Information processing device, information processing method, and information processing program

Abstract

A morphological analysis unit (102) analyzes a plurality of pieces of description information each describing one or more natural-language words, and extracts a word described in each piece of description information from the pieces of description information by identifying the description position of the word in the pieces of description information. A label hierarchization processing unit (103) generates a hierarchical structure between words extracted from neighboring description positions.

Description

Technical Field

[0005] , ,

[0001] The present disclosure relates to a technique for generating a hierarchical structure among words included in descriptive information.

Background Art

[0002] In a system for collecting and storing sensor data (such as power consumption) obtained in a home, factory, building, etc., information regarding sensors is managed. The information to be managed includes static information regarding sensors and dynamic time-series data measured by the sensors. Generally, the static information and the time-series data are stored and managed in separate tables. Here, the static information regarding sensors includes the installation location of the sensors and an explanation of the measurement target of the sensors. Also, the measurement target of the sensors is often managed in a hierarchical structure.

[0003] For example, in Patent Document 1, a method for hierarchically managing sensors is shown. Also, in Patent Document 2, a method for dividing label information given to a sensor into an explanation of the installation location of the sensor and the measurement target by morphological analysis is shown.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] When collecting and storing sensor data (such as power consumption) measured in homes, factories, buildings, etc., the data is saved in a sensor list table and a sensor data storage table. The sensor list table takes the format of [sensor ID, label information]. The sensor data storage table takes the format of [sensor ID, measurement date and time, measured value]. ID stands for Identifier. The label information assigned to each sensor ID is text-based label information assigned by humans for sensor management purposes. Therefore, programs find it difficult to handle this label information. Furthermore, the label information has a flat data structure. On the other hand, the sensors measure things like rooms and electrical equipment, which are arranged hierarchically. Converting flat label information into hierarchical label information allows for the representation of information closer to that of the real environment. Therefore, converting label information into hierarchical label information facilitates data aggregation and analysis.

[0006] Patent Document 1 describes a method for managing sensors in a hierarchical structure. However, Patent Document 1 does not describe a method for generating a hierarchical structure from label information. Patent Document 2 proposes a method for splitting label information using label information and a word dictionary. However, it does not show a method for generating a hierarchical structure from the split label information.

[0007] The primary purpose of this disclosure is to generate a hierarchical structure from descriptive information that describes natural language words, such as label information. [Means for solving the problem]

[0008] The information processing device related to this disclosure is Analyze multiple descriptive information, each containing one or more natural language words, and then analyze each descriptive information... Furthermore, a word extraction unit extracts the words described in each descriptive information by identifying the position of the words in each descriptive information, It includes a hierarchical structure generation unit that generates a hierarchical structure between words extracted from adjacent description positions. [Effects of the Invention]

[0009] According to this disclosure, a hierarchical structure can be generated from descriptive information that describes natural language words, such as label information. [Brief explanation of the drawing]

[0010] [Figure 1] A diagram showing an example of the functional configuration of the information processing device according to Embodiment 1. [Figure 2] A diagram showing an example of a sensor list table according to Embodiment 1. [Figure 3] A diagram showing an example of a sensor data storage table according to Embodiment 1. [Figure 4] A diagram illustrating the morphological analysis process according to Embodiment 1. [Figure 5] A diagram illustrating the label hierarchical processing according to Embodiment 1. [Figure 6] A diagram showing an example of operation in the morphological analysis process and label hierarchical processing according to Embodiment 1. [Figure 7] A diagram showing an example of a hierarchical label information database according to Embodiment 1. [Figure 8] A diagram showing an example of the hardware configuration of the information processing device according to Embodiment 1. [Figure 9] A diagram showing an example of the functional configuration of the information processing device according to Embodiment 2. [Figure 10] A diagram illustrating the label hierarchy processing according to Embodiment 2. [Figure 11] A diagram showing an example of a hierarchical label information database according to Embodiment 2. [Modes for carrying out the invention]

[0011] The embodiments will be described below with reference to the drawings. In the following description of the embodiments and in the drawings, the same reference numerals indicate the same part or a corresponding part.

[0012] Embodiment 1. ***Description of the configuration*** FIG. 1 shows a functional configuration example of the information processing apparatus 100 according to the present embodiment. Also, FIG. 8 shows a hardware configuration example of the information processing apparatus 100 according to the present embodiment. The information processing apparatus 100 according to the present embodiment is a computer. The operation procedure of the information processing apparatus 100 corresponds to an information processing method. Also, a program for realizing the operation of the information processing apparatus 100 corresponds to an information processing program. First, referring to FIG. 8, a hardware configuration example of the information processing apparatus 100 will be described.

[0013] As hardware, the information processing apparatus 100 includes a processor 901, a main memory device 902, an auxiliary storage device 903, and a communication device 904. The functions of the morphological analysis unit 102, the label hierarchical processing unit 103, and the sensor data analysis and aggregation processing unit 106 shown in FIG. 1 are realized by, for example, a program, and the program is stored in the auxiliary storage device 903. Hereinafter, the morphological analysis unit 102, the label hierarchical processing unit 103, and the sensor data analysis and aggregation processing unit 106 are referred to as functional components. The program for realizing the functions of the functional components is loaded from the auxiliary storage device 903 to the main memory device 902. Then, the processor 901 executes the program to perform the operations of the functional components in FIG. 1. FIG. 8 schematically shows a state in which the processor 901 is executing a program for realizing the functions of the functional components in FIG. 1.

[0014] Next, referring to FIG. 1, a functional configuration example of the information processing apparatus 100 will be described.

[0015] The sensor list table 101 stores static information about sensors (hereinafter referred to as sensor static information). The sensor list table 101 is, for example, a table in a CSV file. Alternatively, the sensor list table 101 may be a table in an XML file. Furthermore, the sensor list table 101 may be a table in a binary file. In addition, the sensor list table 101 may be a table in a relational database.

[0016] The morphological analysis unit 102 divides the label information contained in the sensor list table 101 into words through morphological analysis processing. The sensor list table 101 contains information for multiple labels. Each label contains one or more natural language words that represent the sensor's attributes (installation location, measurement target, etc.). The morphological analysis unit 102 analyzes each label information and extracts the words described in each label information by identifying the position of the words. Label information corresponds to descriptive information. Furthermore, the morphological analysis unit 102 corresponds to the word extraction unit. Also, the processing performed by the morphological analysis unit 102 corresponds to the word extraction process.

[0017] The label hierarchical processing unit 103 generates a hierarchical structure from the label information that has been divided into words. More specifically, the label hierarchical processing unit 103 generates a hierarchical structure between words extracted from adjacent descriptive positions by the morphological analysis unit 102. The label hierarchical processing unit 103 corresponds to the hierarchical structure generation unit. Furthermore, the processing performed by the label hierarchical processing unit 103 corresponds to the hierarchical structure generation process.

[0018] The hierarchical label information DB104 stores hierarchical label information. The hierarchical label information DB104 is, for example, a CSV file. Alternatively, the hierarchical label information DB104 may be an XML file or a binary file. Furthermore, the hierarchical label information DB104 may be a table in a relational database, a document in a document database, a key-value in a key-value database, or a node and edge in a graph database.

[0019] The sensor data storage table 105 stores time-series data of sensor measurements. The sensor data storage table 105 may be, for example, a table in a CSV file. Alternatively, the sensor data storage table 105 may be a table in an XML file. Furthermore, the sensor data storage table 105 may be a table in a binary file. In addition, the sensor data storage table 105 may be a table in a relational database.

[0020] The sensor data analysis and aggregation processing unit 106 uses the information from the hierarchical label information DB 104 and the sensor storage data table 105 to perform analysis and aggregation processing of sensor data. By performing analysis and aggregation processing of sensor data, the sensor data analysis and aggregation processing unit 106 provides a variety of services and solutions.

[0021] Figure 2 shows an example of the sensor list table 101.

[0022] The sensor list table 101 stores static information about the sensors. In the example shown in Figure 2, the sensor ID and label information are stored in the sensor list table 101 as static sensor information. Other information such as the sensor model number, installation date and time, installation location, size, and service life may also be stored. If the sensor list table 101 is stored in a relational database table format, the sensor ID will be the primary key. The sensor ID is an identifier used to uniquely identify a sensor. In the example in Figure 2, the sensor ID is a number, but the sensor ID may also be in text format.

[0023] Label information is text-based string information assigned by humans for the purpose of managing sensors. Label information includes natural language words describing the sensor's location and what it measures. In the example in Figure 2, the string (words) are separated by spaces. However, the string (words) do not need to be separated.

[0024] Figure 3 shows an example of the sensor data table 105.

[0025] The sensor data storage table 105 stores time-series data measured by the sensor. In the example shown in Figure 3, the sensor ID, measurement date and time, and measured value are stored. Other data such as measurement accuracy and measurement errors may also be stored. If the sensor data table 105 is stored in a relational database table format, the sensor ID and measurement date and time will be the composite primary key, and the sensor ID will be the foreign key. The sensor ID is an identifier used to uniquely identify a sensor. In the example in Figure 3, the sensor ID is a number, but the sensor ID may also be in text format. The sensor ID in the sensor data storage table 105 uses the same value as the sensor ID in the sensor list table 101. The measurement date and time is the date and time when the sensor performed the measurement. In addition to the date and time, time zone information may also be stored. The measured value is the value measured by the sensor. In the example in Figure 3, the measured value is numerical, but it can also be binary data or text data.

[0026] ***Explanation of operation*** Next, an overview of the operation of the information processing device 100 according to this embodiment will be described.

[0027] First, the user saves static sensor information to the sensor list table 101. Sensor measurements are saved as time-series data in the sensor storage data table 105. Prior to analyzing the sensor data, the user runs the morphological analysis unit 102 and the label hierarchical processing unit 103 to generate hierarchical label information. Subsequently, the sensor data analysis and aggregation processing unit 106 performs analysis and aggregation processing of the sensor data.

[0028] Figure 4 shows the morphological analysis process performed by the morphological analysis unit 102.

[0029] In step S41, the morphological analysis unit 102 determines whether morphological analysis processing has been performed on all label information included in the sensor list table 101. If morphological analysis processing has been performed on all label information (YES), the morphological analysis unit 102 terminates the morphological analysis processing. On the other hand, if there is label information that has not undergone morphological analysis (in the case of NO), the process proceeds to step S42. By repeating steps S41, S42, and S43, the morphological analysis unit 102 performs the processing in steps S42 and S43 for all label information.

[0030] In step S42, the morphological analysis unit 102 obtains unprocessed label information from the sensor list table 101.

[0031] In step S43, the morphological analysis unit 102 performs morphological analysis on the label information obtained in step S42 and divides the label information into meaningful words. The morphological analysis unit 102 can perform morphological analysis using known technologies (for example, open-source morphological analysis software such as MeCab). In step S43, the morphological analysis unit 102 identifies the description position of each word in the label information and divides the label information into words. In the label information for sensor ID: 53 in Figure 2, the morphological analysis unit 102 identifies that "instrument 1" is described first. The morphological analysis unit 102 also identifies that "(A circuit)" is described second. The morphological analysis unit 102 also identifies that "1L-1" is described third. The morphological analysis unit 102 also identifies that "air conditioner_W" is described fourth. After completing the process in step S43, the morphological analysis unit 102 marks the label information as having undergone morphological analysis and returns to step S41.

[0032] Figure 5 shows the label hierarchical processing performed by the label hierarchical processing unit 103.

[0033] In step S51, the label hierarchical processing unit 103 determines whether processing has been performed on all sequences of words that have been divided by the morphological analysis unit 102. If processing has been performed on all columns (YES), the process proceeds to step S53. On the other hand, if there are columns that have not been processed (NO), the process proceeds to step S52.

[0034] The words segmented by the morphological analysis unit 102 consist of multiple words per label information entry. For the purposes of this explanation, one label information entry will be represented as one row, and one segmented word as one column. In other words, for each label information entry, the first segmented word will be represented as the first column, the second segmented word as the second column, and so on. The label hierarchical processing unit 103 repeats steps S51 and S52, thereby executing the process in step S52 for all columns.

[0035] In step S52, the label hierarchical processing unit 103 performs a process to group duplicate words in the target column into identical words. More specifically, the label hierarchical processing unit 103 retains only the first occurrence of a word in the same column (a word extracted from the same description position), and removes any subsequent occurrences of the same word in the same column. After completing the processing in step S52, the label hierarchical processing unit 103 marks the target column as processed and returns to step S51.

[0036] In step S53, it is determined whether the hierarchical structure generation process (S54, S55, S56) has been performed for all columns. If the hierarchical structure generation process has been performed for all columns (YES), the label hierarchical processing unit 103 terminates the label hierarchical processing. On the other hand, if there are columns for which the hierarchical structure generation process has not been performed (i.e., NO), the process proceeds to step S54.

[0037] In step S54, the label hierarchical processing unit 103 compares the number of words in the nth column (A) with the number of words in the (n+1)th column (B). If the comparison result is "(A) < (B)", the process proceeds to step S55. On the other hand, if "(A) ≥ (B)", the process proceeds to step S56. Here, the nth column is a variable value ranging from the first column to "(maximum number of columns - 1)", where one label information item is represented as one row and one divided word as one column. The label hierarchical processing unit 103 changes the value of n by 1 from 1 to "(maximum number of columns - 1)". By repeating steps S53 to S56, the label hierarchical processing unit 103 executes the process in step S55 or step S56 for all columns.

[0038] In step S55, the label hierarchical processing unit 103 generates a hierarchical structure in which the word in the nth column is the parent and the word in the (n+1)th column is the child. If the words in the nth column and the (n+1)th column are extracted from the same label information, the label hierarchical processing unit 103 generates a hierarchical structure between the words in the nth column and the (n+1)th column extracted from the same label information. In other words, if a word exists in the nth column of the target row and a word also exists in the (n+1)th column, the label hierarchical processing unit 103 generates a hierarchical structure in which the word in the nth column is the parent and the word in the (n+1)th column is the child. On the other hand, if there is no word in the nth column of the target row, but there is a word in the (n+1)th column, the label hierarchical processing unit 103 searches for rows above the target row in the nth column. The label hierarchical processing unit 103 then generates a hierarchical structure in which the first word found is the parent and the word in the (n+1)th column is the child. In other words, the label hierarchical processing unit 103 generates a hierarchical structure between the word in the (n+1)th column and the word in the nth column of label information that is different from the label information from which the word in the (n+1)th column was extracted.

[0039] In step S56, unlike in step S55, the label hierarchical processing unit 103 does not generate a hierarchical structure. Columns that do not generate a hierarchical structure will be hierarchical with no children. Also, the (n+1)th column will be hierarchical with no parent.

[0040] Figure 6 shows an example of operation in the morphological analysis process and label hierarchical processing according to Embodiment 1.

[0041] Reference numeral 601 indicates the state before steps S41, S42, and S43 in Figure 4 are performed. In the example in Figure 6, two label entries exist in the sensor list table 101: "Instrument 1 (A circuit) 1L-1 Air conditioner_W" and "Instrument 1 (A circuit) 2L-1 Air conditioner_W".

[0042] Reference numeral 602 indicates the state after performing steps S41, S42, and S43 in Figure 4, and before performing steps S51 and S52 in Figure 5. In the example in Figure 6, "Instrument 1 (A circuit) 1L-1 Air conditioner_W" is divided into four words, "Instrument 1", "(A circuit)", "1L-1", and "Air conditioner_W", by morphological analysis processing (step S43). Similarly, "Instrument 1 (A circuit) 2L-1 Air conditioner_W" is divided into four words, "Instrument 1", "(A circuit)", "2L-1", and "Air conditioner_W", by morphological analysis processing (step S43).

[0043] Reference numeral 603 indicates the state after performing steps S51 and S52 in Figure 5, but before performing steps S53, S54, S55, and S56 in Figure 5. In the example in Figure 6, the first column of code 602 is processed in step S52. The first column contains "instrument 1" and "instrument 1", indicating that the same word appears multiple times. In this case, the first instance of the word, "instrument 1" in the first row, remains, while the subsequent instances of the word, "instrument 1" in the second row, are removed. The second column of code 602 is processed in the same way as the first column. For the third column of code 602, no processing is performed because the same word does not appear multiple times. The same processing as for the first column is performed on the fourth column of code 602.

[0044] Reference numeral 604 indicates the state after steps S53, S54, S55, and S56 in Figure 5 have been performed.

[0045] In the example in Figure 6, the determination in step S54 is first performed on the first column of reference numeral 603. The first column contains only one type of word, so (A)=1. The second column contains only one type of word, so (B)=1. Since "(A)≧(B)", the process proceeds to step S56. In step S56, no hierarchical structure is generated. Therefore, the first column is a hierarchy with no children. Also, since there are no parent nodes, the first column has a hierarchical structure with only one level, having neither parent nor children.

[0046] Next, the determination in step S54 is performed on the second column of code 603. The second column contains only one type of word, so (A)=1. The third column contains two types of words, so (B)=2. Since "(A)<(B)", the process proceeds to step S55. In step S55, a hierarchical structure is generated in which the word in the second column is the parent and the words in the third column are the children. In other words, the word in the second column, "(A circuit)", is the parent, and the two words in the third column, "1L-1" and "2L-1", are the children.

[0047] Next, the determination in step S54 is performed on the third column of code 603. The third column has two types of words, so (A)=2. The fourth column has only one type of word, so (B)=1. Since "(A)≧(B)", the process proceeds to step S56. In step S56, no hierarchical structure is generated. Therefore, the third column will be a hierarchy without children.

[0048] Since processing has been performed on all columns, the determination in step S53 becomes YES, and the process ends.

[0049] Figure 7 shows an example of a hierarchical label information DB104 according to Embodiment 1.

[0050] Figure 7 shows an example in which the hierarchical label information 604 shown in Figure 6 is represented as nodes and edges in a graph database and stored in the hierarchical label information DB 104. The example in Figure 7 shows a graph database, but the hierarchical structure can also be stored in the form of a relational database, document database, etc. Also, in the example in Figure 7, there is a distinction between the start and end points of the edges, but this distinction does not have to be made.

[0051] The word "instrument 1" in the first column has become a separate hierarchical unit due to the processing in step S56. Therefore, no word edges are generated for the word "instrument 1" in the first column. Regarding the relationship between the sensor nodes, an edge is generated between the "Instrument 1" node and the "Sensor ID: 53" node. Additionally, an edge is generated between the "Instrument 1" node and the "Sensor ID: 54" node.

[0052] As a result of the processing in step S55, a hierarchical structure is generated between the word "(A circuit)" in the second column and the words "1L-1" and "2L-1" in the third column, with the second column as the parent and the third column as the child. Therefore, an edge is generated between the node of "(A circuit)" and the node of "1L-1". Also, an edge is generated between the node of "(A circuit)" and the node of "2L-1". Additionally, regarding the relationship between the sensor nodes, an edge is generated between the node "1L-1" and the node "Sensor ID: 53". Also, an edge is generated between the node "2L-1" and the node "Sensor ID: 54".

[0053] The word "air conditioner_W" in the fourth column has become a separate hierarchical unit due to the processing in step S56. Therefore, no word edges are generated for the word "air conditioner_W" in the fourth column. Regarding the relationship between the sensor nodes, an edge is generated between the "Air Conditioner_W" node and the "Sensor ID: 53" node. Also, between the "Air Conditioner_W" node and "Sensor ID: An edge is created between node "54".

[0054] In this way, the hierarchical label information DB104 stores the relationship between the hierarchical label information data and the sensor ID.

[0055] ***Explanation of the effects of the embodiment*** As described above, this embodiment allows for the generation of a hierarchical structure between words from label information. This enables the representation of information closer to the real environment, facilitating data aggregation and analysis. As a result, it becomes possible to provide a variety of services and solutions.

[0056] In this embodiment, label information representing sensor attributes was used as an example. However, the information processing device 100 according to this embodiment is not limited to label information in which it can generate a hierarchical structure between words. As long as the descriptive information contains one or more natural language words, the information processing device 100 according to this embodiment can generate a hierarchical structure between words included in the descriptive information.

[0057] Note that the procedure described in this embodiment is just one example. Therefore, it is acceptable to perform only a portion of the procedure described in this embodiment. Furthermore, at least some of the procedures described in this embodiment may be combined with procedures not described in this embodiment. Furthermore, the configuration and procedures described in this embodiment may be modified as necessary.

[0058] Embodiment 2. In recent years, the use of generative AI based on large-scale language models (LLMs) for information aggregation / analysis or the construction of services / solutions has been widely considered. LLM stands for Large Language Model, and AI stands for Artificial Intelligence. On the other hand, because LLMs learn from publicly available information, they can answer questions about publicly available information, but they cannot answer questions that require internal company information or specific domain knowledge. Therefore, RAG has been proposed as a technology to improve the accuracy of answers by preparing knowledge that LLMs have not learned as external data and using it as a supplement. RAG stands for Retrieval-augmented generation. One type of RAG is GraphRAG, which handles graph-formatted databases as external data.

[0059] When using generative AI and GraphRAG for data analysis, it is necessary to prepare knowledge that the LLM has not yet learned as graph-formatted data beforehand. However, the process of converting knowledge into graph-formatted data is not easy.

[0060] This embodiment describes a method that enables the generative AI and GraphRAG to utilize the hierarchical structure as external knowledge.

[0061] This embodiment will primarily describe the differences from Embodiment 1. Matters not described below are the same as in Embodiment 1.

[0062] ***Explanation of the structure*** Figure 9 shows an example of the functional configuration of the information processing device 100 according to this embodiment. In Figure 9, compared to Figure 1, User 201, Generation AI+GraphRAG202, and LLM203 are added. In this embodiment, Generation AI+GraphRAG202 performs the operation of the sensor data analysis and aggregation processing unit 106 described in Embodiment 1.

[0063] User 201 uses the information processing device 100. User 201 inputs a natural language question for information analysis into the generating AI + GraphRAG202. User 201 then receives the analysis results from the generating AI + GraphRAG202.

[0064] The AI+GraphRAG202 generates and processes sensor data for analysis and aggregation. The generating AI+GraphRAG202 obtains the question text from user 201. Next, the generating AI+GraphRAG202 obtains the knowledge necessary for the analysis from the hierarchical label information DB104. In addition, the generating AI+GraphRAG202 obtains the data to be analyzed from the sensor storage data table 105. Then, the Generating AI+GraphRAG202 sends the question, the knowledge required for the analysis, and the data to be analyzed to the LLM203. After that, the Generating AI+GraphRAG202 receives the analysis results from the LLM203. Finally, the Generating AI+GraphRAG202 responds with the analysis results to the user 201.

[0065] In Figure 9, the sensor list table 101, morphological analysis unit 102, hierarchical label information DB 104, and sensor stored data table 105 are the same as those shown in Figure 1. Therefore, a detailed explanation of these is omitted. In this embodiment, the label hierarchical processing unit 103 generates hierarchical label information after the label hierarchical processing described in Embodiment 1. The hierarchical label information corresponds to hierarchical structure relationship information.

[0066] The information processing device 100 according to this embodiment can also be realized, for example, by the hardware configuration shown in Figure 8.

[0067] ***Explanation of operation*** Next, an overview of the operation of the information processing device 100 according to this embodiment will be described.

[0068] In this embodiment as well, user 201 saves static sensor information to the sensor list table 101. Sensor measurements are saved as time-series data in the sensor storage data table 105. Furthermore, prior to analyzing the sensor data, user 201 executes the morphological analysis unit 102 and the label hierarchical processing unit 103 to generate hierarchical label information. Subsequently, the AI+GraphRAG202 generates and processes the sensor data for analysis and aggregation.

[0069] In this embodiment as well, the morphological analysis unit 102 performs morphological analysis processing according to the procedure shown in Figure 4. The operation of the morphological analysis unit 102 in this embodiment is the same as in Embodiment 1. Therefore, a detailed explanation is omitted.

[0070] Figure 10 shows the label hierarchical processing by the label hierarchical processing unit 103 according to Embodiment 2.

[0071] Steps S51 to S56 are the same as those shown in Figure 5. Therefore, their explanation is omitted.

[0072] In step S95, the label hierarchical processing unit 103 generates hierarchical label information. More specifically, the label hierarchical processing unit 103 analyzes the hierarchical structure generated in step S55 and extracts one or more parent-child relationships between words in the hierarchical structure. Furthermore, the label hierarchical processing unit 103 determines the attributes of each parent-child relationship. The label hierarchical processing unit 103 then generates hierarchical label information that shows one or more parent-child relationships between words in the hierarchical structure along with the attributes of each parent-child relationship. Furthermore, the label hierarchical processing unit 103 sets a parent-child relationship between any word included in the hierarchical structure and the sensor ID. The label hierarchical processing unit 103 then uses at least one of the words that have a parent-child relationship with the sensor ID, and at least one of the words that have a parent-child relationship with the sensor ID, to determine the attribute of the parent-child relationship between the word included in the hierarchical structure and the sensor ID. The label hierarchical processing unit 103 then generates hierarchical label information that shows the parent-child relationship between the word included in the hierarchical structure and the sensor ID, along with the attribute of the parent-child relationship.

[0073] The label hierarchical processing unit 103 generates hierarchical label information, for example, using the following procedure. First, the label hierarchy processing unit 103 generates one node for each parent word. Then, the label hierarchy processing unit 103 sets the node with a type attribute representing the node's type and a name attribute representing the node's name. The label hierarchy processing unit 103 may also set other attributes related to the node. The label hierarchy processing unit 103 sets the value of the type attribute to the column name "Label Information" from the sensor list table 101. The label hierarchy processing unit 103 also sets the value of the name attribute to the word in the nth column. Next, the label hierarchy processing unit 103 similarly generates one node for each child word. Then, the label hierarchy processing unit 103 sets the type attribute, which represents the type of node, and the name attribute, which represents the name of the node, to that node. The label hierarchy processing unit 103 sets the value of the type attribute to the column name "Label Information" from the sensor list table 101. The label hierarchy processing unit 103 also sets the value of the name attribute to the word in the nth column.

[0074] Furthermore, the label hierarchical processing unit 103 also generates nodes for sensor IDs. The label hierarchical processing unit 103 then sets the type attribute, which represents the type of node, and the name attribute, which represents the name of the node, to the node. The label hierarchy processing unit 103 sets the value of the type attribute to the column name "Sensor ID" from the sensor list table 101. The label hierarchy processing unit 103 also sets the value of the name attribute to the "Sensor ID" as is. The sensor ID node generated in this way will be referred to as the sensor ID node below.

[0075] Next, the label hierarchical processing unit 103 generates an edge between the parent node and the child node. The edge direction can be either parent → child or child → parent. However, the label hierarchical processing unit 103 unifies the direction of all edges within the hierarchical label information. The label hierarchical processing unit 103 sets a type attribute on the edge that represents the type of edge. The label hierarchical processing unit 103 may also set other attributes related to the edge on the edge. The label hierarchy processing unit 103 sets the value of the type attribute to a value that represents a parent-child relationship, such as "parent" or "child".

[0076] Furthermore, the label hierarchical processing unit 103 sets a parent-child relationship between the words included in the hierarchical structure generated in step S55 and the sensor ID. Specifically, the label hierarchical processing unit 103 sets up a parent-child relationship between a type:label information node and a sensor ID node if the type:label information node does not have any other child type:label information nodes. In other words, the label hierarchical processing unit 103 adds an edge between a type:label information node that does not have any child type:label information nodes and a sensor ID node, thereby setting up a parent-child relationship between these nodes. The label hierarchical processing unit 103 treats the type:label information node as the parent node and the sensor ID node as the child node. The edge direction can be either from the label information node (parent) to the sensor ID node (child), or from the sensor ID node (child) to the label information node (parent). However, the label hierarchical processing unit 103 unifies the direction of all edges within the hierarchical label information. Furthermore, the label hierarchy processing unit 103 determines the attribute of the parent-child relationship between the label information node (parent node of the sensor ID) and the sensor ID using the words of the label information node (parent node of the sensor ID) that has a parent-child relationship with the sensor ID. Alternatively, the label hierarchy processing unit 103 may determine the attribute of the parent-child relationship between the label information node (parent node of the sensor ID) and the sensor ID using the words of the label information node (parent node of the sensor ID) that has a parent-child relationship with the sensor ID node and the parent node of the sensor ID's parent node, and the parent node of the sensor ID's parent node that has a parent-child relationship with the sensor ID. Alternatively, the label hierarchy processing unit 103 may determine the attribute of the parent-child relationship between the label information node (parent node of the sensor ID) and the sensor ID using the words of other nodes described in the same column number (nth column) in the sensor list table 101 as the parent node of the sensor ID or the parent node of the sensor ID's parent node. Furthermore, the label hierarchical processing unit 103 may use the words of any of the nodes described above to determine the parent-child relationship attribute using an external dictionary or LLM. In other words, the label hierarchical processing unit 103 may use the words of the parent node of the sensor ID to determine the parent-child relationship attribute using an external dictionary or LLM. Alternatively, the label hierarchical processing unit 103 may use the words of the parent node of the parent node of the sensor ID to determine the parent-child relationship attribute using an external dictionary or LLM. Furthermore, the label hierarchical processing unit 103 may use the words of other nodes described in the same column number (nth column) in the sensor list table 101 to determine the parent-child relationship attribute using an external dictionary or LLM. Then, the label hierarchical processing unit 103 adds a value (type attribute) corresponding to the determined attribute to the edge.

[0077] Figure 11 shows an example of a hierarchical label information DB104 according to Embodiment 2.

[0078] Figure 11 shows an example in which the hierarchical label information 604 shown in Figure 6 is represented as nodes and edges in a graph database and stored in the hierarchical label information DB 104. The example in Figure 11 shows a graph database, but the hierarchical structure can also be stored in the form of a relational database, document database, etc. Also, in the example in Figure 11, there is a distinction between the start and end points of the edges, but this distinction does not have to be made.

[0079] For the word "instrument 1" in the first column, the label hierarchical processing unit 103 generates a node with two attributes: "type: label information, name: instrument 1". Furthermore, the label hierarchical processing unit 103 generates a node with two attributes for the word "(A Circuit)" in the second column: "type: label information, name: (A Circuit)". Furthermore, the label hierarchical processing unit 103 generates a node with two attributes for the word "1L-1" in the third column: "type: label information, name: 1L-1". Similarly, the label hierarchical processing unit 103 generates a node with two attributes for the word "2L-1" in the third column: "type: label information, name: 2L-1". Furthermore, the label hierarchical processing unit 103 generates a node with two attributes for the fourth column word "Air Conditioner_W": "type: Label Information, name: Air Conditioner_W". Furthermore, the label hierarchical processing unit 103 generates a node for sensor ID: 53 that has two attributes: "type: sensor ID, name: 53". Similarly, the label hierarchical processing unit 103 generates a node for sensor ID: 54 that has two attributes: "type: sensor ID, name: 54".

[0080] The word "instrument 1" in the first column has become a standalone hierarchical unit due to the processing in step S56. Therefore, the label hierarchical processing unit 103 does not generate any word-inter-word edges for the word "instrument 1" in the first column. In other words, the node for the word "instrument 1" ("type:label information, name:instrument 1") does not have any child nodes of type:label information. Therefore, the label hierarchical processing unit 103 sets up a parent-child relationship between the node for the word "instrument 1" and each of the two sensor ID nodes ("type:sensor ID, name:53", "type:sensor ID, name:54"). Specifically, the label hierarchical processing unit 103 generates an edge between the node for the word "instrument 1" and each of the two sensor ID nodes. Here, the label hierarchical processing unit 103 uses the value "instrument1" of the name attribute of the word "instrument1", which is the parent node of the two sensor ID nodes, and uses an external dictionary or LLM to determine "instrument" as the attribute of the parent-child relationship. Then, the label hierarchical processing unit 103 sets the attribute value "instrument" corresponding to the determined parent-child relationship attribute to the type attribute value of each edge.

[0081] Furthermore, the label hierarchical processing unit 103 sets up a parent-child relationship between the node of the word "(A Circuit)" in the second column ("type: Label Information, name: (A Circuit)") and the node of the word "1L-1" in the third column ("type: Label Information, name: 1L-1"). Specifically, the label hierarchical processing unit 103 generates an edge between the two nodes. Similarly, the label hierarchical processing unit 103 sets up a parent-child relationship between the node of the word "(A circuit)" in the second column and the node of the word "2L-1" in the third column ("type: label information, name: 2L-1"). Specifically, the label hierarchical processing unit 103 generates an edge between the two nodes. Then, the label hierarchical processing unit 103 sets "type:child" as the attribute value for the parent-child relationship for each edge.

[0082] Furthermore, the node for the word "1L-1" does not have any child nodes with type:label information. Therefore, the label hierarchical processing unit 103 sets up a parent-child relationship between the node for the word "1L-1" and the sensor ID node ("type:sensor ID, name:53"). Specifically, the label hierarchical processing unit 103 generates an edge between the node for the word "1L-1" and the sensor ID node. Here, the label hierarchical processing unit 103 uses the value of the name attribute "(A Circuit)" of the word "(A Circuit)", which is the parent node of the word "1L-1", which is the parent node of the sensor ID node, and uses an external dictionary or LLM to determine "Circuit" as the attribute of the parent-child relationship. Then, the label hierarchical processing unit 103 sets the attribute value "Circuit", which corresponds to the determined parent-child relationship attribute, as the type attribute value of the edge.

[0083] Similarly, the node for the word "2L-1" does not have any child nodes with type:label information. Therefore, the label hierarchical processing unit 103 sets up a parent-child relationship between the node for the word "2L-1" and the sensor ID node ("type:sensor ID, name:54"). Specifically, the label hierarchical processing unit 103 generates an edge between the node for the word "2L-1" and the sensor ID node. Here, the label hierarchical processing unit 103 uses the value of the name attribute "(A Circuit)" of the word "(A Circuit)", which is the parent node of the word "2L-1", which is the parent node of the sensor ID node, and uses an external dictionary or LLM to determine "Circuit" as the attribute of the parent-child relationship. Then, the label hierarchical processing unit 103 sets the attribute value "Circuit", which corresponds to the determined parent-child relationship attribute, as the type attribute value of the edge.

[0084] The word "air conditioner_W" in the fourth column has become a separate hierarchical unit due to the processing in step S56. Therefore, the label hierarchical processing unit 103 does not generate an edge between words for the word "air conditioner_W" in the fourth column. In other words, the node for the word "air conditioner_W" ("type: label information, name: air conditioner_W") does not have any child nodes of type: label information. Therefore, the label hierarchical processing unit 103 sets up a parent-child relationship between the node for the word "air conditioner_W" and each of the two sensor ID nodes ("type: sensor ID, name: 53", "type: sensor ID, name: 54"). Specifically, the label hierarchical processing unit 103 generates an edge between the node for the word "air conditioner_W" and each of the two sensor ID nodes. Here, the label hierarchical processing unit 103 uses the value of the name attribute "Air Conditioner_W" of the parent node of the two sensor ID nodes, and uses an external dictionary or LLM to determine "Equipment" as the parent-child relationship attribute. Then, the label hierarchical processing unit 103 sets the attribute value "Equipment" corresponding to the determined parent-child relationship attribute to the type attribute value of each edge.

[0085] In this way, the label hierarchical processing unit 103 sets the relationship between the hierarchical label information data and the sensor information. The hierarchical label information DB 104 then stores the relationship between the hierarchical label information data and the sensor information.

[0086] ***Explanation of the effects of the embodiment*** According to this embodiment, the generative AI and GraphRAG can utilize the hierarchical structure as external knowledge. Furthermore, by utilizing generative AI that uses hierarchical structures as external knowledge and leveraging GraphRAG, more accurate data analysis and aggregation processing becomes possible. As a result, a variety of services and solutions can be provided.

[0087] Although Embodiments 1 and 2 have been described above, these two embodiments may be implemented in combination. Alternatively, one of these two embodiments may be partially implemented. Alternatively, these two embodiments may be partially combined and implemented. Furthermore, the configurations and procedures described in these two embodiments may be modified as needed.

[0088] ***Supplementary explanation of hardware configuration*** Here, we will provide a supplementary explanation of the hardware configuration of the information processing device 100. The processor 901 shown in Figure 8 is an integrated circuit (IC) that performs processing. Processor 901 includes components such as a CPU (Central Processing Unit) and a DSP (Digital Signal Processor). The main memory 902 shown in Figure 8 is RAM (Random Access Memory). The auxiliary storage device 903 shown in Figure 8 includes ROM (Read Only Memory), flash memory, HDD (Hard Disk Drive), etc. The communication device 904 shown in Figure 8 is an electronic circuit that performs data communication processing. The communication device 904 is, for example, a communication chip or a NIC (Network Interface Card).

[0089] Furthermore, the auxiliary storage device 903 also stores the OS (Operating System). Then, at least a portion of the OS is executed by processor 901. The processor 901 executes programs that realize the functions of the functional components shown in Figures 1 and 9, while executing at least a portion of the OS. Processor 901 executes the OS, handling task management, memory management, file management, communication control, and other functions. Furthermore, at least one of the information, data, signal values, and variable values ​​indicating the processing results of the functional components shown in Figures 1 and 9 is stored in at least one of the main memory 902, auxiliary memory 903, registers in the processor 901, and cache memory. Furthermore, the programs that realize the functions of the functional components shown in Figures 1 and 9 may be stored on portable recording media such as magnetic disks, flexible disks, optical disks, compact disks, Blu-ray® discs, and DVDs. Portable recording media containing the programs that realize the functions of the functional components shown in Figures 1 and 9 may also be distributed.

[0090] Furthermore, at least one of the "parts" in the functional components shown in Figures 1 and 9 may be read as "circuit," "process," "procedure," "process," or "circuitry." Furthermore, the information processing device 100 may be implemented by a processing circuit. Examples of processing circuits include logic ICs (Integrated Circuits), GAs (Gate Arrays), ASICs (Application Specific Integrated Circuits), and FPGAs (Field-Programmable Gate Arrays). In this case, the functional components shown in Figures 1 and 9 are each implemented as part of the processing circuit. In this specification, the higher-level concept encompassing both the processor and the processing circuit is referred to as "processing circuitry." In other words, a processor and a processing circuit are specific examples of "processing circuits," respectively.

[0091] Finally, the various aspects of this disclosure are summarized in the appendix. (Note 1) A word extraction unit analyzes multiple descriptive information, each containing one or more natural language words, and extracts the words described in each descriptive information by identifying the word's position in each descriptive information. An information processing device having a hierarchical structure generation unit that generates a hierarchical structure between words extracted from adjacent description positions. (Note 2) The hierarchical structure generation unit, The information processing device described in Appendix 1, which eliminates duplication among multiple words extracted from the same description location, and after the duplication has been eliminated, generates a hierarchical structure between words extracted from adjacent description locations. (Note 3) The hierarchical structure generation unit, The information processing device according to Appendix 1 or 2, which generates a hierarchical structure between the words extracted from the nth description position and the words extracted from the (n+1)th description position when the number of types of words extracted from the nth description position is greater than the number of words extracted from the nth description position. (Note 4) The hierarchical structure generation unit, The information processing device according to Appendix 3, which generates a hierarchical structure between the word extracted from the nth description position and the word extracted from the (n+1)th description position, with the word extracted from the nth description position as the parent and the word extracted from the (n+1)th description position as the child. (Note 5) The hierarchical structure generation unit, The information processing device according to Appendix 4, which generates a hierarchical structure between the word extracted from the nth description position and the word extracted from the (n+1)th description position, when there are both a word extracted from the same description information and a word extracted from the (n+1)th description position. (Note 6) The hierarchical structure generation unit, The information processing device according to Appendix 4, which generates a hierarchical structure between the word extracted from the (n+1)th description position and the word extracted from the nth description position of a description information different from the description information from which the word extracted from the (n+1)th description position was derived, when a word exists extracted from the (n+1)th description position for any of the description information but no word exists extracted from the nth description position. (Note 7) The word extraction unit, An information processing device described in any one of the appendices 1 to 6, which analyzes multiple descriptive information in which one or more natural language words representing the attributes of a sensor are written. (Note 8) The computer analyzes multiple descriptive information, each containing one or more natural language words, and extracts the words described in each descriptive information by identifying their positions within that information. An information processing method in which the computer generates a hierarchical structure between words extracted from adjacent descriptive positions. (Note 9) This process involves analyzing multiple descriptive information, each containing one or more natural language words, and extracting the words described in each descriptive information by identifying their positions within that information. An information processing program that causes a computer to perform a hierarchical structure generation process, which generates a hierarchical structure between words extracted from adjacent descriptive positions. (Note 10) The hierarchical structure generation unit, The information processing device according to Appendix 1, which analyzes the aforementioned hierarchical structure, extracts one or more parent-child relationships between words in the hierarchical structure, determines the attributes of each parent-child relationship, and generates hierarchical structure relationship information that shows one or more parent-child relationships between words in the hierarchical structure along with the attributes of each parent-child relationship. (Note 11) The word extraction unit, The sensor ID, which is the sensor's identifier, is analyzed by analyzing the multiple descriptive pieces of information written to each sensor. The hierarchical structure generation unit, The information processing device according to Appendix 10, which sets a parent-child relationship between any word included in the hierarchical structure and the sensor ID, determines the attribute of the parent-child relationship between any word included in the hierarchical structure and the sensor ID using at least one of the words that have a parent-child relationship with the sensor ID and the words that have a parent-child relationship with the sensor ID, and generates hierarchical structure relationship information that shows the parent-child relationship between any word included in the hierarchical structure and the sensor ID together with the attribute of the parent-child relationship between any word included in the hierarchical structure and the sensor ID. (Note 12) The hierarchical structure generation unit, An information processing device described in Appendix 10 that generates hierarchical relationship information in graph format. [Explanation of Symbols]

[0092] 100 Information processing unit, 101 Sensor list table, 102 Morphological analysis unit, 103 Label hierarchical processing unit, 104 Hierarchical label information DB, 105 Sensor stored data table, 106 Sensor data analysis and aggregation processing unit, 201 User, 202 Generation AI + GraphRAG, 203 LLM, 901 Processor, 902 Main memory, 903 Auxiliary memory, 904 Communication device.

Claims

1. A word extraction unit analyzes multiple descriptive information, each containing one or more natural language words, and extracts the words described in each descriptive information by identifying the word's position in each descriptive information. An information processing device having a hierarchical structure generation unit that, when the number of types of words extracted from the (n+1)th description position is greater than the number of words extracted from the nth description position, generates a hierarchical structure between the words extracted from the nth description position and the words extracted from the (n+1)th description position, and generates a hierarchical structure between words extracted from adjacent description positions.

2. The hierarchical structure generation unit, The information processing apparatus according to claim 1, which eliminates duplication in multiple words extracted from the same description position, and after the duplication has been eliminated, generates a hierarchical structure between words extracted from adjacent description positions.

3. The hierarchical structure generation unit, The information processing apparatus according to claim 1, wherein the word extracted from the nth description position is the parent and the word extracted from the (n+1)th description position is the child, and a hierarchical structure is generated between the word extracted from the nth description position and the word extracted from the (n+1)th description position.

4. The hierarchical structure generation unit, The information processing device according to claim 3, which generates a hierarchical structure between the word extracted from the nth description position and the word extracted from the (n+1)th description position, when there are words extracted from the same description information and words extracted from the (n+1)th description position.

5. The hierarchical structure generation unit, The information processing device according to claim 3, which generates a hierarchical structure between the word extracted from the (n+1)th description position and the word extracted from the nth description position of a description information different from the description information from which the word extracted from the (n+1)th description position was derived, when a word exists extracted from the (n+1)th description position for any of the description information but no word exists extracted from the nth description position.

6. The word extraction unit, The information processing apparatus according to claim 1, which analyzes a plurality of descriptive pieces of information in which each piece contains one or more natural language words that represent the attributes of a sensor.

7. The computer analyzes multiple descriptive information, each containing one or more natural language words, and extracts the words described in each descriptive information by identifying their positions within that information. An information processing method which, when the computer generates a hierarchical structure between the words extracted from the nth description position and the words extracted from the (n+1)th description position, and generates a hierarchical structure between words extracted from adjacent description positions, if the number of types of words extracted from the nth description position is greater than the number of words extracted from the nth description position.

8. This process involves analyzing multiple descriptive information, each containing one or more natural language words, and extracting the words described in each descriptive information by identifying their positions within that information. An information processing program that causes a computer to perform a hierarchical structure generation process, which generates a hierarchical structure between the words extracted from the nth description position and the words extracted from the (n+1)th description position, and generates a hierarchical structure between words extracted from adjacent description positions, when the number of types of words extracted from the nth description position is greater than the number of types of words extracted from the nth description position.

9. The hierarchical structure generation unit, The information processing apparatus according to claim 1, which analyzes the hierarchical structure, extracts one or more parent-child relationships between words in the hierarchical structure, determines the attributes of each parent-child relationship, and generates hierarchical structure relationship information that shows one or more parent-child relationships between words in the hierarchical structure along with the attributes of each parent-child relationship.

10. The word extraction unit, The sensor ID, which is the sensor's ID (identifier), is analyzed by analyzing the multiple descriptive pieces of information written to each sensor. The hierarchical structure generation unit, The information processing apparatus according to claim 9, which sets a parent-child relationship between any word included in the hierarchical structure and the sensor ID, determines the attribute of the parent-child relationship between any word included in the hierarchical structure and the sensor ID using at least one of the words that have a parent-child relationship with the sensor ID and the words that have a parent-child relationship with the sensor ID, and generates hierarchical structure relationship information that shows the parent-child relationship between any word included in the hierarchical structure and the sensor ID together with the attribute of the parent-child relationship between any word included in the hierarchical structure and the sensor ID.

11. The hierarchical structure generation unit, The information processing apparatus according to claim 9, which generates hierarchical relationship information in graph format.

Images