Analysis system and estimation system

Abstract

Provided is an analysis system capable of providing a selection item having a high possibility of being selected as a selection item to be given to a failure response record of a device. An analysis system for analyzing a failure response record including a free statement input by a person who performs a failure response of a device, the analysis system comprising: an extraction unit for extracting a free statement input by a person who performs a failure response of the device; a word extraction unit that extracts, from the free statements included in the fault response record, a specific target word and a word indicating the content of a treatment performed at the time of fault response; a degree-of-association calculation unit that calculates the degree of association between each of a plurality of candidate codes, which are candidates of the determination code given to the failure response record, and the target word extracted by the extraction unit; a degree of importance calculation unit that calculates the degree of importance of the treatment extracted by the extraction unit and associates the degree of importance with any one of the plurality of candidate codes; and a list unit that creates, on the basis of the degree of association calculated by the degree-of-association calculation unit and the degree of importance calculated by the degree-of-importance calculation unit, a selection list in which the plurality of candidate codes are arranged such that the higher the likelihood of being selected as a determination code, the higher the order is.

Description

Technical Field

[0001] This disclosure relates to analytical systems and estimation systems. Background Technology

[0002] Patent Document 1 discloses a maintenance record input assistance device as an analysis system. This analysis system assists maintenance personnel when inputting maintenance records for equipment maintenance. The analysis system extracts keywords from the freely described fault status information input by the maintenance personnel. Based on the extracted keywords, the analysis system determines the selection options corresponding to phenomena with a high probability of occurrence from the options selected for the maintenance record. The analysis system then presents the determined selection options to the maintenance personnel, thereby assisting them.

[0003] Existing technical documents

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2023-019574 Summary of the Invention

[0006] The problem the invention aims to solve

[0007] However, in the analysis system described in Patent Document 1, the selection option is determined solely based on the probability of association with the extracted keywords. In free-form statements, there are often words that, although extracted as keywords, have low relevance to the selected option. Therefore, the probability of the proposed option being selected is low.

[0008] This disclosure was made to solve the aforementioned problems. The purpose of this disclosure is to provide an analysis system and an estimation system capable of providing, or estimating, a highly probable option that can be selected as an option assigned to a fault response record of the device.

[0009] means for solving problems

[0010] The analysis system disclosed herein analyzes fault response records containing free statements input by a person who has performed fault response on the device. The analysis system comprises: an extraction unit that extracts specific object terms and words representing the actions taken during fault response from the free statements contained in the fault response records; a correlation calculation unit that calculates the correlation degree between each of a plurality of candidate codes that become candidates for judgment codes assigned to the fault response records and the object terms extracted by the extraction unit; an importance calculation unit that calculates the importance of the actions extracted by the extraction unit and correlates them with any one of the plurality of candidate codes; and a list unit that, based on the correlation degree calculated by the correlation calculation unit and the importance calculated by the importance calculation unit, creates a selection list that arranges the plurality of candidate codes in a manner where the higher the probability of being selected as a judgment code, the higher the priority.

[0011] The estimation system disclosed herein analyzes fault response records containing free statements input by a person who has performed fault response on the device. The estimation system comprises: an extraction unit that extracts specific object terms and words representing the actions taken during fault response from the free statements contained in the fault response records; a correlation calculation unit that calculates the correlation degree between each of a plurality of candidate codes that become candidates for a judgment code assigned to the fault response record and the object terms extracted by the extraction unit; an importance calculation unit that calculates the importance of the actions extracted by the extraction unit and correlates them with any one of the plurality of candidate codes; a listing unit that, based on the correlation degree calculated by the correlation calculation unit and the importance calculated by the importance calculation unit, creates a selection list in which the plurality of candidate codes are arranged in order of increasing probability of being selected as a judgment code; and an estimation unit that estimates the candidate code ranked highest in the selection list created by the listing unit as the judgment code.

[0012] Invention Effects

[0013] According to this disclosure, a selection list is created by arranging multiple candidate codes in a manner that prioritizes them based on the relevance of object terms extracted from free statements and the importance of treatments extracted from free statements. Therefore, it is possible to provide selections with a high probability of being selected. Furthermore, the estimation system estimates the candidate code ranked highest in the selection list as the judgment code. In this way, it is possible to estimate selections with a high probability of being chosen. Attached Figure Description

[0014] Figure 1 This is a diagram showing an outline of the elevator device utilizing the analysis system in Embodiment 1.

[0015] Figure 2 This is a cross-sectional view of the cold storage using the analysis system in Implementation Method 1.

[0016] Figure 3 This is a hardware structure diagram of the analysis system in Implementation Method 1.

[0017] Figure 4 This is a functional block diagram of the analysis system in Implementation Method 1.

[0018] Figure 5 This is a diagram showing an example of a retrieval user interface displayed by the analysis system in Implementation 1.

[0019] Figure 6 This is a diagram showing an example of a setting request screen displayed by the analysis system in Implementation 1.

[0020] Figure 7 This is a flowchart used to illustrate the operation summary of the analysis system in Implementation Method 1.

[0021] Figure 8 This is a hardware structure diagram of the analysis system in Implementation Method 2.

[0022] Figure 9 This is a flowchart used to explain the operation summary of the analysis system in Implementation Method 2. Detailed Implementation

[0023] The embodiments for implementing this disclosure are described with reference to the accompanying drawings. Furthermore, in the drawings, identical or equivalent parts are labeled with the same reference numerals. Repetitive descriptions of these parts are appropriately simplified or omitted.

[0024] Implementation method 1.

[0025] Figure 1 This is a diagram showing an outline of the elevator device utilizing the analysis system in Embodiment 1. Figure 2 This is a cross-sectional view of the cold storage using the analysis system in Implementation Method 1.

[0026] exist Figure 1The diagram shows a first example of a device utilizing the analysis system 1, namely an elevator unit 100. In the elevator unit 100, a shaft 101 runs through each floor of a building 102. Multiple landings 103 are respectively provided on each floor of the building 102. Multiple landing doors 104 are respectively provided at the entrances and exits of the multiple landings 103. The landing doors 104 are sliding doors that move along door tracks 105. A car 106 is located inside the shaft 101. The car 106 can move up and down inside the shaft 101. A control panel 107 is located in the machine room at the top of the shaft 101. The control panel 107 provides overall control of the operation of the elevator unit 100. When a malfunction occurs in the elevator unit 100, the control panel 107 can display an error code corresponding to the malfunction. A field terminal 108 is a terminal capable of inputting information and visually displaying information. For example, the field terminal 108 is a laptop computer. The field terminal 108 can communicate with a network N via wireless communication or the like.

[0027] Maintenance worker H1 belongs to the company that performs maintenance work on elevator unit 100. Upon receiving a notification that elevator unit 100 has malfunctioned, maintenance worker H1 proceeds to building 102 to perform response work, such as replacing the faulty part of elevator unit 100. Maintenance worker H1 inputs information indicating the results of the response work into the field terminal 108. At this time, maintenance worker H1 inputs information such as the error code displayed on the control panel 107 and the type of equipment that malfunctioned, as part of the response work results. In addition, maintenance worker H1 inputs information such as the condition upon arrival at building 102, the equipment condition obtained from the owner of building 102, the cause of the malfunction, and the content of the fault-handling work performed, using free-form statements based on natural language, as part of the response work results. Free-form statements are strings input using natural language and are strings that are not pre-set as selection options. Based on the input information, the field terminal 108 generates a fault response record containing information including the input free-form statements.

[0028] The analysis system 1 includes a retrieval terminal 2 and a server device 3. Additionally, the analysis system 1 may also include a field terminal 108. Figure 1 In this example, retrieval terminal 2 and server device 3 are located in a company performing maintenance work on elevator device 100. Specifically, for example, retrieval terminal 2 and server device 3 are located in the company's information center S. Server device 3 can communicate with control panel 107 via network N and a monitoring device (not shown) located near control panel 107. Server device 3 can receive fault response records from field terminal 108 via the network. Retrieval terminal 2 can display the fault response records received by server device 3 and accept input of information related to the fault response records.

[0029] Administrator H2 works in Information Center S. Administrator H2 operates Retrieval Terminal 2 to view fault response records. For each fault response record, Administrator H2 inputs a judgment code as information for information management. In this case, Server Device 3 generates information including the judgment code assigned to the fault response record and the fault response record itself, and registers it in the fault response history database.

[0030] Judgment codes are statistical codes used in the database related to fault response. One or more types of judgment codes are set for each fault response history. For example, the types of judgment codes may include statistical equipment names, cause codes, etc. Statistical equipment names are the names of equipment that were addressed in a particular response operation. Statistical equipment names use terms such as "door," "in storage," or "switch" to represent a broad concept. For example, if a part of a door was replaced during a response operation, "door" would be entered as the statistical equipment name. Cause codes are codes indicating the cause of a fault that was addressed in a particular response operation. These can be terms such as "wire break" or "deterioration over time" to represent a broad concept, or strings such as numbers indicating the cause. Multiple candidate codes are set for each judgment code to become input candidates.

[0031] Server device 3 analyzes the contents of the fault response log containing free-form statements based on natural language, and creates a selection list that rearranges multiple candidate codes for a specific judgment code. The selection list is a list that arranges multiple candidate codes in descending order of their probability of being selected as the judgment code.

[0032] Administrator H2 refers to the selection list displayed on search terminal 2 and selects candidate codes corresponding to each of the one or more judgment codes. In this case, information on the fault response history of the selected candidate codes as judgment codes is generated. Furthermore, administrator H2 inputs sorting criteria via search terminal 2, thereby being able to refer to a selection list that has been further filtered through multiple candidate codes included in the selection list. By using such a selection list, administrator H2 can easily select the appropriate judgment code.

[0033] Alternatively, maintenance personnel H1 can input the results of the response operation into the field terminal 108, and record the results in other formats such as reports. That is, the information from the fault response record is essentially summarized in a report, etc. In this case, administrator H2 can also input the content recorded in the report, etc., into the retrieval terminal 2. Retrieval terminal 2 and server device 3 can also treat this input information as a fault response record.

[0034] exist Figure 2The diagram shows a second example of the device utilizing analysis system 1, namely a cold storage 200. The cold storage 200 has a storage compartment 201 inside its housing. For example, the storage compartment 201 is divided into storage chambers 202, 203, 204, and 205. For example, storage chamber 204 can be opened and closed via a sliding door 206. The sliding door 206 is achieved by the housing 208 sliding along a door track 207.

[0035] The cold storage 200 generates cold air through a refrigeration cycle including a compressor 209 and a cooler 210. The generated cold air is supplied to the storage compartment 201. The control board 211 is electrically connected to the compressor 209, cooler 210, and other equipment via wiring (not shown). The control board 211 can perform overall control of the operation of the cold storage 200. A switch 212 covered by a switch cover is provided inside the storage compartment 201, i.e., on the wall surface. For example, when the switch 212 is operated, the control board 211 can change the operating state of the cold storage 200. In the event of a malfunction in the cold storage 200, the control board 211 can also display an error code indicating the malfunction on an LCD panel 213 or similar device.

[0036] exist Figure 2 In the example, the maintenance worker belongs to the company that repairs the cold storage 200. Upon receiving a notification that the cold storage 200 has malfunctioned, the maintenance worker goes to the building where the cold storage 200 is located and performs corresponding tasks such as replacing the faulty parts of the cold storage 200.

[0037] exist Figure 2 Not shown in the image, but, with Figure 1 Similarly, the maintenance worker inputs information indicating the result of the response operation into the field terminal 108. Based on the input information, the field terminal 108 generates a fault response record containing the input free statements. The administrator and maintenance worker belong to the same company. The administrator inputs information related to the fault response record into the retrieval terminal 2. At this time, the server device 3 generates a selection list.

[0038] Alternatively, after the owner of cold storage 200 responds to a malfunction in cold storage 200, information related to the malfunction response record can be sent to server device 3. In this case, for example, the owner can also input information related to the malfunction response record via a device such as a smartphone. This information can include free statements entered by the owner. The smartphone or server device 3 can also create a malfunction response record based on the input information. Alternatively, even in this case, the administrator can attach a judgment code to the malfunction response record.

[0039] Next, use Figure 3 and Figure 4 This explains the analysis of system 1. Additionally, the following section will discuss... Figure 2 The second example shown illustrates the analysis system 1 during fault response record processing.

[0040] Figure 3 This is a hardware structure diagram of the analysis system in Implementation Method 1. Figure 4 This is a functional block diagram of the analysis system in Implementation Method 1.

[0041] In the analysis system 1, one or more retrieval terminals 2 are connected to the server device 3 via a network to enable communication.

[0042] For example, the retrieval terminal 2 is composed of a personal computer (PC). The retrieval terminal 2 has a communication interface 2a, a processing circuit, an input interface 2b, and an output interface 2c. The processing circuit includes a processor 2d and a memory. The memory consists of a main storage device 2e and an auxiliary storage device 2f. Alternatively, the processing circuit can also refer to multiple electronic circuits or other devices housed in different casings.

[0043] Communication interface 2a is used to connect to networks such as wired LAN and wireless LAN. Retrieval terminal 2 can communicate with other devices such as server device 3 through communication interface 2a. Input interface 2b is the interface for receiving input operations from the administrator from input device 2g. Input devices include mice, keyboards, styluses, microphones, etc. Output interface 2c is the interface for displaying various information to output device 2h. Output device 2h includes monitors, etc.

[0044] For example, processor 2d is a device that performs calculations, such as a CPU (Central Processing Unit) or a central processing unit. The various functions of retrieval terminal 2 are implemented by processor 2d reading and executing programs for information processing stored in at least one of the main storage device 2e and auxiliary storage device 2f, which serve as memory. Figure 2 In this representation, the functions implemented by processor 2d are depicted as the structure of processor 2d. Various information used to implement the functions of retrieval terminal 2 is stored in the memory. Figure 2 In this example, the information stored in the memory is represented by the structure of the auxiliary storage device 2f.

[0045] For example, server device 3 is composed of a PC. For example, server device 3 has a communication interface 3a and processing circuitry. The communication interface 3a and processing circuitry may also have the same structure and function as the communication interface 2a and processing circuitry of retrieval terminal 2, respectively. The processing circuitry includes a processor 3b, a main storage device 3c as memory, and an auxiliary storage device 3d. The processor 3b, main storage device 3c, and auxiliary storage device 3d may also have the same structure and function as the processor 2d, main storage device 2e, and auxiliary storage device 2f of retrieval terminal 2, respectively. Figure 3 In this context, the functions implemented by processor 3b are represented by the structure of processor 3b. Furthermore, in... Figure 3 In this context, as an example, the information stored in the memory is represented by the structure of an auxiliary storage device in 3D.

[0046] Alternatively, some of the functions of the retrieval terminal 2 can be implemented using dedicated hardware, while other functions can be implemented using the processing circuitry of the retrieval terminal 2. Furthermore, some of the functions of the server device 3 can be implemented using dedicated hardware, while other functions can be implemented using the processing circuitry of the server device 3.

[0047] Alternatively, server device 3 can also be implemented on a cloud server. In this case, the processing circuit consists of multiple circuit components. These multiple processing circuit components are respectively located in multiple devices constituting the cloud server. The multiple devices constituting the cloud server can also be located in different buildings.

[0048] Functionally, the retrieval terminal 2 includes a display unit 20, an input unit 21, a reading unit 22, and an acquisition unit 23. The display unit 20 controls the content displayed on the output interface 2c. For example, the display unit 20 causes the output interface 2c to display a retrieval user interface screen for creating fault response history. The input unit 21 detects information input through the input interface 2b and reflects this in the device's operation. The reading unit 22 causes the server device 3 to read information from the fault response records selected by the administrator through the input unit 21. The acquisition unit 23 sends information such as information added to or assigned to the fault response records to the server device 3 based on the information selected in the retrieval user interface and information obtained from the network, etc.

[0049] As functions, the server device 3 includes a receiving unit 30, an extraction unit 31, a correlation calculation unit 32, a performance learning unit 33, an importance calculation unit 34, a priority calculation unit 35, a response learning unit 36, a list unit 37, a display control unit 38, a history production unit 39, and a setting request unit 40.

[0050] The receiving unit 30 receives fault response record information from the field terminal 108 or the retrieval terminal 2. The fault response record may include identification information of the equipment that responded to the fault, the error code at the time of the fault, the signal data transmitted from the equipment, the location of the equipment, and identification information of the maintenance personnel who responded. Furthermore, the receiving unit 30 may also collect this information from other databases or the equipment itself and correlate it with the fault response record.

[0051] Furthermore, the fault response log includes multiple fields for recording free statements. These fields include at least a first field containing a free statement indicating the cause of the fault, and a second field containing free statements detailing the actions taken to address the fault during the response process. A third field may also contain free statements describing the arrival status of information gathered before the maintenance personnel arrive on site and ascertain the cause.

[0052] The following explains an example of including a free statement as the cause of the fault in the first record column and a free statement as the handling content in the second record column of the fault response record. Furthermore, the processing when selecting "Statistical Equipment Name" as the judgment code is explained. "Statistical Equipment Name" refers to the equipment name selected as valid for the statistical classification, not the actual product name, etc. Additionally, record columns other than those shown in this example may be included, as long as the first and second record columns contain different meanings of the recorded free statements.

[0053] Extraction unit 31 has a first extraction unit 31a and a second extraction unit 31b. The first extraction unit 31a extracts specified object terms from the free statements contained in the fault response record. The object terms are terms needed when creating the selection list, and are determined according to the situation. In this example, the first extraction unit 31a extracts the device name as the object term. At this time, the first extraction unit 31a matches the extracted object term with which of the multiple record fields the object term is recorded in. The second extraction unit 31b extracts words indicating the handling of the fault from the free statements contained in the fault response record. At this time, the second extraction unit 31b extracts words indicating actions as the handling. The handling content basically records the handling for the device. The second extraction unit 31b matches the extracted handling with the device name as the object term. An example of the operation of extraction unit 31 is explained using the following Tables 1 and 2.

[0054] [Table 1]

[0055] [Table 2]

[0056] Table 1 shows examples of free statements included in three failure response records respectively. The second column is the free statement of the failure cause as the first record column. The third column is the free statement of the handling content as the second record column. The first extraction unit 31a extracts the device name shown in boldface in Table 1 as the target word. The second extraction unit 31b extracts the words of the handling shown underlined in Table 1. Table 2 shows the record positions and handling corresponding to the target words extracted in the example of Table 1. In addition, when the extraction unit 31 extracts words from the free statement, a known method for extracting words can also be used.

[0057] The relevance calculation unit 32 refers to the code DB50 and calls the candidate list related to the judgment code that is the current object. In the code DB50, candidate lists corresponding to multiple judgment codes that may be objects are stored. The candidate list is a list in which multiple candidate codes corresponding to the judgment code are arranged in a prescribed initial order. Hereinafter, the multiple candidate codes corresponding to the judgment code that is the object will be described. Specifically, each candidate code included in the statistical device name as the judgment code will be described. In this case, the candidate codes "inside the library", "thermometer inside the library", "cold storage", "sliding door", "door rail", "switch", and "switch cover" are included in the candidate list in the order of the record.

[0058] The relevance calculation unit 32 calculates the relevance between the target word extracted by the extraction unit 31 and each of the multiple candidate codes called. When the extraction unit 31 extracts multiple target words from the failure response record, the relevance calculation unit 32 calculates the relevance between the multiple target words and the candidate codes respectively. The relevance is an index value indicating the literal or semantic relevance between the target word and the candidate code. For example, the relevance can be the literal similarity between the target word and the candidate code, or the attribute similarity between the target word and the candidate code. This similarity can also be obtained using ontology based on information such as attributes attached to the target word and the candidate code.

[0059] In addition, the edit distance between the target word and the candidate code can be used as the literal similarity. Specifically, when the target word is a device name and the candidate code is a statistical device name, the relevance calculation unit 32 calculates the edit distance between the device name and the statistical device name as the relevance. For example, when the device name is "door" and the statistical device name is "door rail", considering one character insertion, deletion, or replacement as one step, converting "door" (Japanese: "ドア") to "door rail" (Japanese: "ドアレ") The minimum number of steps when "ル" is 3 is required. In this case, the correlation calculation unit 32 calculates the correlation between the device name and the statistical device name as 3, which is the edit distance.

[0060] The correlation calculation unit 32 associates the candidate code with the highest correlation among multiple candidate codes with a specific object term. Alternatively, the correlation calculation unit 32 can also associate the object term with one or more candidate codes whose correlation exceeds a predetermined threshold. In this case, the correlation calculation unit 32 can also temporarily associate one or more object terms extracted from the currently processed fault response record with multiple candidate codes in the candidate list.

[0061] The Performance Learning Department 33 calculates the performance importance of each action extracted from multiple past fault response histories stored in the historical DB51. Performance importance is calculated based on the number of times an action becomes associated with a judgment code selected in the fault response history.

[0062] The importance calculation unit 34 matches the handling extracted from the fault response record by the extraction unit 31 with any candidate code from a plurality of candidate codes. At this time, the importance calculation unit 34 matches the candidate code corresponding to the object term corresponding to the handling with that handling. Table 3 below shows examples of the device name, handling, and object device name matched by the importance calculation unit 34 as object terms.

[0063] [Table 3]

[0064] As shown in Table 3, if based on row #3-1, then the action of "inspection" corresponds to the equipment name such as "door" in the fault response record. In this case, the importance calculation unit 34 associates the action of "inspection" with the statistical equipment name such as "door track".

[0065] The importance calculation unit 34 calculates the overall importance of each extracted process as its overall importance. First, the importance calculation unit 34 retrieves the basic importance from the process DB52 for each process. The basic importance is information stored in the process DB52 and represents a pre-set importance for each process. Alternatively, the basic importance can be manually set or updated beforehand.

[0066] Here, the processing of the Performance Learning Department 33 is explained in more detail. The Performance Learning Department 33 retrieves multiple fault response histories stored in the History DB51. In each fault response history, according to multiple judgment codes, a candidate code is included as that judgment code. As shown in Table 3, the handling corresponds to the candidate code. The Performance Learning Department 33 determines the handling corresponding to the judgment code as associated handling in each of the multiple fault response histories. Alternatively, in the associated handling, not only the word itself corresponding to the candidate code can be determined, but also its synonyms. The Performance Learning Department 33 calculates the number of associated handlings, i.e., the number of times it becomes an associated handling, for each handling stored in the Handling DB52. The Performance Learning Department 33 calculates the performance importance for each handling in a manner where the more times it becomes an associated handling, the higher its importance. For example, the performance importance value can be calculated in a manner where the more times it becomes an associated handling, the lower its importance. Alternatively, the performance importance can also be calculated based on the proportion of handlings that become associated handlings, rather than the number of times they become associated handlings. The performance learning unit 33 maps the calculated performance importance to the multiple actions stored in the action DB52. For example, the performance learning unit 33 can also update the performance importance whenever a new fault response record is stored in the history DB51.

[0067] The importance calculation unit 34 calculates the overall importance based on at least one of the basic importance and the performance importance calculated by the performance learning unit 33. Furthermore, the basic importance, performance importance, and overall importance can all be processed such that the smaller the value, the higher the importance. For example, the smaller the importance value of a certain action, the higher the probability that the candidate code corresponding to that action will be selected as the judgment code. As an example, as shown in equation (1) below, the importance calculation unit 34 calculates the overall importance by calculating the weighted sum of the basic importance and the performance importance for a certain action.

[0068] Overall importance = A × basic importance + B × actual performance importance (1)

[0069] A and B are constants preset based on human input or machine learning. Alternatively, A or B can be set to 0. When B is set to 0, importance is determined solely by the base importance. When A is set to 0, importance is determined solely by performance importance. Table 4 below shows an example of how the overall importance is calculated.

[0070] [Table 4]

[0071] In Table 4, the number of times of becoming an associated disposition is shown in parentheses in the row of the actual importance. In this example, the relationship of y < x < z holds. Therefore, for example, the value of the actual importance of the disposition of "maintenance" is the smallest and the most important. In Table 4, constants of A = B = 1 are set.

[0072] The priority calculation unit 35 calculates a comprehensive priority according to at least one of the basic priority and the immediate priority for each object word extracted by the extraction unit 31. Specifically, first, the priority calculation unit 35 calls the basic priority corresponding to the extracted object word from the object word DB 53 that stores the assumed object words. The basic priority is a preset priority stored in the object word DB 53. In addition, the basic priority can also be manually set or updated by a person in advance.

[0073] The priority calculation unit 35 calculates the immediate priority for each object word using the fault response record that is the object. The priority calculation unit 35 calculates the immediate priority in the following manner: the value of the immediate priority is different depending on which of the multiple record columns the object word is included in, where the multiple record columns are the multiple record columns included in the fault response record that is the object. For example, when the priority calculation unit 35 records the object word only in the first record column, the immediate priority is calculated to be higher than when the object word is only recorded in the second record column. When the priority calculation unit 35 records the object word in both the first record column and the second record column, the immediate priority is calculated to be higher than when the object word is only recorded in the first record column. In addition, regarding the value of the basic priority and the value of the immediate priority, they can also be calculated in such a way that the higher the priority, the smaller the value. The following Tables 5 to 7 respectively show examples of the basic priority, the immediate priority, and the comprehensive priority.

[0074] [Table 5]

[0075] [Table 6]

[0076] [Table 7]

[0077] Table 5 shows examples of the set base priorities. According to the examples, the object term "switch" has the highest base priority and the lowest value. Table 6 shows examples of the set values ​​for the real-time priority calculated based on the location of the object term. According to Table 6, the real-time priority is highest when the object term is recorded in both the fault cause in the first record column and the handling content in the second record column. The priority is lowest when the object term is recorded only in the handling content in the second record column. Furthermore, the values ​​of the base priority and the real-time priority are not limited to the examples shown in Tables 5 and 6. For example, when the object term is recorded only in the second record column, the real-time priority value can be set to "5" instead of "3".

[0078] As shown in Table 7, the priority calculation unit 35 calculates the comprehensive priority by weighting the basic priority and the real-time priority of a certain object. Specifically, the priority calculation unit 35 calculates the comprehensive priority according to the following formula (2).

[0079] Overall priority = C × basic priority + D × on-demand priority (2)

[0080] C and D are constants preset based on human or machine learning. Alternatively, C or D can be set to 0. When D is set to 0, the priority is determined solely by the base priority. When C is set to 0, the priority is determined solely by the current priority. In the examples in Table 7, constants such as C=D=1 are set.

[0081] The response learning unit 36 ​​calculates response indicator values ​​for each fault response history based on multiple fault response histories stored in the historical database 51, for each candidate code. The response indicator value can be the number of responses assigned to a fault response record, or a response ratio calculated based on the number of responses. Furthermore, if there is information corresponding to a candidate code contained in an error code within multiple fault response histories, the response learning unit 36 ​​can add the number of instances of information corresponding to that candidate code to the number of responses in the associated indicators. For example, if the candidate code is a statistical device name, it can be added to the number of responses to the statistical device name corresponding to the device contained in the error code. Similarly, if the candidate code is a cause code, it can be added to the number of responses to the cause code corresponding to the cause contained in the error code. The response learning unit 36 ​​stores the information obtained by mapping response indicator values ​​to candidate codes in the associated database 54. For example, the response learning unit 36 ​​can update the response indicator values ​​whenever a new fault response record is stored in the historical database 51. Tables 8 and 9 below show examples of response counts and response ratios, respectively.

[0082] [Table 8]

[0083] [Table 9]

[0084] Table 8 shows the number of responses recorded for each candidate code, i.e., the statistical device name, in the fault response history, including multiple error codes A to Z. In this example, no devices corresponding to the candidate codes are shown in codes B to Y. For example, the response learning unit 36 ​​calculates the number of times the statistical device name "inside the warehouse" is recorded in codes A to Z as 5 times. For example, the response learning unit 36 ​​calculates the number of responses recorded for the statistical device name when the error code is code Z as follows: "door track", "switch", "switch cover" as 2 times, "inside the warehouse", "inside the warehouse thermometer and hygrometer", "cold storage", "sliding door" as 0 times. Alternatively, the number of responses can also be calculated as a weighted sum of the number of times selected as a candidate code and the number of times included in the error code, as shown in the following formula (3).

[0085] Number of responses = E × number of times selected as a candidate code + F × number of times included in error codes (3)

[0086] E and F are constants that are pre-set based on human or machine learning. In addition, E or F can also be set to 0.

[0087] Table 9 shows the response ratios calculated using the same number of responses as in Table 8. The response learning unit 36 ​​can also use the response ratio instead of the number of responses. The response ratio is the sum of the individual response ratios calculated for each error code for all error codes. By using the response ratio as a response indicator value, as shown in the "switch shield" in Table 9, sometimes the response indicator value for candidate codes that appear frequently only in a particular error code is larger than when using the number of responses.

[0088] In addition to error codes, the number of responses or the response ratio can also be calculated for each device's installation location, and then added to the response index value of the candidate code corresponding to the device. For example, the device's installation location can also be divided into "indoor" and "outdoor" to calculate the number of responses or the response ratio.

[0089] Listing unit 37 calls the candidate list corresponding to the judgment code that becomes the object, and rearranges the order of the multiple candidate codes contained in the candidate list to create a selection list. At this time, as shown in Table 10 below, listing unit 37 rearranges the codes multiple times to create the selection list. Table 10 shows an example of the following situation: the free statement for the fault cause in the first record column states "liquid overflowed in the warehouse", and the free statement for the handling content in the second record column states "door inspection. switch maintenance".

[0090] [Table 10]

[0091] First, the listing unit 37 determines one or more containing candidate codes from a plurality of candidate codes. These one or more containing candidate codes contain at least one of the object terms extracted by the extraction unit 31 as a string. The listing unit 37 creates a first list arranged in such a way that the containing candidate codes are ranked in order of priority that the non-containing candidate codes precede. Furthermore, Table 10 shows the order of the candidates in the first list as ":1st position". In the first list, all containing candidate codes have the same priority. In the first list, all non-containing candidate codes have the same priority.

[0092] In the example in Table 10, the object terms "inside the cold storage," "door," and "switch" were extracted from the fault response records that became objects. Therefore, the candidate codes included were "inside the cold storage," "inside the cold storage thermometer and hygrometer," "sliding door," "door track," "switch," and "switch cover." In List 1, the device names of these six objects containing candidate codes were all set to one position each. "Cold storage," which did not contain a candidate code, was set to seven positions.

[0093] Then, the listing unit 37 further assigns a ranking to candidate codes with the same position among the multiple candidate codes included in the first list and rearranges them, thereby creating a second list. Additionally, Table 10 shows the ranking in the second list as ": 1st place". The listing unit 37 rearranges candidate codes with the same position based on at least one of the relevance calculated by the relevance calculation unit 32 and the importance calculated by the importance calculation unit 34. In this case, the listing unit 37 rearranges the multiple candidate codes in such a way that the higher the relevance to the object term corresponding to each candidate code and the higher the importance of the corresponding action, the higher the ranking. Furthermore, in addition to relevance and importance, the listing unit 37 can also rearrange candidate codes with the same position based on the priority calculated by the priority calculation unit 35. In this case, the listing unit 37 rearranges the multiple candidate codes in such a way that the higher the relevance to the object term corresponding to each candidate code and the higher the importance of the corresponding action and the higher the priority of the corresponding object term, the higher the ranking.

[0094] Specifically, the listing unit 37 calculates an index value for each of the multiple candidate codes. This index value is a weighted sum of relevance, importance, and priority. The listing unit 37 compares the index values ​​assigned to candidate codes of the same priority and rearranges the list so that the smaller the index value, the higher the priority, as the second list. Alternatively, it can be calculated as a weighted sum of the number of times a code is selected as a candidate code and the number of times it is included in the error code, as shown in equation (4) below.

[0095] Indicator value = G × Relevance + H × Importance + I × Priority (4)

[0096] G, H, and I are constants preset based on human or machine learning. Additionally, I can be set to 0. When I is set to 0, the indicator value is calculated based on relevance and importance. In the example of Table 10, H=G=I=1. Listing section 37 arranges the six statistical device names that have the same 1st position in the first list in order of increasing rank with smaller indicator values. Multiple candidate codes that have the same rank in the first list and the same indicator value in the second list are assigned the same rank in the second list. In the example of Table 10, listing section 37 sets "Inside the Library" and "Sliding Door" to 4 digits.

[0097] Then, the listing unit 37 further assigns a ranking to candidate codes with the same position among the multiple candidate codes included in the second list and rearranges them, thereby creating a third list. Additionally, Table 10 shows the ranking in the third list as ":1st position". The listing unit 37 rearranges the multiple candidate codes with the same position in the second list in such a way that the higher the response index value calculated by the response learning unit 36, the higher the ranking, thus creating the third list.

[0098] In the example in Table 10, the number of responses is used as the response index value. In the second list, the same 4-digit "in-the-library" and "sliding door" have response counts of 5 and 8, respectively. The list unit 37 sets the "sliding door" with the higher response index value, i.e., more response counts, to 4 digits, and sets the "in-the-library" with fewer response counts to 5 digits. The list unit 37 then rearranges the multiple candidate codes according to the updated order as the third list. Table 11 below shows the third list created based on the example in Table 10.

[0099] [Table 11]

[0100] Then, the listing unit 37 removes candidate codes that are not eligible from among the multiple candidate codes present in the third list, based on at least one of the model information and error code information contained in the fault response record, thereby creating a fourth list. Alternatively, if no candidate codes are not eligible, the listing unit 37 creates the same fourth list as the third list. For example, if the fourth list is created solely based on model information, the listing unit 37 obtains model correspondence information from the model DB55, indicating the correspondence between the model and the multiple candidate codes.

[0101] For example, since some devices may be listed as candidate codes in the candidate list but not installed in the actual device, there may be codes that are not selected as candidate codes. The model compatibility information indicates whether a candidate code is "object" or "not object" for each model. The list unit 37 compares the model compatibility information with the models included in the fault response record and removes candidate codes that are "not object" for that model from the third list. Tables 12, 13, and 14 below are examples of the case where the fourth list is created solely based on model information.

[0102] [Table 12]

[0103] [Table 13]

[0104] [Table 14]

[0105] In the example of Table 12, it is assumed that the fault response record for this incident includes information about model a of the device that was fault-responded to. In this case, as shown by the underlined section in Table 13, "inside the warehouse," "inside the warehouse thermometer and hygrometer," "cold storage," and "sliding door" are designated as "objects" based on the model information. Furthermore, "door track," "switch," and "switch cover" are designated as "outside objects." As shown in Table 14, the list section 37 creates a fourth list by deleting the statistical equipment names designated as "outside objects" from the third list.

[0106] Furthermore, when creating a fourth list based on the error codes, the listing unit 37 similarly removes candidate codes from the list shown in Table 14 based on the "error response information" to obtain the fourth list. Specifically, the listing unit 37 obtains error response information from error DB56. Error response information is information indicating the correspondence between error codes and multiple candidate codes. That is, error response information indicates whether each candidate code is selected as an "object" or "not an object" for each error code. The listing unit 37 removes candidate codes that are "not an object" from the list shown in Table 14 based on the error codes shown in the fault response record that are now objects, thus creating the fourth list.

[0107] In addition, when the fourth list is created solely based on the error codes, the list unit 37 similarly creates a fourth list based on the error codes contained in the error response information and fault response records, by removing candidate codes from the third list that have become "non-object".

[0108] Then, list section 37 outputs the fourth list as a selection list.

[0109] Alternatively, list unit 37 may not output list 4 as the selection list, but instead output any one of list 1, list 2, or list 3 as the selection list. In list 1, list 2, list 3, list 4, and any selection list, multiple candidate codes are arranged such that the higher the probability of being selected as the judgment code, the higher its priority. Furthermore, the more recently created a list is, the higher the precision of the probability of it being selected as the judgment code.

[0110] The display control unit 38 causes the search terminal 2 to display the selection list output by the display list unit 37. For example, the display control unit 38 causes the display unit 20 of the search terminal 2 to display the search user interface screen. For example, the display control unit 38 first displays only the candidate code that is ranked first among the multiple candidate codes contained in the selection list. A button for accepting the selection operation is also displayed near the candidate code that is ranked first. If the administrator operates the button for accepting the selection operation, the display control unit 38 displays the multiple candidate codes contained in the selection list in such a way that the earlier ones are displayed are placed at the top. The display control unit 38 accepts the selection of the candidate code corresponding to a certain judgment code from the administrator via the search terminal 2. In addition, the display control unit 38 can also accept the selection of multiple candidate codes for one judgment code.

[0111] The display control unit 38 can further accept input of sorting conditions or input of rearrangement conditions via the search user interface.

[0112] The sorting criteria are used to remove candidate codes that do not meet the sorting criteria from the selection list while maintaining the order of the candidate codes contained in the selection list. The display control unit 38 removes candidate codes from the selection list that do not meet the sorting criteria, and displays the candidate codes in the selection list after the removal process in a manner where the earlier the code appears, the higher it is in the order.

[0113] For example, the sorting criteria could be as follows: accepting input from at least one of the first and second record fields, and displaying candidate codes corresponding to the object terms recorded in the record field of the field that received the input. In this case, the display control unit 38 can also delete candidate codes corresponding to object terms not recorded in the record field of the field that received the input from the selection list, and display the multiple candidate codes included in the selection list after the deletion process is completed.

[0114] For example, the sorting criteria could also be as follows: Inputting a boundary value for the importance of the accepted performance, displaying candidate codes corresponding to actions with a performance importance exceeding the boundary value. In this case, the display control unit 38 can also delete candidate codes corresponding to actions with a performance importance less than the boundary value from the selection list, and display multiple candidate codes included in the selection list after the deletion process is completed.

[0115] The rearrangement condition is used to rearrange the order of multiple candidate codes contained in the selection list. When the display control unit 38 receives input of the rearrangement condition, the list unit 37 rearranges the multiple candidate codes contained in the candidate list according to the rearrangement condition to create a selection list. Then, the display control unit 38 causes the retrieval terminal 2 to display the rearranged selection list.

[0116] For example, the rearrangement condition is a condition that changes a constant such as the weighted ratio of values ​​for importance, priority, and index. Specifically, the display control unit 38 receives input as a condition for changing the weighted ratio of the base priority to the current priority as the rearrangement condition. When the list unit 37 creates a selection list based on the candidate list, especially when calculating priorities, it calculates the comprehensive priority of the target words based on constants C and D corresponding to the input ratio. After creating the selection list again, the display control unit 38 displays the selection list reflecting the rearranged priority conditions.

[0117] The history production department 39 assigns the candidate codes selected by the display control department 38 as corresponding judgment codes to the fault response records that become the targets. Furthermore, if multiple candidate codes are selected, the history production department 39 can also assign multiple candidate codes as judgment codes. When all judgment codes associated with the fault response records that become the targets are assigned, the history production department 39 creates fault response history information including the fault response records and the assigned judgment codes. The history production department 39 stores the created fault response history in the history DB51. That is, the history DB51 stores multiple fault response histories created by the history production department 39 in the past.

[0118] The setting request unit 40 compares the fault handling history contained in the historical DB51 with the handling DB52, and extracts the handling procedures for which no basic importance has been set. The setting request unit 40 creates an importance request list that ranks the extracted handling procedures. The setting request unit 40 compares the fault handling history contained in the historical DB51 with the object terminology DB53, and extracts the object terms for which no basic priority has been set. The setting request unit 40 creates a priority request list that ranks the extracted object terms. The setting request unit 40 compares the fault handling history contained in the historical DB51 with the machine model DB55, and extracts combinations of statistical equipment names and machine models for which no object status has been set. The setting request unit 40 creates a combination request list that ranks the extracted combinations.

[0119] The setting request unit 40 can also cause the search terminal 2 to display a setting request screen at any time, such as when operated by the search terminal 2. This setting request screen displays at least one of the following: an importance request list, a priority request list, and a combination request list. When a setting is entered in the setting request screen, the information of each database is updated.

[0120] Next, use Figure 5 This section provides an example of a user interface for retrieving data.

[0121] Figure 5 This is a diagram showing an example of a retrieval user interface displayed by the analysis system in Implementation 1.

[0122] Figure 5 The display control unit 38 displays a search user interface (hereinafter also referred to as the "selection UI") that shows multiple candidate codes Ex1 related to a certain judgment code, sorting conditions Ex2, and rearrangement conditions Ex3. The selection UI displays a button Ex4 for accepting selection operations. When the "change" button Ex4 is operated, not only is the first-ranked candidate code, i.e., "switch," displayed, but also multiple candidate codes Ex1 are also displayed.

[0123] In this example, as sorting criterion Ex2, the input "location of device name" was selected as the condition to display only candidate codes corresponding to the object terms recorded in "disposal content".

[0124] Furthermore, in this example, the following condition was input as the rearrangement condition Ex3: Rearrangement is performed based on a comprehensive priority calculated to make the weighting constant D of the current priority greater than the weighting constant C of the base priority. Specifically, multiple candidate codes Ex1 are displayed after rearrangement using the following comprehensive priority, which is calculated by the priority calculation unit 35 with the base priority weighting constant C = 0.5 and the current priority weighting constant D = 1.0. Additionally, when the option "Rearrange with priority given to important device names" is selected in the next section, rearrangement is performed based on a comprehensive priority calculated to make the weighting constant C of the base priority greater than the weighting constant D of the current priority.

[0125] Next, use Figure 6 This section provides an example of setting up a request screen.

[0126] Figure 6 This is a diagram showing an example of a setting request screen displayed by the analysis system in Implementation 1.

[0127] like Figure 6 As shown, the setting request screen displays the importance request list Ex5, the priority request list Ex6, and the combination request list Ex7, respectively. The setting request unit 40 processes updates to setting values ​​in each list. When a setting is entered, the setting request unit 40 updates the DB information corresponding to the entered setting value to the entered information.

[0128] Next, use Figure 7 This explains the actions of the analysis system 1.

[0129] Figure 7 This is a flowchart illustrating the outline of the operation of the analysis system in Implementation 1.

[0130] For example, the flowchart begins when the maintenance personnel input information into the field terminal 108 after completing the fault response work. In step S1, the field terminal 108 creates a fault response record and sends it to the server device 3.

[0131] Then, in step S2, server device 3 stores fault response records. The administrator retrieves the fault response records as objects from retrieval terminal 2.

[0132] Then, in step S3, the server device 3 creates a selection list for one or more judgment codes corresponding to the fault response record that becomes the object.

[0133] Then, in step S4, the server device 3 causes the retrieval terminal 2 to display the retrieval UI. The selection list created in step S3 is displayed in the retrieval UI. The display control unit 38 accepts candidate code input from the administrator in the retrieval UI. Furthermore, if a rearrangement condition is selected in the retrieval UI, the server device 3 creates a selection list based on the rearrangement condition and displays it again in the retrieval UI.

[0134] Then, in step S5, the display control unit 38 of the server device 3 determines whether a registration operation has been performed on the fault response record that is the target. At this time, the administrator performs a registration operation after all the management codes associated with the fault response record that is the target are entered. If no registration operation is performed in step S5, the server device 3 repeats the actions after step S4.

[0135] If the registration operation was performed in step S5, the action in step S6 is performed. In step S6, server device 3 creates a fault response history related to the fault response record that became the object. Server device 3 stores the fault response history in history DB51.

[0136] Then, in step S7, the performance learning unit 33 of the server device 3 updates the performance importance of handling DB52 based on the fault response record registered in step S6. The response learning unit 36 ​​of the server device 3 updates the response index value associated with DB54 based on the fault response history registered in step S6.

[0137] Then, the flowchart's actions end.

[0138] According to Embodiment 1 described above, the analysis system 1 includes an extraction unit 31, a correlation calculation unit 32, an importance calculation unit 34, and a listing unit 37. The listing unit 37 creates a selection list by rearranging the order of multiple candidate codes based on correlation and importance. In the selection list, multiple candidate codes are arranged in order of increasing probability of being selected as a decision code. Therefore, the analysis system 1 can provide selections with a high probability of being chosen.

[0139] Furthermore, the analysis system 1 also includes a display control unit 38. The display control unit 38 displays multiple candidate codes included in the selection list in a manner where the codes listed earlier are placed at the top. Therefore, the analysis system 1 can, for example, visually present multiple options with high probability of selection to the administrator in a manner including their order. Furthermore, for example, when searching for a suitable candidate code from the candidate list, the administrator can refer to it sequentially starting with the most likely candidate codes. As a result, in management tasks such as creating fault response history, the administrator's work time can be reduced. Moreover, since the most likely candidate codes are displayed, the phenomenon of the administrator selecting codes with low management value, such as "other," can be suppressed.

[0140] Furthermore, the display control unit 38 first displays only the candidate code that appears first in the selection list. Therefore, in management tasks such as creating fault response history, administrator time can be reduced.

[0141] Furthermore, when creating the selection list, the list unit 37 arranges multiple candidate codes in a manner that prioritizes those with higher relevance and importance. Therefore, in the selection list, multiple candidate codes can be arranged in a way that prioritizes those with a higher probability of being selected as the judgment code. For example, the object term is a device name, the candidate code is a statistical device name, and the relevance is the edit distance between the device name and the statistical device name. In this case, given the characteristics of the free statement, the selection list can be created based on the edit distance that is assumed to have high relevance. As a result, multiple candidate codes can be arranged in a way that prioritizes those with a higher probability of being selected as the judgment code.

[0142] Furthermore, the analysis system 1 also includes a historical production unit 39 and a performance learning unit 33. The performance learning unit 33 calculates the number of times a handling action becomes a related action based on past failure handling history. The importance calculation unit 34 calculates the importance of the handling action using a performance importance rating that increases with the number of times it becomes a related action. In other words, the importance calculated by the analysis system 1 reflects past performance such as the number of times a handling action has become a related action. As a result, the analysis system 1 can further increase the likelihood that the provided options will be selected.

[0143] Furthermore, the importance calculation unit 34 calculates the weighted sum of the actual importance and the set base importance as the importance of the action. Therefore, the importance is based not only on the actual performance but also on settings such as what actions are considered important by people. For example, even without accumulating a sufficient amount of fault handling history to ensure accuracy, it is possible to set actions that should be emphasized based on the base importance. As a result, the likelihood of the provided options being selected can be further increased.

[0144] Furthermore, the analysis system 1 also includes a priority calculation unit 35. The priority calculation unit 35 calculates the priority of the target term. The listing unit 37 arranges multiple candidate codes in order of priority, with higher priority terms appearing earlier. Therefore, the analysis system 1 can provide options that are highly likely to be selected.

[0145] Furthermore, the priority calculation unit 35 calculates the real-time priority, the value of which varies depending on whether the object term is contained in the first or second record field. The priority calculation unit 35 calculates the weighted sum of the real-time priority and the basic priority as the priority of the object term. For example, it is possible that when the first object term is recorded in the "Cause of Fault" field of the first record field and the second object term is recorded in the "Handling Content" field of the second record field, the candidate code corresponding to the first object term is more likely to be selected than the candidate code corresponding to the second object term. By calculating the priority based on the recording position of the object term, the analysis system 1 can further increase the probability that the provided options are selected.

[0146] Furthermore, the list unit 37 calculates the weighted sum of relevance, importance, and priority, which is the index value. Therefore, the list unit 37 can create a list that arranges multiple candidate codes in a more specific order, with higher relevance, higher importance, and higher priority being ranked higher.

[0147] Furthermore, after displaying multiple candidate codes, the display control unit 38 accepts input of rearrangement conditions. The list unit 37 rearranges the order of the selection list according to the rearrangement conditions. Therefore, administrators can input rearrangement conditions after confirming the display of the selection list. Furthermore, administrators can refer to the multiple candidate codes rearranged according to the rearrangement conditions. Specifically, the rearrangement conditions can also be conditions that change the weighted ratio of the base priority and the current priority. In this case, candidate codes that meet the rearrangement conditions deemed necessary by the administrator and have a high probability of being selected can be displayed. As a result, the analysis system 1 can reduce the administrator's work time.

[0148] Furthermore, the display control unit 38 displays multiple candidate codes in the selection list after deleting candidate codes that do not meet the acceptance sorting criteria, with the earlier-ranked codes appearing in the higher position. Therefore, it is possible to display candidate codes that meet the sorting criteria deemed necessary by the administrator and are highly likely to be selected. As a result, the analysis system 1 can reduce the administrator's work time.

[0149] Furthermore, the analysis system 1 also includes a response learning unit 36. The response learning unit 36 ​​calculates response index values ​​based on the fault response history. When there are two or more candidate codes with the same priority determined by information including relevance and importance, the listing unit 37 arranges these two or more candidate codes in a manner that the higher the response index value, the higher the priority, and sets them up as a third list. Therefore, the analysis system 1 can provide options with a high probability of being selected.

[0150] Furthermore, the list unit 37 identifies multiple candidate codes containing multiple object terms as strings and creates a first list. The list unit 37 then creates a second list based on the first list. Therefore, the analysis system 1 is able to provide options that are highly likely to be selected.

[0151] Furthermore, the processing disclosed herein can be applied even when other codes are selected as the judgment codes instead of "statistical device name". In this case, regarding the words extracted or used in each process, only words according to the judgment code need to be selected. For example, when "cause code" is selected as the judgment code, the second extraction unit 31b can extract words indicating cause from free statements instead of words indicating disposal. As for relevance, the relevance calculation unit 32 can also calculate the relevance between the cause code and the cause as an object term. The performance learning unit 33 can also calculate the number of times that words indicating cause but not disposal are associated with the object code assigned to the fault response history to become associated causes, and calculate the performance importance. The importance calculation unit 34 can also calculate the importance by using words indicating cause but not disposal as object terms.

[0152] In addition, when creating the fourth list, the list unit 37 may determine whether it is an "object" or "non-object" based on a database associated with other codes issued by the device, content entered by the maintenance personnel, building names, and other information contained in the fault response records, rather than on the error code.

[0153] In addition, the settings contained in each database can also be updated by importing the corresponding database information from external sources.

[0154] Implementation method 2.

[0155] Figure 8 This is a hardware structure diagram of the analysis system in Implementation Method 2. Figure 9 This is a flowchart outlining the operation of the analysis system in Embodiment 2. Furthermore, parts that are identical or equivalent to those in Embodiment 1 are labeled with the same reference numerals. Descriptions of these parts are omitted.

[0156] In implementation method 2, the analysis system 1 functions as an estimation system. Based on the fault response record, the estimation system estimates the candidate code most suitable for the judgment code. The estimation system then creates a fault response history that assigns the estimated candidate code as the judgment code to the fault response record.

[0157] Specifically, such as Figure 8 As shown, the estimation system, which is the analysis system 1, also has an estimation unit 41 as its function. For example, in embodiment 2, the display control unit 38 may not be provided.

[0158] When the list unit 37 creates a selection list based on fault response records, the estimation unit 41 estimates the candidate code that ranks first in the selection list as the judgment code to be assigned. In this case, the history production unit 39 assigns the candidate code estimated as the judgment code by the estimation unit 41 as the corresponding judgment code to the fault response record that becomes the object, and creates a fault response history.

[0159] Figure 9 This diagram illustrates a summary of the operation of the analysis system 1, or estimation system, in Embodiment 2. Step S1 in this flowchart is similar to that in Embodiment 1. Figure 7 The steps are the same as S1 in the flowchart.

[0160] Following step S1, in step S11, server device 3 stores the sent fault response record. Server device 3 then retrieves the fault response record as an object.

[0161] Then, in step S3, in accordance with embodiment 1 Figure 7 Similarly, the server device 3 creates a selection list in the flowchart.

[0162] Then, in step S12, the estimation unit 41 of the server device 3 estimates the candidate code with the highest priority from the corresponding selection list as the judgment code to be assigned for all judgment codes related to the fault response record that is the target.

[0163] Then, in step S13, the history production unit 39 of the server device 3 produces a fault response history that uses the candidate codes estimated by the estimation unit 41 in step S12 as judgment codes.

[0164] Then, in step S7, compared with embodiment 1 Figure 7 Similarly, in the flowchart, server device 3 updates each database.

[0165] Then, the flowchart's actions end.

[0166] According to Embodiment 2 described above, the estimation system includes an extraction unit 31, a correlation calculation unit 32, an importance calculation unit 34, a listing unit 37, and an estimation unit 41. The listing unit 37 creates a selection list by rearranging the order of multiple candidate codes based on correlation and importance. The selection list is arranged in a manner where the higher the probability of being selected as a decision code, the higher its ranking. The estimation unit 41 estimates the candidate code ranked highest in the selection list as the decision code. Therefore, the estimation system can estimate the selection item with a high probability of being selected.

[0167] Industrial availability

[0168] As described above, the analysis system disclosed herein can be used as a system for managing the maintenance records of an apparatus.

[0169] Label Explanation

[0170] 1: Analysis system; 2: Retrieval terminal; Communication interface: 2a; Input interface: 2b; Output interfaces: 2c, 2d: Processor; 2e: Main storage device; 2f: Auxiliary storage device; 2g: Input device; 2h: Output device; 3: Server device; Communication interfaces: 3a, 3b: Processor; 3c: Main storage device; 3d: Auxiliary storage device; 20: Display unit; 21: Input unit; 22: Read-in unit; 23: Acquisition unit; 30: Receiving unit; 31: Extraction unit; 31a: First extraction unit; 31b: Second extraction unit; 32: Correlation calculation unit; 33: Performance learning unit; 34: Importance calculation unit; 35: Priority calculation unit; 36: Response learning unit; 37: List unit; 38: Display control unit; 39: History production unit; 40: Setting request unit; 41: Estimation unit; 50: Code database; 51: History database; 52: Processing database; 53: Object terminology DB; 54: Association DB; 55: Model DB; 56: Error DB; 100: Elevator device; 101: Shaft; 102: Building; 103: Landing; 104: Landing door; 105: Door track; 106: Car; 107: Control panel; 108: Field terminal; 200: Cold storage; 201: Storage room; 202, 203, 204, 205: Storage room; 206: Sliding door; 207: Door track; 208: Housing; 209: Compressor; 210: Cooler; 211: Control board; 212: Switch; 213: LCD panel; Ex1: Candidate code; Ex2: Sorting condition; Ex3: Rearrangement condition; Ex4: Button; Ex5: Importance request list; Ex6: Priority request list; Ex7: Request list; H1: Maintenance personnel; H2: Administrator; N: Network; S: Information center.

Claims

1. An analysis system that analyzes fault response records containing free statements entered by a person who performed a fault response to the device, wherein, This analysis system has the following features: The extraction unit extracts specific object terms and words indicating the actions taken during fault response from the free statements contained in the fault response record; The correlation calculation unit calculates the correlation between each of the multiple candidate codes that become the judgment codes assigned to the fault response record and the object words extracted by the extraction unit. An importance calculation unit calculates the importance of the processing extracted by the extraction unit and assigns it to any one of the plurality of candidate codes; as well as The list section generates a selection list based on the relevance calculated by the relevance calculation section and the importance calculated by the importance calculation section, arranging the multiple candidate codes in a manner that the higher the probability of being selected as the judgment code, the higher the ranking.

2. The analysis system according to claim 1, wherein, The analysis system also has a display control unit that displays the multiple candidate codes included in the selection list in a manner that the earlier the code appears, the higher it is placed.

3. The analysis system according to claim 2, wherein, The display control unit only displays the candidate code that is first in the order of the candidate codes in the selection list. When a selection operation is subsequently processed, the candidate codes in the selection list are displayed in such a way that the earlier the order, the higher up they are.

4. The analysis system according to claim 2 or 3, wherein, The correlation calculation unit matches the object words extracted by the extraction unit and the correlation degree of those object words with the candidate code with the highest correlation degree among the multiple candidate codes. When creating the selection list, the list section arranges the multiple candidate codes in a manner that the higher the relevance of the corresponding object's terminology and the higher the importance of the corresponding treatment, the higher their priority.

5. The analysis system according to any one of claims 2 to 4, wherein, The analysis system also has: The history production department produces a fault response history, which is information obtained by assigning candidate codes selected from the selection list as the judgment codes to the fault response record; as well as The Performance Learning Department, from multiple fault response histories previously produced by the History Production Department, determines the following actions, which are associated actions corresponding to the assigned judgment codes. The Performance Learning Department calculates the number of times each action becomes an associated action in the multiple fault response histories. The importance calculation unit uses the performance importance, which increases the more times the related action is applied, to calculate the importance of the action.

6. The analysis system according to claim 5, wherein, The importance calculation unit calculates the weighted sum of the base importance set for each action and the actual performance importance calculated for each action as the importance of the action.

7. The analysis system according to any one of claims 2 to 6, wherein, The analysis coefficients also include a priority calculation unit, which calculates the priority of the object words extracted by the extraction unit. When creating the selection list, the list section arranges the multiple candidate codes in a manner that the higher the relevance of the corresponding object's terminology, the higher the importance of the corresponding action, and the higher the corresponding priority, the earlier they are ranked.

8. The analysis system according to claim 7, wherein, When creating the selection list, the list unit calculates an index value for each of the plurality of candidate codes. The index value includes a weighted sum of the relevance of the corresponding object term, the importance of the corresponding treatment, and the priority of the corresponding object term. The list unit arranges the plurality of candidate codes in such a way that the more favorable the index value, the higher its ranking.

9. The analysis system according to claim 7 or 8, wherein, The fault response record includes a first record column and a second record column that can record free statements. The priority calculation unit calculates the real-time priority based on whether the object term extracted by the extraction unit is contained in the first or second record column, and calculates the weighted sum of the real-time priority and the basic priority set according to each object term, as the priority of the object term extracted by the extraction unit.

10. The analysis system according to claim 9, wherein, After displaying the plurality of candidate codes, the display control unit accepts input of rearrangement conditions, so that the plurality of candidate codes in the selection list rearranged by the list unit according to the rearrangement conditions are displayed in a manner in which the earlier the order is, the higher the position.

11. The analysis system according to claim 10, wherein, The display control unit receives input as a condition for changing the weighted ratio of the base priority to the real-time priority, which is used as the rearrangement condition. The list section calculates the priority of the object words according to the weighted ratio shown by the rearrangement conditions, and rearranges the multiple candidate codes according to the calculated priority, so that the higher the relevance of the corresponding object words, the higher the importance of the corresponding disposal, and the higher the corresponding priority, the earlier they are placed.

12. The analysis system according to any one of claims 2 to 11, wherein, The display control unit accepts the input of sorting conditions, and displays the candidate codes in the selection list after deleting candidate codes that do not meet the sorting conditions, in a manner where the earlier the candidate code appears, the higher it is displayed.

13. The analysis system according to any one of claims 2 to 12, wherein, The analysis system also has: The history production department produces a fault response history, which is information obtained by assigning candidate codes selected from the selection list as the judgment codes to the fault response record; as well as The response learning department calculates a response index value for each of the plurality of candidate codes, based on at least one of the number of times assigned to a plurality of fault response histories previously created by the history production department and the number of times corresponding to error codes contained in the plurality of fault response histories. If, in the selection list, there are two or more candidate codes with the same ranking determined by information including the relevance and importance, the list will arrange the two or more candidate codes with the same ranking in such a way that the higher the response indicator value, the higher the ranking.

14. The analysis system according to claim 13, wherein, The list section arranges the candidate codes containing the corresponding object terms as strings in a manner that prioritizes candidate codes that are not among the candidate codes containing the corresponding object terms. Then, it further arranges the candidate codes containing the corresponding object terms in a manner that prioritizes them based on the degree of relevance of the corresponding object terms and the importance of the corresponding processing.

15. The analysis system according to any one of claims 2 to 14, wherein, The extraction unit extracts the names of the devices that constitute the apparatus. Each of the candidate codes is a statistical device name selected for statistical purposes. The correlation degree is the edit distance between the device name and the statistical device name.

16. An estimation system that analyzes fault response records containing free statements input by a person who performed a fault response to the device, wherein, This estimation system has the following characteristics: The extraction unit extracts specific object terms and words indicating the actions taken during fault response from the free statements contained in the fault response record; The correlation calculation unit calculates the correlation between each of the multiple candidate codes that become the judgment codes assigned to the fault response record and the object words extracted by the extraction unit. An importance calculation unit calculates the importance of the processing extracted by the extraction unit and assigns it to any one of the plurality of candidate codes; The list section generates a selection list based on the relevance calculated by the relevance calculation section and the importance calculated by the importance calculation section, which arranges the multiple candidate codes in a manner that the higher the probability of being selected as the judgment code, the higher the ranking. as well as The estimation unit estimates the candidate code that ranks first in the selection list created by the list unit as the judgment code.

Images