Apparatus and methods for preparing data

Abstract

This invention relates to an apparatus and method for preparing data from a relational database to form data blocks in a graph database and / or allocating them to data blocks in a graph database using R2RML. "R2RML" stands for "RDB to RDF Mapping Language," where "RDB" stands for "Relational Database" and "RDF" is an abbreviation for "Resource Description Framework." According to the invention, implementations of at least one R2RML unit, and if necessary, other R2RML units, are proposed. Identifying and improving the quality of R2RML mapping files can initiate optimization of the R2RML mapping process and its efficiency, and, if necessary, visualize the R2RML mapping after quality improvement and / or after optimizing the R2RML method.

Description

Technical Field

[0001] This invention relates to an apparatus and method for preparing data from a relational database, particularly data containing sensor data, to form data blocks in a graph database and / or allocating them to data blocks in a graph database using R2RML. "R2RML" stands for "RDB to RDF Mapping Language," where "RDB" stands for "Relational Database" and "RDF" is an abbreviation for "Resource Description Framework."

[0002] "RDF" (Resource Description Frame) is an internet protocol and a technical interpretation of logical statements used on the internet to formulate information about any transaction (resource). To have globally unique identifiers for resources, these identifiers are formed according to conventions similar to "URLs" (Uniform Resource Locators). In this way, data can be automatically and reasonably presented to humans in a uniform manner worldwide.

[0003] This preparation of relational data, such as sensor data, for example, for forwarding sensor data stored in columns and rows in industrial facilities, i.e., data from a relational database, to train artificial intelligence, wherein, optionally, the apparatus may also include a display device for presenting transformations or the progress of transformations, and the method includes allocating data from the relational database to RDF data blocks and thereby enabling the generation of RDF data blocks and / or including transferring data to a graph database. Background Technology

[0004] In modern IT infrastructure, relational databases are a core component. Data stored in relational databases not only directly controls and regulates equipment and / or automated devices, but also controls and regulates entire power supply systems, traffic control systems, industrial automation facilities, and more.

[0005] All heterogeneous data resides in relational database systems, which are based on the collection of tables. Data sets exist in these tables in a structured manner, with columns and rows being recorded and / or stored. This data exists in a structured way and is stored in the database. This data can be automatically identified and used based on its coordinates in these tables.

[0006] To prepare this data, the data in the relational database is transformed into data in the graph database. This involves synchronizing, transforming, and / or converting the data into a form that can be further processed for storage in the form of a graph database. Data preparation can also be the first step in a data analysis project and can include multiple discrete tasks.

[0007] In the technological process, relational data—in RDB format—is prepared to form RDF format data, which serves as the basis for automated control and regulation. This is because the resulting RDF format data, which can be used as the basis for a graph database, can be read and processed by machines in a way that differs from the generated and / or collected RDB format data. Therefore, it is absolutely essential in the automation process that:

[0008] First, data is collected, such as records of surface properties, roughness, rheological measurements, stress modulus, etc., through sensors and monitoring devices like cameras. This data—like much of the data from long-lived equipment like steam turbines—is typically stored as relational data in RDB format, and...

[0009] - Secondly, these source relational data in RDB format are converted into machine-processable and / or machine-readable data in RDF format, because industrial regulation and control equipment, such as heating, conveyor belt speed, robot movement, valve opening and closing, etc., mainly use graph data in RDF format to work.

[0010] Here, the data from the relational database is first allocated to RDF data blocks in a user-defined manner. Allocation to RDF data blocks provides the possibility of displaying existing relational data within the RDF data model.

[0011] An RDF data block always consists of three units: subject, predicate, and object, where the resource acting as the subject is described in more detail by another resource or value acting as the object. Together with the other resource acting as the predicate, these three units form a triple. The RDF model is a data model based on directed graphs with well-defined formal semantics. Data in an RDF is statements about resources. Here, these statements are modeled as triples. The collection of triples forms a graph—such as a mathematical graph—and is called an RDF model. A triple in an RDF model is a statement consisting of a subject, predicate, and object. The collection of triples forms a mathematical graph and is called an RDF model. For example, graph databases and / or the Google Knowledge Graph are based on this representation and storage of data.

[0012] A resource is something explicitly identified and requires some form of declaration. The subject and predicate are always the resource. The object can be the resource itself or simply text, where the text is an additional indicator, such as a date indicator or a truth value.

[0013] Data preparation is typically the first step in any approach aimed at generating data-driven applications based on heterogeneous data. However, preparing data stored in relational databases still encounters some errors because the transformation of relational data and data interconnected through graphs is difficult and complex.

[0014] Various methods have been used to transform data from relational databases into graph databases, but R2RML, a user-defined language for expressing assignments from relational databases to RDF data blocks, has proven useful here. This assignment provides the possibility of displaying existing relational data in an RDF data model on the output device and / or in memory, expressed in a structure and target vocabulary chosen by the assigner. The R2RML assignment itself is an RDF graph and is written in Turtle syntax.

[0015] R2RML supports different types of mapping implementations. For example, a processor or, more generally, an R2RML unit can provide a virtual SPARQL endpoint through allocated relational data, or it can generate RDF dumps or automatically provide an interface for relational data.

[0016] SPARQL (Protocol and RDF Query Language) is a graph-based RDF query language. An RDF model is a data model with defined formal semantics, based on a directed graph—unlike table-based relational databases.

[0017] "Mapping" or "data mapping" refers to the process of mapping data elements between different data models. For example, data mapping can be achieved using graph mapping tools, which automatically create executable transformation procedures. In the case of so-called "data-driven mapping," heuristics and statistics are used to attempt to automatically identify complex mappings between two data sources, including identifying data components, data merging, or arithmetic relationships. This also allows for the identification of anomalies that do not correspond to the identified mapping logic, i.e., those that may be errors.

[0018] The R2RML assignments discussed here assign data from specified relational data sources, such as sensors, measuring devices, industrial controllers, etc., to a set of terms within a specific subject area and the relationships that exist between these terms, in one or more languages ​​compiled and / or formally ordered “graph” representations. These automatically performed R2RML assignments then form the basis for storing this data in a graph database or RDF file.

[0019] An example of an industrial knowledge graph is a graphical representation of the paths from and between nodes. These nodes, in an industrial system, represent physical objects such as sensors, industrial controllers and / or control systems like PLCs (Programmable Logic Controllers), robots, machines, drives, tools, parts lists, or other hardware components. However, they can also correspond to more abstract entities, such as the attributes and configurations of these physical objects, like production schedules and machine, robot, and / or equipment characteristics. Abstract entities include, for example, IP addresses, data types, or applications executed by the industrial system.

[0020] In Figures 1a to 1c—the prior art—it is shown how the R2RML map of Figure 1a is represented in the form of triples—in a manner that reproduces the Turtle syntax, as shown in Figure 1b. This R2RML map is then generated graphically as a triple map, i.e., a TripleMap as shown in Figure 1c.

[0021] This type of R2RML triple map can be mapped onto output and / or display devices such as monitors, for example, like... Figure 2 As shown in the diagram. For example, R2RML assignments are visualized graphically on a display device. Here, nodes connected via paths are shown, such as in... Figure 2 As shown in the prior art.

[0022] Figure 3A known scheme of R2RML process is illustrated: Data from relational database 1 is transferred to R2RML module 4 via a first interface 3 through a first R2RML unit 2 for pre-processing and / or data mining, for example, in the form of R2RML mapping files. In R2RML module 4, the data is prepared by an R2RML parser 5, and RDF data 6 is generated from it. This RDF data 6 is stored in memory 8 as a data pool for the graph database or as RDF data blocks through the final interface 7 of R2RML module 4. Optionally, the allocation performed can be depicted here—as shown by the dashed lines—by a computer-implemented ontology process 9. This depiction 9 in… Figure 2 It reappears in the middle.

[0023] The known methods and known devices including R2RML modules have the disadvantage that the allocation is prone to errors due to the various complex method steps. However, these errors only manifest when reading memory 8. Unfortunately, visualization via ontology process 9 is also problematic due to the complexity of this ontology process—see [link to documentation] Figure 2 — And it is not possible to mark the error here.

[0024] Thus, when transforming relational database data into graph database using an R2RML processor, quality checks are required—preferably automatically. Summary of the Invention

[0025] Therefore, the object of the present invention is to overcome the shortcomings of the prior art and to describe an apparatus for preparing relational data by means of R2RML to form data blocks as a data pool for a graph database, and a method for transforming relational data by means of R2RML, which makes it possible to identify the quality of R2RML mapping and provide improvements and optimizations, thereby enabling the preparation of relational data to be performed optimally, more economically, more energy-efficiently and more quickly.

[0026] This task is solved by the subject matter of the present invention, which is described in the specification, Figures 4 to 9 —Because of Figures 1a to 1c and Figure 2 and 3 This is an illustration of the prior art—and disclosed in the claims.

[0027] Correspondingly, the subject of this invention is an R2RML module for preparing relational data to form RDF data blocks, the R2RML module having:

[0028] At least one interface with at least one storage area used to store the generated RDF data blocks;

[0029] ο and at least one interface with preprocessing;

[0030] at least one interface with a display device;

[0031] o at least one interface with the parser; and o one or more processors configured to generate at least one generation of RDF data blocks;

[0032] and

[0033] one or more R2RML parsers, the R2RML parsers being adapted to receive relational data, decompose the relational data, and convert the relational data into a first R2RML mapping;

[0034] at least one storage area for storing the generated RDF data blocks.

[0035] in

[0036] With the aid of a first R2RML parser, the R2RML module is adapted to allocate the received relational data to RDF data blocks via a first R2RML mapping, and store them as corresponding RDF data blocks in the storage area.

[0037] Its features are,

[0038] o Provide one or more processors, said processors being configured to:

[0039] A second R2RML mapping is generated as a result of a quality check performed downstream of the first R2RML parser on the first R2RML mapping by the R2RML unit; and

[0040] The system provides at least one additional interface through which the second R2RML mapping can be transmitted, such that the energy efficiency of preparing relational data is optimized relative to the energy efficiency without generating the second R2RML mapping, because the second R2RML mapping automatically stops processing unresolved relational data when preparing relational data to form RDF data blocks, thereby optimizing the energy efficiency of the preparation.

[0041] Furthermore, the subject of this invention is a computer-aided method for transforming data from a relational database containing sensor measurements into RDF data blocks of a graph database, the method comprising the following steps:

[0042] - Provides relational data from a relational database;

[0043] - Provide the first R2RML mapping file;

[0044] - The relational data is decomposed and transformed using the first R2RML mapping parser and the first R2RML mapping file;

[0045] - Generate at least one generation of RDF data blocks; and

[0046] - Store this generation of RDF data blocks as a data pool and / or RDF file for the graph database.

[0047] Its features are,

[0048] - After the data is decomposed and transformed by the first R2RML parser, the quality of the obtained R2RML mapping is checked and improved, and at least one second R2RML mapping file is created. Based on the second R2RML mapping file, the relational data is decomposed and transformed into RDF data blocks. The second R2RML mapping automatically stops processing unparseable relational data when preparing relational data to form RDF data blocks, thereby optimizing the energy efficiency of the preparation.

[0049] The computer program according to the present invention runs in a processor that executes the method. The computer program product according to the present invention provides a machine-readable data carrier on which a computer program is stored, and when the computer program runs in a processor, the method is executed by the computer program.

[0050] This method prepares the data into RDF data in the first R2RML mapping step of the prior art. Following this first step—as first proposed in this invention—a check and / or optimization of the intermediate result is performed in the form of a "second R2RML mapping." Here, the check determines which nodes and / or paths actually need to be transformed because they can be converted into graph data, and for which nodes and / or paths are scattered because they are not connected to other paths and / or nodes, or cannot be reasonably connected, further transformation is unreasonable. Using the technique according to this invention, the time, cost, and overhead of these "unproductive" R2RML mapping transformations can be saved.

[0051] For example, the present invention relates to an apparatus for preparing relational data to form data blocks as a data pool for a graph database and / or to form RDF data blocks, the apparatus comprising one or more processors configured to:

[0052] Preprocessing is performed on the R2RML mapping of relational data such as sensor data;

[0053] ο Describe cached data of one or more R2RML mappings using an ontology for visualization.

[0054] Furthermore, the processor has:

[0055] one or more R2RML modules, the R2RML modules

[0056] The system includes one or more R2RML parsers and a storage area, wherein the R2RML parsers and the storage area are adapted to receive relational data, decompose the relational data, and convert the relational data into R2RML mappings;

[0057] ο and storage units used to store the generated data blocks;

[0058] one or more display devices for optically reproducing the generated first and / or second R2RML mappings, wherein the device further includes

[0059] one or more storage areas (8) for storing data in a graph database; and

[0060] Related interfaces (3, 7, 10, 12, and 15),

[0061] in

[0062] With the aid of an R2RML mapping parser, the R2RML module of the device is adapted to: allocate received relational data to RDF data blocks according to R2RML mapping rules and store them as corresponding RDF data blocks in a storage unit, particularly in such a way that one or more processors of the device are also configured to generate a second R2RML mapping file as a result of a quality check performed downstream of the first R2RML parser by the R2RML unit, and

[0063] ο Provides another interface through which a second R2RML mapping file can be transferred to an R2RML module within a preprocessing framework and / or transferred to one or more R2RML units within that R2RML module via another interface.

[0064] According to an advantageous embodiment of the device, the R2RML module includes at least one additional R2RML unit that initiates an optimized transformation method for data mapping derived from a second R2RML mapping file (14), particularly for generating and storing optimized and / or multi-generational RDF data blocks.

[0065] According to another advantageous implementation of this method, a depiction corresponding to the ontology is provided from a second R2RML mapping file, see [link to relevant documentation]. Figures 6 to 9 .

[0066] According to an advantageous embodiment of the invention, the method is a computer-implemented method for error analysis via one or more R2RML units, performing quality checks and / or quality improvement when transforming data from a relational database containing columns and rows of sensor data into data from a graph database containing relations via an R2RML method, wherein the R2RML method, if necessary, visualizes intermediate results in the form of mappings. The computer-implemented method comprises at least the following method steps:

[0067] - Check if each column in the relational database has a corresponding node in the graph database;

[0068] - Check if each row in the relational database has a path in the graph database.

[0069] The resulting graph, in the form of a second R2RML mapping file, addresses the technical issues of erroneous transformations and / or assignments from relational database metadata to graph databases.

[0070] An “R2RML unit” refers to a computer-aided element that automatically transforms and / or maps relational data using R2RML mapping rules. An R2RML unit receives data via one or more interfaces, processes this data in a computer-aided manner while incorporating R2RML rules, and forwards the results via one or more other interfaces. An R2RML unit (11) can be part of data preprocessing, part of an R2RML module, or part of a visualization of R2RML mappings.

[0071] In conjunction with this invention, "processor" can be understood, for example, as a machine or electronic circuit. A processor can be, in particular, a central processing unit (CPU), a microprocessor, or a microcontroller, such as an application-specific integrated circuit (ASIC) or a digital signal processor, which may be combined with a memory unit for storing program instructions. A processor can also be, for example, an integrated circuit (IC), especially a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), a digital signal processor (DSP), or a graphics processing unit (GPU). A processor can also be understood as a virtual processor, a virtual machine, or a soft CPU. For example, a programmable processor may also be involved, equipped with configuration steps for implementing the methods according to the invention, or a programmable processor configured using configuration steps to implement the features of the methods or modules or other aspects and / or sub-aspects of the invention according to the invention.

[0072] In conjunction with this invention, a "module" can be understood, for example, as a processor and / or a storage unit for storing program code. For example, a processor is specifically configured to execute program code so that it performs functions to implement the method or steps of the method according to this invention. Corresponding modules can also be constructed as separate or independent modules. For this purpose, corresponding modules may include other elements, for example. These elements are, for example, one or more interfaces (e.g., database interfaces, communication interfaces—e.g., network interfaces, WLAN interfaces) and / or evaluation units (e.g., processors) and / or storage units. By means of these interfaces, for example, data can be exchanged (e.g., received, transmitted, sent, or provided). By means of the evaluation unit, for example, data can be compared, examined, processed, allocated, or calculated in a computer-aided and / or automated manner. By means of the storage unit, for example, data can be stored, retrieved, or provided in a computer-aided and / or automated manner.

[0073] An ontology is a multi-language compiled and formally ordered representation of a set of terms within a specific subject area and the relationships between those terms. An ontology is a network of information with logical relationships. In recent years, ontology has emerged with the idea of ​​semantic extension and has thus become part of knowledge representation in the subfield of artificial intelligence.

[0074] In the current context, the "semantic web" refers to the extended World Wide Web, in which data can be more easily exchanged between computers and more easily used by computers or processors.

[0075] According to this invention, it is proposed in principle that mappings created based on automated transformations using R2RML should be interpreted as graphs themselves as much as possible. For example, graphs are created here that visualize the improved, second R2RML mapping file.

[0076] A "parser" is a computer program that can break down and transform data into a format more suitable for further processing.

[0077] "Pre-processing" refers to data cleaning, and in particular, it also refers to "data mining." This includes, for example, identifying missing or erroneous data combinations, such as removing impossible data combinations like: "Gender: Male; Pregnant: Yes".

[0078] "Data mining" refers to the systematic application of computer-aided methods to find patterns, trends, or relationships in an existing database.

[0079] In conjunction with this invention, "computer-aided" or "computer-supported" can be understood, for example, as an implementation of the method, wherein, in particular, the processor implements at least one method step of the method. For example, "computer-aided" or "computer-supported" can also be understood as "computer implementation".

[0080] In conjunction with this invention, "providing," particularly concerning data, metadata, and / or other information, can be understood, for example, as providing computer-aided information. This provision is achieved, for example, through an interface (e.g., a database interface, a network interface, an interface with a storage unit). For example, in the case of this provision, the interface can be used to transmit and / or send and / or invoke and / or receive corresponding data and / or information. In conjunction with this invention, "providing" can also be understood, for example, as loading or storing, e.g., loading or storing a transaction with corresponding data. "Providing" can also be understood, for example, as transferring (or sending or transmitting) corresponding data from one node to another.

[0081] exist Figure 4 China — based on Figure 3 The illustration of the prior art shows a scheme extended by the R2RML unit 13 according to the invention. In the pre-processing, i.e., pre-processing, which includes at least elements or method steps 10 to 14, according to this embodiment of the invention, based on... Figure 3The known element "2", i.e., the R2RML unit used for R2RML mapping, is used to create a second, checked, and improved R2RML unit 14 for R2RML mapping. For this purpose, the original unit of the first R2RML mapping file 2 is transmitted via interface 10 to an additional R2RML parser 11, which forwards its result via interface 12 to R2RML unit 13 for quality checking and improvement. This R2RML unit then creates the second R2RML mapping file 14, which is transmitted via interface 3 to R2RML module 4. According to this embodiment of the invention, as... Figure 3 As shown in the example, the elements of R2RML module 4 are presented unchanged from the prior art.

[0082] Figure 5 Another embodiment of the invention is shown, in which the preprocessing is not supplemented, but rather the elements of the conventional R2RML module 4 are supplemented. Of course, both embodiments of the invention, as... Figure 4 and 5 As shown, they can also be used in combination. Here, after the relational data including sensor data is decomposed and transformed by one or more R2RML parsers 5 and / or 11, the first R2RML file is checked and improved in the R2RML unit 13 for quality checking and / or quality improvement. The results are transmitted via interface 12 and / or 16, and the second, improved R2RML mapping file 14 is thus fed into the R2RML module 4 and / or can be transmitted to the R2RML unit (16) within the R2RML module 4.

[0083] exist Figure 5 Instead of using an additional R2RML parser, R2RML parser 5 or 11 transmits the transformation result according to the first R2RML mapping to R2RML unit 13 via interface 12 for quality checking and improvement. Figure 4 In the illustrated embodiment, the R2RML unit generates a second R2RML mapping 14, which controls another R2RML unit 16 via interface 15. In R2RML unit 16, optimization of the R2RML mapping begins, by comparing the received results with calculated data.

[0084] - Initiate parallelism and / or distribution of ongoing assignments;

[0085] - Perform clustering of paths / nodes;

[0086] - To identify isolated and / or unconnected nodes, and may, for example, cause the deletion of these nodes and / or the removal of unused classes / groups from the ontology.

[0087] The particular advantages of the data transformation disclosed herein, performed using the improved second R2RML mapping file 14, are especially:

[0088] Mapping transformations can be performed in parallel or in a distributed manner. This not only improves performance but also enhances scalability.

[0089] - It can implement a plan for the execution of assignments by identifying unconnected nodes relative to the assignments between nodes in the R2RML map.

[0090] - Enables efficient verification of quality during the mapping phase itself, not just after data / knowledge graph generation.

[0091] - It becomes possible to identify and distinguish various types of differences in semantic mapping.

[0092] Finally, it can also identify and distinguish various types of differences in ontology-based mappings.

[0093] With the aid of the second mapping file 14, the R2RML unit 13, used for quality checking and improvement of the R2RML mapping, automatically performs the following method steps:

[0094] - Adjust the size of the node to fit the number of paths connected to that node;

[0095] - For example, using the Louvain method for optimizing modularity, communities are identified, i.e., groups of connected nodes.

[0096] According to an advantageous embodiment of the invention, a second R2RML mapping file (14) is represented on a display device (such as an imaging device, like a monitor) by an R2RML unit 9 having a suitable body. This representation corresponds to... Figure 6 and 9 .

[0097] Here, the representation of a node that can be filled or empty is adapted to the number of paths connected to the node, such that a node from which multiple paths originate is larger than a node connected to only a few paths.

[0098] Furthermore, the communities identified, including both nodes and paths, were represented by the same color.

[0099] Thus, instead of the currently common representation of R2RML mapping files, such as Figure 2As shown, clustering and / or identifying these communities can generate results such as... Figure 6 The representation of the second R2RML mapping file 14 shown.

[0100] Figure 7 It shows Figure 6 The document shows labeled and magnified details. Here, we can see how these clusters and / or communities can be identified more clearly using the method first presented here for generating the second R2RML mapping file.

[0101] exist Figure 7 The representation shown in the figure provides some useful insights into mapping, and the most interesting of these insights are listed below.

[0102] 1. Dependencies within and between mapping rules become apparent.

[0103] 2. Different clusters / communities involve separate assignment rules—corresponding to one TriplesMap, or a set of related assignment rules—corresponding to multiple TriplesMaps.

[0104] According to the present invention, such as Figures 4 to 8 As shown, the R2RML unit 13 for quality inspection and / or quality improvement and the second R2RML mapping file 14 generated therefrom can optimize the R2RML allocation process in particular in different ways:

[0105] 1. Especially when performing and / or generating mappings in the R2RML mapping resolver:

[0106] - Automating allocation creation via R2RML is a capacity-intensive process. According to the present invention... Figure 8 The embodiments shown in the figure use clusters / communities identified in the allocation graph as blocks that are executed in parallel or sequentially, based on the amount of data and allocations, thereby improving performance and scalability when creating mappings in the processor.

[0107] -B) Because dependencies within and between allocation rules (mapping rules) are represented by—compared to, for example— Figure 2 In terms of representation—obviously—the assignment graph can be interpreted as a dependency graph or even a dependency acyclic graph, enabling the planned execution of assignment rules. For example, bounded blocks are all independent blocks that can be executed in any order. After all these blocks have been completed, the dependent blocks can be assigned—shown as large circles with IDs ending with "12879TriplesMap"—see [reference needed]. Figure 8 .

[0108] 2. In the case of improving the quality of these allocations:

[0109] According to the R2RML procedure in the R2RML map resolver, each assignment rule is required to contain at least one TriplesMap (subject, predicate, object), which in turn contains references to a logical table, a subject map, and at least one predicate / object map. According to a preferred embodiment of the invention, this is checked via a SPARQL query. Therefore, assignments that do not meet these requirements can be easily identified.

[0110] - Because visualization can also be performed automatically within the processor without a display device, unconnected paths and / or nodes can be easily identified. These interruptions can have various causes:

[0111] Incomplete assignments, such as when the subject map, object map, and predicate map are not fully mapped. ) and / or object / predicate map ( Although defined, nodes are not executed, linked, and / or referenced during allocation. After these errors are identified, they can be corrected by adding references to the unconnected nodes using a SPARQL update query. This can be done automatically or manually through visualization of the erroneous mappings; see [link to relevant documentation]. Figure 8 .

[0112] οB) Unused R2RML blocks: When the subject map, object map, and predicate map are unused. ) and / or object / predicate map ( This occurs when an R2RML block, though defined, is no longer needed, i.e., redundant. This can happen, for example, in legacy assignments where multiple changes have been made but not consistently updated. These errors can be identified—again, through the corresponding SPARQL update query—and eliminated by removing the R2RML mapping block from the second R2RML mapping file.

[0113] 3. In the case of quality management of R2RML mapping files related to the ontology:

[0114] During the preparation process using the R2RML mapping resolver, according to an advantageous embodiment of the invention, an ontology is applied, wherein, for example, the specification of "OWL - Web Ontology Language"—the World Wide Web Consortium—may be applied. This allows for the possibility of obtaining additional knowledge from the second R2RML mapping file and / or also provides further quality checks.

[0115] After applying the ontology, it is additionally suggested, for example, that the same methodological steps be performed as when looking up clusters and / or communities—e.g., via the Louvain method. If the assignment is consistent and compatible with the ontology, it can be assumed that the following premises are met: for example, the R2RML predicates of the mapping should correspond to valid data features and / or object features in the ontology, and the R2RML classes in the mapping should correspond to valid classes in the ontology.

[0116] Here, disconnected nodes and / or paths may indicate the following:

[0117] - Ontology extensions may be required, such as the ontology being extended and / or modified, especially by adding and / or renaming class and / or data / object attributes based on incorrect or missing references in the mapping.

[0118] - Ontology updates may be required. For example, disconnected nodes may indicate the removal of obsolete classes and / or properties, ontological concepts (i.e., classes, data, and / or object properties) from the ontology, which would make the ontology up-to-date. Therefore, classes, data, and / or object properties not referenced in these assignments may be valid and necessary—i.e., up-to-date—or may be redundant concepts—i.e., obsolete.

[0119] These relationships can also be particularly evident through [their] sources Figure 4 and 5 The processor 13 is invoked to process the results of the R2RML mapping resolver, thereby creating a second R2RML mapping file 14 according to the present invention, which can be used for further allocation—see [link to invention]. Figure 4 and 5 If the processor used in this invention for quality inspection and / or quality improvement is located within the R2RML module—see [link to R2RML module]. Figure 5 In other words, these results can be used automatically to optimize the method, namely, the parallelism of independently running cluster lookups and the removal of unconnected nodes and / or paths after the corresponding SPRQL update query.

[0120] "Mapping" corresponds to "assignment". Correspondingly, R2RML mapping is also R2RML assignment. An R2RML mapping file is an R2RML assignment map, such as in... Figure 2 , 6 Visualized in 7 (details only), 8, and 9, i.e., shown.

[0121] according to Figure 5 The embodiment shown optimizes the mapping file (Mapping Data), and in this embodiment, an R2RML module is used, which includes a processor 16 for optimization based on the second mapping file 14.

[0122] For optimizing the entire R2RML transformation process using processor 16 for optimization, it is not necessary to visualize these relationships.

[0123] The R2RML method illustrated herein extracts information from the representation and / or data of the R2RML mapping method. This information allows for error analysis of the automated R2RML allocation from relational data to RDF data blocks, and, if necessary, the creation of at least one optimization possibility for the automated allocation from relational data to RDF data blocks later. Optionally, a presentation, visualization, and / or representation of the R2RML mapping, referred to herein as a second R2RML mapping file 14, is also provided, offering the possibility of measuring the integrity and / or quality of the R2RML mapping that has occurred.

[0124] In particular, evidence can be provided by identification information—for example, in the second R2RML mapping file (14)—that indicates that the corresponding data block was inserted by a specific node or indicates the source of the data block.

[0125] Relational data can be, for example, a database of devices in technical systems and / or industrial facilities and / or automated networks and / or production facilities. These devices can be, for example, field devices or devices in the Internet of Things (IoT).

[0126] The output and / or display device may optionally be a component of currently disclosed apparatus for preparing relational data to form data blocks as a data pool for a graph database and / or to form RDF data blocks, and the output and / or display device is particularly suitable for displaying R2RML triple maps with nodes and paths that represent R2RML triple links in a graph manner.

[0127] Here, display devices specifically include imaging components, such as screens, monitors, and so on.

[0128] In particular, the number of paths for each node can be identified by the node's size and the path's common attributes, and can be made visible through color when necessary during visualization.

[0129] -A) It can automatically—in processor 13—identify which paths “enter the void,” i.e., have errors, and / or

[0130] -B) can automatically—also in processor 13—compare whether all data received from all relations has also been allocated to at least one RDF data block, here,

[0131] - Provide at least one communication interface configured to receive results from A) and / or B), and thereby forward the second R2RML mapping file 14 during preprocessing or within the R2RML transformation method to another processor 16 within the R2RML module, which processor

[0132] - Compare the received results with the calculated data.

[0133] - Can enable parallelism in ongoing allocations,

[0134] - It can cluster paths / nodes.

[0135] - It can identify isolated and / or unconnected nodes, and, for example, cause the deletion of these nodes and / or the removal of unused classes / groups from the ontology. On the other hand, the identification can also be used to correspondingly extend, update, and / or modify the ontology.

[0136] According to the present invention, an implementation of at least one R2RML unit and, if necessary, other R2RML units is proposed, wherein identifying and improving the quality of the R2RML mapping file can initiate the optimization of the R2RML mapping process, and, if necessary, visualize the R2RML mapping after quality improvement and, if necessary, after optimizing the R2RML method.

[0137] This invention typically achieves energy savings in the transformation system by avoiding some R2RML mappings for erroneous and / or non-extending nodes and / or paths through an optimized second R2RML mapping. Since—according to this invention—in principle less data is analyzed, because this data is deleted and / or removed through the second mapping, and because the second mapping eliminates redundant paths and / or nodes, the transformation method becomes faster overall, thereby saving more time, energy, and cost. The magnitude of savings or optimization varies. For some data, the second mapping deletes a lot; for others, it doesn't delete much, depending on the form of the received relational data assigned in the first parser R2RML mapping step.

[0138] The method and R2RML module according to the present invention can improve the energy efficiency of automated transformation from relational data to machine-readable data in RDF format. This transformation is required whenever first-generation data—whether "freshly" generated by monitoring equipment such as sensors and analytical devices in an existing production chain or data generated in outdated formats from "older" equipment with long operating histories—for example, in the case of gas turbines and / or energy distribution units with operating histories of 30 years and longer—is transformed and thus can be further automated and, if necessary, controlled and regulated.

[0139] List of reference numerals

[0140] 1. Relational Databases

[0141] 2. R2RML unit with R2RML mapping file

[0142] 3. Interface with R2RML module

[0143] 4 R2RML modules

[0144] 5. R2RML Mapping Parser of the R2RML Module

[0145] 6 RDF generation memory

[0146] 7. The end interface of the R2RML module

[0147] 8. A data pool or storage for RDF files used in graph databases.

[0148] 9. R2RML units with ontology for making mappings visible.

[0149] 10. Second interface with the preprocessed R2RML parser

[0150] 11. Preprocessed R2RML mapping parser

[0151] 12. A third interface with the quality inspector, which is part of the preprocessing or located within the R2RML module.

[0152] 13 R2RML units used for quality checking and / or quality improvement of R2RML mappings

[0153] 14 Improved second R2RML mapping file

[0154] 15. The fourth interface within the R2RML module

[0155] 16 Optimized R2RML Parser

[0156] 17. Memory 1, 2, ... n for RDF generation 1, RDF generation 2... RDF generation n

Claims

1. An R2RML module (4) for preparing relational data to form RDF data blocks, the R2RML module having: At least one interface connected to at least one storage area used to store the generated RDF data blocks; At least one interface connected to a preprocessing device for performing preprocessing; At least one interface for connecting to a display device; At least one interface for connecting to the parser; as well as One or more processors are configured to generate at least one generation of RDF data blocks; and One or more R2RML parsers, the R2RML parsers being adapted to receive relational data, decompose the relational data, and convert the relational data into a first R2RML mapping; At least one storage area for storing the generated RDF data blocks. in With the aid of a first R2RML parser, the R2RML module (4) is adapted to allocate the received relational data to RDF data blocks using R2RML mapping rules, and store the relational data as corresponding RDF data blocks in the storage area. Its features are, Provide one or more processors, said processors being configured to: A second R2RML mapping file is generated using the second R2RML mapping, as a result of the quality check performed downstream of the first R2RML parser by the R2RML unit of the R2RML module (4); and Provide at least one additional interface through which the second R2RML mapping file can be transmitted. The second R2RML mapping automatically stops processing unparseable relational data when preparing the relational data to form RDF data blocks, thereby optimizing the energy efficiency of the preparation.

2. The R2RML module (4) according to claim 1, characterized in that, An interface is provided so that the R2RML unit within the R2RML module (4) can receive the second R2RML mapping file via the interface.

3. The R2RML module (4) according to claim 2, wherein, The R2RML unit automatically begins the optimization process after obtaining the second R2RML mapping file.

4. The R2RML module (4) according to any one of claims 1 to 3, characterized in that, The R2RML unit is used to optimize the mapping process based on the second R2RML mapping file.

5. The R2RML module (4) according to claim 4, characterized in that, The optimization process in the R2RML module (4) triggers the generation of multi-generation RDF data blocks, which are stored in corresponding storage areas.

6. The R2RML module (4) according to claim 4, characterized in that, The optimization process in the R2RML module (4) triggers the parallel and simultaneous generation of multiple generations of RDF data blocks, which are stored in corresponding storage areas.

7. The R2RML module (4) according to any one of claims 1 to 3, wherein the R2RML module is connected to a processor via an interface, the processor being adapted to preprocess the R2RML mapping of relational data.

8. The R2RML module (4) according to claim 7, wherein, The relationship data is sensor data.

9. The R2RML module (4) according to any one of claims 1 to 3, wherein the R2RML module is connected to a processor adapted to depict cached data of one or more first and / or second R2RML maps through an ontology for visualization.

10. The R2RML module (4) according to any one of claims 1 to 3, wherein, An ontology is provided in the R2RML unit for comparing the ontology with the second R2RML mapping file.

11. The R2RML module (4) according to claim 9, characterized in that, An output device is connected to the R2RML unit, and the output device uses the body to visualize the comparison results of the second R2RML mapping file.

12. A computer-aided method for transforming data from a relational database containing sensor measurements into RDF data blocks of a graph database, the method comprising the following steps: - Provides relational data from a relational database; - Provide the first R2RML mapping file; - The relational data is decomposed and transformed using the first R2RML mapping parser and the first R2RML mapping file; - Generate at least one generation of RDF data blocks; and - Store this generation of RDF data blocks as a data pool and / or RDF file for the graph database. Its features are, After the data is decomposed and transformed by the first R2RML parser, the quality of the obtained R2RML mapping is checked and improved, and at least one second R2RML mapping file is created by the second R2RML mapping. Based on the second R2RML mapping file, the relational data is decomposed and transformed into RDF data blocks. The second R2RML mapping automatically stops processing unparseable relational data when preparing the relational data to form RDF data blocks, thereby optimizing the energy efficiency of the preparation.

13. The method according to claim 12, characterized in that, The second R2RML mapping file has already been generated during preprocessing.

14. The method according to claim 12, characterized in that, The second R2RML mapping file is generated within the R2RML module (4).

15. The method according to any one of claims 12 to 14, characterized in that, An optimized transformation method for data mapping is initiated using an additional R2RML unit.

16. The method according to claim 15, characterized in that, Multigenerational RDF data blocks are generated using the optimized transformation method of the R2RML mapping.

17. The method according to claim 16, characterized in that, Within the method, multiple generations of RDF data blocks are generated simultaneously, i.e. in parallel, through the optimized transformation method of the R2RML mapping.

18. The method according to any one of claims 12 to 14, characterized in that, The result of at least one R2RML mapping resolver is passed to another R2RML unit for comparison with the ontology and for visualization.

19. A computer program product having program code stored on a machine-readable medium for performing the method according to any one of claims 12 to 18 when the program code is executed on a computer.

Images