, and the quality and efficiency of entity relationship extraction of the medical text can be improved.

The invention discloses a relation extraction method based on combination of an attention mechanism and a graph long-short-term memory neural network. The method comprises the following steps of extracting context information in sentences through BiLSTM, and entity position information and entity label information are introduced to expand word vector features; extracting the sentence dependency structure tree through a Stanford Parser tool to generate an initial sentence structure matrix, and introducing an attention mechanism to perform attention calculation on the initial sentence structurematrix to obtain weight information of the structure matrix in the sentence; and taking the extracted sentence context information and the weight information of the sentence structure as input, and performing relationship extraction on the input by using a relationship extraction model based on the combination of an attention mechanism and a graph long-short-term memory neural network to finally obtain triple information of an entity. According to the method, evaluation is carried out on a TACRED data junction and a Semeval2010 task8 data set respectively, and the performance of the model is superior to that of an existing mainstream deep learning extraction model.

A method of operating a service registry and repository based on a triplestore comprises: receiving a request to aggregate a service document; shredding elements of the service document to create logical objects within the triplestore; for each logical object, searching for all policy attachments logical objects that have a relationship with the logical object; for each located policy attachment, retrieving details of the policy and building a list of policies and associated logical objects in the repository; and returning an indication of the list of polices and associated logical objects. The list of logical objects and associated polices is used to compile a service document containing details of policies that have relationships with the selected service document logical objects. Objects that have associated polices are rendered with a hypertext policy icon next to the object and selection of the hypertext policy icon opens a new window with the policy details.

The invention discloses a method for carrying out compression storage on adjacent matrixes of a sparse directed graph. The method comprises the steps of carrying out compression storage on the adjacent matrixes of the sparse directed graph by using a triple table; recording the positions of the first effective data of each row in the adjacent matrixes in the triple table to realize application algorithms of a part of related graphs. For the compression storage on the adjacent matrixes of the sparse directed graph, only the row number, the column number and the element value of each effective element in the adjacent matrixes are stored, in certain algorithms, a small space is required to store the number of effective elements of each row in the adjacent matrixes and the subscript of the first effective element of each row in the triple table. According to the method, on one hand, the advantages of the adjacent matrix representation method of the graph are kept, and on other hand, the problems of space waste when the sparse graph is represented by the adjacent matrixes are solved, the operation is simplified in certain algorithms, the time complexity of the algorithm based on the adjacent matrixes is reduced, and the software performance is effectively improved.

A Data Constraint Engine (100) for enforcing data constraints in a polyglot data tier (20) having a plurality of database-specific data stores (21, 22, 23) of various types such as an RDBMS (21), a Triplestore (22), and a MongoDB (23). The Data Constraint Engine uses the concept of a unified data model based on “records” in order to allow data constraints to be defined (using so-called “record shapes”) in a store-agnostic way. The Data Constraint Engine includes APIs (130) for processing incoming requests from remote clients (30) relating to data in the polyglot data tier, for example a request to create or update data in a data store. The APIs extract, from such a request, a record corresponding to the data specified in the request and a data source identifier identifying the data store holding the specified data. Then, on the basis of the record extracted by the interface, an appropriate record shape is extracted from a shapes catalogue (110), the record shape determining the structure of the record. Validators (120) each validate the record against the record shape according to various criteria such as format, data type, cardinality and slot count. If the record is validated, a record dispatcher (140) directs the specified data to the appropriate data store using the data source identifier. Data read from a data store can be validated in the same way.

The invention relates to a data storage method and query method for an RDF (Resource Description Framework). The data storage method comprises the steps of designing entity-oriented RDF data storage structure and storage mapping; converting a URI (Uniform Resource Identifier) and a literal amount of RDF data into 64-bit binary data; and storing the 64-bit binary data according to the designed storage structure. The data query method comprises the steps of analyzing and converting an SPARQL query statement; estimating single query cost according to an analysis result of the whole data set and a connection relationship among queries by a plurality of query triples in the SPARQL statement; and finally generating a least-cost query process. According to the data storage method and query method, the data comparison speed can be greatly increased and the storage space can be reduced; and compared with a conventional method for directly converting SPARQL into SQL to perform query, the query method provided by the invention has the advantages that the query efficiency is greatly improved and the query method can be used in the fields of Web data management, Web semantic retrieval and the like.

The embodiments described herein comprise a free association engine running on a server for capturing input data and recorded observations received from one or more electronic devices or other sources, optionally during user encounters with physical objects associated to the input data and observations; deriving underlying entities and associations of the input data using rules applied to the input data and observations; storing the derived entities and associations in a graph or triplestore database; executing dynamic operations based on the derived entities and associations; and generating visualizations concerning the express and inferred entities and associations. The embodiments allow a user to create rules that are applied to input data and observations to generate entities and associations. The embodiments provide the ability to generate dynamic views of the derived entities and associations based upon a user's selection of sources of the input data and observations and use these views to visualize the data in different ways to different users under different circumstances and power semantically precise user searches and automated recommendations.

The invention discloses an entity relation joint extraction method and device based on a neural network. The method comprises the following steps: firstly, fusing a pre-trained ERNIE word vector, a pre-trained CWV word vector and relative position information of words by utilizing a single-layer Transform network; secondly, improving a handshake labeling strategy, and introducing a vectorized entity type label to make full use of entity type information; then, obtaining candidate entity relationship triples in the sentences through an annotation decoding method; and finally, indexing articles, sentences, entities and relationships by using UCL according to the characteristics that the UCL national standard can efficiently organize contents and effectively associate information. The entity relationship triple can be directly extracted from the sentence, the problems of redundant entities, nested entities, overlapping relationships and the like are avoided, a small amount of annotation data can be used for quickly checking the entity relationship, and the data can be objectively and normatively indexed.

A client machine writes to and reads from a virtual disk on a remote storage platform using a storage protocol. Metadata is generated and is stored in replicas on different metadata nodes of the storage platform. A modified log-structured merge tree is used to store and compact string-sorted tables of metadata. During file storage and compaction, a consistent file identification scheme is used across all metadata nodes. A fingerprint file is calculated for each SST file on disk that includes hash values corresponding to regions of the SST file. To synchronize, the fingerprint files of two SST files are compared, and if any hash values are missing from a fingerprint file then the key-value-timestamp triples corresponding to these missing hash values are sent to the SST file that is missing them in the SST file is compacted with the missing triples to create a new version of the SST file. The other fingerprint file is then analyzed the same way.

The invention provides an entity relationship extraction method based on dependency syntactic analysis and rules. According to the method, the dependency relationship between the syntactic structure of the sentence and the entity in the sentence is determined through dependency syntactic analysis, the entity relationship in the text is recognized and extracted by making a rule, formation of the triad is restrained, and the accuracy of entity relationship extraction is improved. According to the method, the input text is directly extracted, a keyword library does not need to be formed, and universality is high, the triple output speed is high, and the operation efficiency is improved.

The invention belongs to the technical field of natural language processing, particularly relates to a machine reading understanding method, system and device based on external knowledge enhancement,and aims to solve the problem that the existing machine reading understanding method does not utilize graph structure information among triples, so that the answer prediction accuracy is relatively low. The method of the system comprises the following steps: generating context representations of entities in a question and an original text; based on an external knowledge base, obtaining a triple set of each entity in the question and original text and a triple set of adjacent entities of each entity in the original text; based on the triple set, obtaining knowledge sub-graphs of each entity through an external knowledge graph; updating the fused knowledge sub-graph through a graph attention network to obtain knowledge representation; and splicing the context representation and the knowledgerepresentation through a sentry mechanism, and obtaining answers to the to-be-answered questions through a multilayer perceptron and a softmax classifier. According to the method, the graph structureinformation among the triples is utilized, so that the answer prediction accuracy is improved.

The invention provides an automatic knowledge graph construction method based on multi-source heterogeneous power equipment data, and aims at structured data such as table data, various types of table data are transferred into an Excel table, and a data conversion rule and a data processing rule are defined, so that mapping from the structured data to a knowledge graph is completed; automatic extraction from a database to a knowledge base is performed, information extraction is performed on power text and webpage information by mainly adopting a deep learning method aiming at unstructured text data, an entity identification relation extraction process is completed, SPO triple extraction on the unstructured data is achieved, and finally integrated data is stored into Neo4j in a form of an RDF (Resource Description Framework) file. Visual display of the knowledge graph is achieved through a Neo4j graph database, and semantic query is performed by using a cypher query language.

The invention provides an entity relationship extraction method and device, electronic equipment and a storage medium, and relates to the technical field of natural language processing. The method comprises the steps of obtaining a text of a to-be-extracted entity relationship, and processing the text through an entity and relationship joint extraction model to obtain a first entity relationship triple; determining words in the text and labels corresponding to the words, matching the words and the labels according to the relation rule set, and determining a second entity relation triple; constructing entity pairs according to the words and the labels, processing the text through a relationship classification model, and determining third entity relationship triples for the entity pairs; anddetermining an entity relationship extraction result according to the first entity relationship triad, the second entity relationship triad and the third entity relationship triad. According to the invention, the entity relationship extraction accuracy can be improved.

The invention discloses a legal knowledge graph construction method and equipment based on entity relationship joint extraction. The construction method comprises the following steps: constructing a triple data set; designing a model architecture and training a model, wherein the model architecture comprises a model coding layer, a head entity extraction layer and a relation-tail entity extraction layer; judging a relationship between text sentences; and carrying out triple compounding and map visualization. According to the design of the model architecture, a Chinese bert pre-training model is adopted as an encoder, and the Chinese text encoding effect is good. According to the entity extraction part, two BiLSTM binary classifiers are adopted to judge the starting position and the ending position of an entity, and the entity in a phrase form in a text can be effectively extracted. According to the method, the head entity is firstly extracted, then the tail entity corresponding to the entity relationship is extracted from the extracted head entity, and when the entity relationship and the tail entity are extracted, not only is coded information of sentences used, but also coded information of the head entity is fused. According to the method, the legal knowledge graph with relatively high accuracy can be obtained.

An automatic construction method of an information security knowledge graph based on deep learning comprises the following steps: step 1, constructing an ontology database of the information securityknowledge graph; step 2, labeling the collected data based on a heuristic rule mode to obtain a training set of entity recognition, designing a model structure based on Bi-LSTM for learning to serve as a model of entity recognition, and extracting entity words of the text; and step 3, performing automatic labeling based on triple knowledge to obtain a training set of relationship extraction, and designing an attention mechanism-based LSTM structure as a model of relationship extraction so as to extract triples (relationships) in the text.

The invention discloses a knowledge graph construction method for log data. The knowledge graph construction method comprises the following steps: S1, acquiring log data in a specified time window; S2, preprocessing the log data, and converting the log data into structured log data; establishing a template tree according to the template tree establishment strategy; S3, acquiring a log template based on the established template tree; S4, based on the structured log data and the log template, carrying out entity and relationship extraction, acquiring multiple pieces of triple data used for output, wherein the triple data is data in the form of 'entity-relationship-entity '; and S5, fusing the plurality of triple data according to a time sequence to obtain a knowledge graph of the log data. The unstructured text data can be converted into the structured data, the operation and maintenance log knowledge graph is constructed, and visual services with different functions are provided for theentity information of the business from different latitudes around the entity information in the knowledge graph.

The invention discloses a named entity linking method based on knowledge base feature extraction. The method comprises the following steps of (1), extracting triple entries with specified features from Freebase data dump to form a relational data table, and saving the relational data table in a knowledge base; (2), designing a complex rule, and searching for a plurality of Freebase Objects which are closely related to entity reference from the knowledge base as candidate entities; (3), using a statistical-based method to design and extract the entity reference and features of the candidate entities, and Embedding is conducted on the features; (4), Embedding for feature extraction is used as an input of a multi-layer neural network to obtain the probability of each candidate entity as a target entity, and returning to a Freebase MID of the candidate entity with the highest probability is conducted. The method combines a complex rule-based candidate generation technology and a statistical learning-based candidate sorting technology, and a set of processing framework suitable for entity linking is established for a specific type of named entities, so that convenience is provided for auser to obtain an entity linking result by adopting a batch processing manner.

A semantic document generation system is described. The semantic document is composed of document details, people and meta-data. The semantic document is self-aware of the information it contains. The semantic document's structure and terms are governed by legal, logical and party related rules. A semantic contract can be created from a semantic document generation system. The semantic document generation system receives an indication of a type of a document to be generated and plurality of terms for the document from a plurality of sources. The terms are converted into triples. A plurality of rules governing the terms of the document is applied to the triples to generate a knowledge graph and determine whether terms from the different parties are compatible. The terms are determined to be compatible in a case where the plurality of rules governing terms of the document is satisfied. If at least one set of terms is non-compatible, the system reconciles the non-compatible terms in the generated knowledge graph until all the terms are compatible, and generates the document based at least on the reconciled knowledge graph.

The invention provides a method for analyzing a sparse semantic relationship by combining a BiLSTM-CRF algorithm and an R-BERT algorithm, which comprises the following steps: acquiring text data of emerging industries through a web crawler, and performing semi-supervised annotation on the text data; preprocessing the labeled text data, and constructing a training data set and a verification data set; training a BiLSTM-CRF algorithm model and an R-BERT algorithm model according to the training data set and the verification data set; extracting entities contained in the text data to be predictedthrough the trained BiLSTM-CRF algorithm model; predicting the relationship between the text data to be predicted and the entities through the trained RBERT algorithm model, and establishing relationship connection between the related entities; and extracting the triad pair of the semantic relationship of the text data to be tested according to the established relationship connection, and completing semantic analysis of the text data to be tested. The invention provides a high-precision semantic relation extraction method for information extraction of an unstructured text, a BiLSTM-CRF algorithm model is used for extracting required entities in the text, and the relation between the text and the extracted entities is predicted through an R-BERT model.

The invention discloses a knowledge graph intelligent question-answering method fusing a pointer generation network, and belongs to the field of artificial intelligence question-answering. According to the technical scheme, a word segmentation tool is used for carrying out word segmentation and checking on original text and question parts in a WebQA data set; performing named entity recognition on the data after correct word segmentation by using a BiLSTM-CRF model; querying a triple corresponding to the identified entity in a Neo4j database; counting the occurrence frequency of each word in the corresponding triad, and storing the queried words in the triad into a knowledge word list according to a word frequency sequence; using a deep learning method to obtain word vectors of the question sentences; and constructing a generative model, and returning an answer. The method has the beneficial effects that entity recognition is performed on texts by using a deep learning technology, knowledge is quickly queried by using a knowledge graph technology, and the problems that returned answers are stiff and single and the storage space in a knowledge base is incomplete are effectively solved in combination with a generative model; the time for obtaining the answer is saved, the intention of the user is more fully understood, and the answer more conforming to the reading mode of the user is returned.

The knowledge graph construction method for the multi-source Chinese financial announcement document comprises the following steps: structuring the hierarchical relationship of each chapter of the document, and constructing a relatively complete document structure tree; labeling all the title data; unifying the length of the title to a preset word number, and carrying out word embedding coding at a character level by using BERT to obtain a corresponding vector representation; dividing the processed data set into a training set and a test set, and training to obtain a title classification model; classifying the document titles by using a title classification model; the complex and effective knowledge of the effective text blocks is masked; constructing a semantic model with a mask, constructing a multi-source similar generalization mask Bi-LSTM semantic model (M-MST model), and feeding the M-MST model for training to obtain a knowledge extraction model; according to the knowledge extraction model, combining with an external knowledge base to obtain an entity relationship triple; and constructing a multi-source financial announcement document knowledge graph and realizing incremental updating or expansion. The invention further discloses a system for implementing the knowledge graph construction method for the multi-source Chinese financial announcement document.

The invention relates to a method and a device for detecting authenticity of triples in a knowledge graph. The method comprises the following steps of: forming a substrate; extracting a first vector for describing triad structure information in the knowledge graph and a second vector for describing triad description information in the knowledge graph from the text; fusing the first vector and thesecond vector; obtaining fusion vectors, according to a preset network structure of a capsule network, constructing a scoring function used for defining a triple score, scoring function, constructinga loss function for training a capsule network; constructing a sample set according to the fusion vector, training the capsule network model according to the sample set and the loss function to obtaina trained capsule network, inputting the fusion vector corresponding to the a to-be-predicted triad into the trained capsule network, and determining whether the relationship of the triad is accurateor not according to the output value of the scoring function. By adopting the method, tThe accuracy of triple authenticity detection can be improved.

The invention discloses a knowledge graph pre-training method based on structured context information, and the method comprises the steps: constructing an instance composed of context triples for a target triple, employing a triple integration module to code each context triple of the instance, and obtaining an integration vector; forming a context vector sequence by the integrated vectors of allthe context triples of the instance, and encoding the context vector sequence by adopting a structured information module to obtain structural representation vectors of the triples; and calculating the structure representation vector of the triple by adopting a universal task module to obtain a label prediction value of the triple, and updating the structure representation vector of the triple based on the cross entropy loss of the label prediction value of the triple and the label true value until the training is finished to obtain an optimized structure representation vector of the target triple. The structural representation vector of the triple obtained by the method combines contextual information.