Information processing device, information processing method, and program
By acquiring and filtering search queries and meta-information of facility inspections in the information processing device, the reliability of search results is calculated, and the problem of non-professional personnel obtaining low-precision search results in facility inspections is solved, and the accuracy of facility inspections is improved. efficiency.
Patent Information
- Application Number
- CN202380034447.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-04-18
- Filing Date
- 2023-03-24
- Publication Date
- 2024-11-26
AI Technical Summary
Due to increased costs and insufficient labor, non-professional personnel obtain low accuracy in search results when performing facility inspections, making it difficult to effectively utilize them, especially the search results contain content that is less relevant to the search object.
By obtaining search queries and meta information in the information processing device, filtering the database using the processor to calculate the reliability of the search results, including using methods such as cosine similarity and Jaccard coefficients to improve the reliability of the search results.
Effectively improve the reliability of search results, enable non-professional personnel to make more use of search results, and improve the accuracy and efficiency of facility inspections.
Smart Images

Figure CN119032353A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an information processing device, an information processing method and a program. Background Art
[0002] Conventionally, facility managers have outsourced inspection and other tasks to building consultants, and inspectors with specialized knowledge have conducted inspections of various structures (also called buildings, structures, or infrastructure) such as bridges, roads, tunnels, dams, and buildings.
[0003] In patent document 1, a first inspection result is obtained, a second inspection result is searched from a database based on image features included in the first inspection result, a specific inspection result corresponding to the building condition is searched from the searched second inspection result, and the specific inspection result is preferentially displayed on a display unit.
[0004] Previous technical literature
[0005] Patent Literature
[0006] Patent Document 1: International Publication No. 2017 / 056804 Summary of the invention
[0007] Technical issues to be solved by the invention
[0008] In recent years, due to the increase in costs and the shortage of labor, the demand for non-professional inspections is increasing. However, the search accuracy of the search results obtained by executing the search query is low, and the search results sometimes include content with low relevance to the search object. Therefore, it is not easy for non-professionals to use the search results.
[0009] The present invention has been made in view of such circumstances, and an object of the present invention is to provide an information processing device, an information processing method, and a program that can effectively utilize search results by obtaining the reliability of the search results.
[0010] Means for solving technical problems
[0011] In the first method, in an information processing device equipped with a processor, the processor performs the following processing: obtains a search query; obtains a first search result of a database based on the search query; obtains meta-information of the search query; obtains a second search result of a database filtered by the meta-information based on the search query; and calculates the reliability of the first search result based on the second search result.
[0012] In the information processing device of the second aspect, the processor acquires the second search result of the database filtered step by step using the meta information based on the search query.
[0013] In the information processing device according to the third aspect, the processor calculates the similarity when obtaining the reliability.
[0014] In the information processing device of the fourth mode, when calculating the reliability, the processor calculates the similarity according to a method selected from the group consisting of cosine similarity, deviation pattern similarity, Jaccard coefficient, Dyss coefficient, Simpson coefficient, Pearson correlation coefficient, Spearman correlation coefficient, bulldozer distance, Euclidean distance, weighted Euclidean distance, Hamming distance, Mahalanobis distance or Canberra distance.
[0015] In the information processing device of the fifth aspect, when obtaining the reliability, the processor obtains statistical information for the first search result or the second search result based on the first search result or the second search result, and determines whether the statistical information is within a threshold range.
[0016] In the information processing device according to the sixth aspect, the processor outputs the reliability.
[0017] In the information processing device according to the seventh aspect, the processor feeds back the reliability to the acquisition of the first search result.
[0018] In the information processing device according to the eighth aspect, the meta-information includes information on the structure.
[0019] In the information processing device of the ninth aspect, the information related to the structure of the meta-information includes at least one selected from the group consisting of a damaged image, specifications, damage information, repair information, peripheral information, weather information, and statistical information.
[0020] In the information processing device according to the tenth aspect, the search query includes information related to the structure.
[0021] In the information processing device of the eleventh aspect, the information related to the structure to be searched includes at least one selected from the group consisting of damage images, specifications, damage information, repair information, surrounding information, weather information, and statistical information.
[0022] In the 12th mode, the information processing method is an information processing method executed by a processor, which includes the following steps: obtaining a search query; obtaining a first search result of a database based on the search query; obtaining meta-information of the search query; obtaining a second search result of a database filtered by the meta-information based on the search query; and calculating the reliability of the first search result based on the second search result.
[0023] The program of the 13th method is a program for executing an information processing method executed by a processor, which enables the processor to execute the following steps: obtain a search query; obtain a first search result of a database based on the search query; obtain meta-information of the search query; obtain a second search result of a database filtered by the meta-information based on the search query; and based on the second search result, calculate the reliability of the first search result.
[0024] Effects of the Invention
[0025] According to the present invention, by obtaining the reliability of the search result, the search result can be effectively utilized. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 This is a schematic diagram of an information processing device.
[0027] Figure 2 This is a block diagram showing an example of the hardware configuration of the information processing device.
[0028] Figure 3 This is a block diagram showing the processing functions implemented by the CPU.
[0029] Figure 4 It is a diagram showing a database stored in a storage unit.
[0030] Figure 5 It is a flowchart showing the information processing method.
[0031] Figure 6 This is a diagram for explaining an example of a display screen displaying a search query.
[0032] Figure 7 This is a diagram for explaining an example of a display screen displaying a first search result.
[0033] Figure 8 This is a diagram for explaining an example of a display screen displaying meta-information.
[0034] Fig. 9 FIG. 2 is a diagram for explaining an example of a display screen displaying a second search result.
[0035] Fig.10 This is a diagram for explaining an example of a method for obtaining reliability.
[0036] Fig.11 This is a diagram for explaining an example of a display screen that displays reliability.
[0037] Fig.12 This is a flowchart showing the first embodiment.
[0038] Fig.13 This is a flowchart showing the second embodiment. DETAILED DESCRIPTION
[0039] Hereinafter, preferred embodiments of the information processing device, information processing method and program according to the present invention will be described with reference to the accompanying drawings. In this specification, "structure" includes buildings, such as civil engineering structures such as bridges, tunnels, and dams, and also includes buildings such as walls, columns, and beams of buildings, houses, and structures.
[0040] Figure 1 1 is a schematic diagram of the information processing device 10. The information processing device 10 is connected to the operation unit 18 and the display device 30. Figure 1 As shown, the user obtains the inspection result of the structure 1 via the camera 3 or the mobile terminal 5 .
[0041] As the mobile terminal 5 , for example, various mobile terminals having a photographing function and an information inputting function, such as a smartphone, a tablet terminal, and a portable personal computer, are preferably used.
[0042] The user inputs a search query based on the inspection result into the information processing device 10 via the operation unit 18. The information processing device 10 acquires the input search query. The information processing device 10 acquires the first search result of the database 40 according to the search query. The database 40 includes, for example, past inspection results on structures.
[0043] The information processing device 10 acquires meta information of the search query, and acquires second search results based on the search query for the database 40 filtered by the meta information.
[0044] Next, the information processing device 10 obtains the reliability of the first search result based on the second search result. Furthermore, the information processing device 10 displays the first search result and the reliability on the display device 30.
[0045] Hereinafter, this embodiment will be described in detail.
[0046] [Hardware structure of information processing device]
[0047] Figure 2 This is a block diagram showing an example of the hardware configuration of the information processing device according to the present embodiment.
[0048] As Figure 2The information processing device 10 shown can be a computer or a workstation. The information processing device 10 mainly includes an input / output interface 12, a storage unit 16, an operation unit 18, a CPU (Central Processing Unit) 20, a RAM (Random Access Memory) 22, a ROM (Read Only Memory) 24, and a display control unit 26. A display device 30 constituting a display is connected to the information processing device 10. Under the instruction of the CPU 20, the display device 30 displays various information through the control of the display control unit 26.
[0049] The input / output interface 12 can input various data (information) to the information processing device 10. For example, data stored in the storage unit 16 is input via the input / output interface 12.
[0050] The CPU (processor) 20 reads various programs stored in the storage unit 16 or the ROM 24 , etc., and develops the programs in the RAM 22 to perform calculations, thereby executing various processes of the information processing device 10 .
[0051] Figure 3 It is a block diagram showing the processing functions implemented by the CPU 20 .
[0052] The CPU 20 mainly includes a search query acquisition unit 51, a first search result acquisition unit 52, a meta-information acquisition unit 53, a filter unit 54, a second search result acquisition unit 55, a reliability calculation unit 56, and an output unit 57, and executes processing in each unit. The processing function of each unit will be described later.
[0053] Return to Figure 2 The storage unit (memory) 16 is composed of a non-temporary storage medium such as a hard disk device, various semiconductor memories, and a control unit thereof. The storage unit 16 stores programs that operate the information processing device 10, such as an operating system and a program for executing an information processing method. In addition, the storage unit 16 stores information used in the embodiments described below.
[0054] The operation unit 18 includes a keyboard and a mouse, and the user can cause the information processing device 10 to perform a desired process via these devices. By using a touch panel type device, the display device 30 and the operation unit 18 can be used as both.
[0055] The display device 30 is a device such as a liquid crystal display, and displays various information from the information processing device 10 .
[0056] Figure 41 is a diagram for explaining the database 40 stored in the storage unit 16. In the embodiment, the database 40 includes inspection data 41 which is an inspection result of a structure acquired by a past inspection.
[0057] like Figure 4 As shown, the inspection data 41 includes information of the structure, for example, at least one selected from the group consisting of "specifications", "damage information", "repair information", "surrounding information", "weather information", "statistical information" and "damage image". "Specifications", "damage information", "repair information", "surrounding information", "weather information" and "statistical information" are text data 42, and "damage image" is image data 43. In addition, the inspection data 41 may also include text data such as "inspection date and time", "shooting date and time", and "repair date and time".
[0058] The "specification" includes at least one of the number of years, the structural form, the name of the component or the material.
[0059] The number of years that have passed can be exemplified by the number of years that have passed since the completion date or the start date of use.
[0060] In the case of a bridge, the structural form may be a beam bridge, a rigid frame bridge, a truss bridge, an arch bridge, a cable-stayed bridge, or a suspension bridge.
[0061] In the case of a bridge, the component name may include, for example, a bridge deck, a pier, an abutment, or a beam.
[0062] · Examples of the material include steel, reinforced concrete, and PC (Prestressed Concrete).
[0063] The "damage information" includes at least one of the damage type, damage degree, healthiness, or countermeasure classification.
[0064] The damage types include, for example, cracks (fissures), water leakage, corrosion, breakage, or peeling that occur in the structure.
[0065] The damage degree is information showing the objective state of the damage according to the damage type, and is classified according to size, depth and type, for example, from a to d.
[0066] The health index can be exemplified by an index indicating the diagnosis results of four stages I to IV described in the inspection guidelines prescribed by the Ministry of Land, Infrastructure, Transport and Tourism.
[0067] · Countermeasure classification The countermeasure classification described in the inspection guidelines prescribed by the Ministry of Land, Infrastructure, Transport and Tourism can be exemplified.
[0068] The “repair information” can exemplify, for example, past repair contents.
[0069] The “surrounding information” can include, for example, traffic volume (daily, monthly, annually, cumulatively, etc.) or location (distance from the sea).
[0070] Examples of “weather information” include average temperature, average humidity, rainfall, snowfall, and the like.
[0071] The “statistical information” may include, for example, the type of damage or the ratio of the size of damage.
[0072] exist Figure 4 In the above description, the case where the database 40 is stored in the storage unit 16 is described. However, if the information processing device 10 can access the database 40 via a wired or wireless network, the database 40 may be stored in an external storage device.
[0073] <Information processing method>
[0074] Figure 5 1 is a flowchart showing an information processing method using the information processing device 10. Figure 5 As shown, as an example, the information processing method includes the following steps: obtaining a search query (step S1); obtaining the first search result (step S2); obtaining meta-information of the search query (step S3); filtering the database according to the meta-information (step S4); obtaining the second search result (step S5); calculating the reliability of the first search result based on the second search result (step S6); and outputting the reliability (step S7).
[0075] <Steps to obtain search query>
[0076] In the step of acquiring a search query (step S1), the search query acquisition unit 51 acquires the search query. The user inputs the search query into the information processing device 10 via the operation unit 18, for example. At this time, the search query is created by the user based on the inspection result of the structure 1. Then, the search query acquisition unit 51 acquires the search query. In addition, as another method, the search query can also be created by the user using the mobile terminal 5. In addition, as another method, the search query can also be automatically created based on the inspection result. Here, the search query includes information for determining the search object and is a processing request to the information processing device 10.
[0077] The search query may include, for example, text data and / or image data. In the case where the structure 1 is the object, the search query includes at least one selected from the group consisting of "specifications", "damage information", "repair information", "surrounding information", "weather information", "statistical information" and "damage image". "Specifications", "damage information", "repair information", "surrounding information", "weather information" and "statistical information" are examples of text data, and "damage image" is an example of image data. The search query is not limited to the above-mentioned text data and / or image data.
[0078] In addition, the image data and text data included in the search query are preferably of the same type as the image data 43 and text data 42 included in the inspection data 41 .
[0079] Figure 6 This is an example of a display screen showing a search query. Figure 6 The display screen 100 shown is displayed on the display device 30 under the control of the display control unit 26. The display screen 100 includes an input display screen 102 and a result display screen 104.
[0080] The input display screen 102 displays the search query Qu acquired by the search query acquisition unit 51, and the characters of "search query" are displayed in the category area 106. In the case where the search query Qu includes image data, the image data ID is displayed on the input display screen 102, and the user can confirm the image data ID. In the case where the search query Qu includes text data, the text data TD is displayed on the input display screen 102, and the user can confirm the text data TD. The user can input the search query Qu (image data ID and / or text data TD) from the input display screen 102, and the input search query Qu is acquired by the search query acquisition unit 51.
[0081] Figure 6 The input display screen 102 shown in FIG. 1 can display or input meta information described later. In the area where meta information is displayed or input, the characters "meta information" are displayed in the category area 107.
[0082] The input display screen 102 includes an execution button 108. If the execution button 108 is operated, the next step is entered. Figure 6 In the process, the process proceeds to step S2 of obtaining the first search result.
[0083] <Steps to obtain the first search result>
[0084] In the step of acquiring the first search result (step S2), the first search result acquisition unit 52 acquires the first search result of the database 40 according to the search query Qu. Figure 7As shown, the first search result acquisition unit 52 acquires, for example, inspection data 41 of the database 40 based on the search query Qu. The acquired inspection data 41 is an example of the first search result. Figure 7 In FIG. 1 , a search query Qu including text data TD and image data ID is displayed. The search query Qu only needs to include at least one of the text data TD and the image data ID.
[0085] When the search query Qu includes the image data ID, the first search result acquisition unit 52 acquires the image data 43 from the database 40 according to the image data ID. The first search result acquisition unit 52 calculates the feature amount of the image data ID and the image data 43 by an image recognition algorithm or a machine learning model. Next, the first search result acquisition unit 52 compares the feature amount of the image data ID with the feature amount of the image data 43, and acquires the image data 43 corresponding to the image data ID as the first search result according to the search query Qu.
[0086] When the search query Qu includes text data TD, the first search result acquisition unit 52 acquires text data 42 from the database 40 according to the text data TD. The first search result acquisition unit 52 compares the text data TD with the text data 42 by a text search algorithm or a machine learning model, and acquires the text data 42 corresponding to the text data TD of the search query Qu as the first search result according to the search query Qu.
[0087] Furthermore, the term "performed according to a search query" in this specification means that the database 40 is processed in some way using the information of the search query, such as including a search. The search includes concepts such as "consistent", "similar", "dissimilar" and "evaluation" (for example, "from near to far" or "from far to near").
[0088] like Figure 7 As shown, the first search result obtained by the first search result acquisition unit 52 is displayed on the result display screen 104 under the control of the display control unit 26. When the search query Qu is an image data ID, the image data 43 corresponding to the image data ID is displayed on the result display screen 104. In addition, the text data 42 associated with the image data 43 on the database 40 is also displayed.
[0089] When the search query Qu is text data TD, text data 42 corresponding to the text data TD is displayed on the result display screen 104. In addition, image data 43 associated with the text data 42 on the database 40 is also displayed.
[0090] Figure 7In the figure, the image data ID and the text data TD are shown as the search query Qu, but as mentioned above, the search query Qu only needs to include at least one of the image data ID and the text data TD. In addition, the first search result acquisition unit 52 can also acquire the first search result based on the search query composed of the image data ID and the text data TD.
[0091] In addition, the acquisition of the first search result can apply, for example, the technology of International Publication No. 2020 / 071216, International Publication No. 2020 / 255227, Japanese Patent Application Publication No. 2018-165926, or Japanese Patent Application Publication No. 2017-167987.
[0092] Next, the processing flow enters step S3 of obtaining meta information of the search query Qu.
[0093] <Steps for obtaining meta information of search query>
[0094] In the step of acquiring meta information of the search query Qu (step S3 ), the meta information acquisition section 53 acquires meta information of the search query Qu. Figure 8 FIG. 1 shows an example of a display screen 100 displaying meta information Me. The acquired meta information Me is displayed on an input display screen 102. Figure 8 In the process, the first search result obtained in step S2 is displayed on the result display screen 104.
[0095] Similar to the search query Qu, the meta-information Me may include text data TD and / or image data ID, for example, at least one selected from the group consisting of "specifications", "damage information", "repair information", "peripheral information", "weather information", "statistical information" and "damaged image". "Specifications", "damage information", "repair information", "peripheral information", "weather information" and "statistical information" are examples of text data TD, and "damaged image" is an example of image data ID. The search query Qu and meta-information Me may include common types of image data ID and text data TD. The information included in the meta-information Me is not limited. Figure 8 , meta information Me including text data TD and image data ID is displayed. However, the meta information Me only needs to include either the text data TD or the image data ID.
[0096] Here, the meta information Me is information related to the search query Qu, not the search query Qu itself.
[0097] Next, the combination of the search query Qu and the meta-information Me is shown. Table 1 shows an example of the combination of the search query Qu and the meta-information Me. When the object is a structure, the following combinations can be exemplified. No. 1 shows the case where the search query Qu is image data (damaged image) and the meta-information Me is text data (specifications, damage information, repair information, statistical information, etc.). No. 2 shows the case where the search query Qu is text data (specifications, damage information, repair information, statistical information, etc.) and the meta-information Me is image data (damaged image). No. 3 represents the case where the search query Qu is image data (damaged image) and the meta-information Me is image data (damaged image). In No. 3, when the search query Qu and the meta-information Me respectively include a damaged image as image data, as the damaged image of the meta-information Me, a damaged image different from the search query Qu is applied. As the damaged image of the meta-information Me, an image completely different from the search query Qu, a past image, a slightly processed image, an image shot remotely, etc. can be exemplified.
[0098] Table 1 is an example of a combination of a search query Qu and meta information Me, but the present invention is not limited thereto.
[0099] [Table 1]
[0100]
[0101] The meta-information acquisition unit 53 can automatically acquire the meta-information Me of the search query Qu. For example, when the search query Qu is image data, Exif (Exchangeable Image File Format) is automatically acquired by the meta-information acquisition unit 53 as the meta-information Me. Exif is information attached to the image data when shooting, including information such as the conditions and settings when shooting. Furthermore, when the image data of the search query Qu is shot in a GPS reception environment, the meta-information Me may include location information, latitude and longitude, or altitude, etc.
[0102] Furthermore, when the image data of the search query Qu is a "damaged image", the meta information Me can be acquired using a learning model that has been machine-learned. For example, the "damaged information" can be identified based on the "damaged image" of the search query Qu and through the learning model. The identified "damage information" is acquired as the meta information Me by the meta information acquisition unit 53.
[0103] The meta information Me can be manually input into the information processing device 10 by a user. Figure 8 This is an example of a display screen showing an input example of meta-information Me. Figure 8The display screen 100 can confirm the acquired meta-information Me and input the meta-information Me. That is, the user can manually input the image data ID and / or text data TD as the meta-information Me into the input display screen 102 of the display screen 100. The input meta-information Me is acquired by the meta-information acquisition unit 53. Figure 8 2 shows a case where image data ID and text data TD are input as meta information Me, and the image data ID and text data TD are displayed. The user can input either text data TD or image data ID as meta information Me.
[0104] When the user manually inputs meta information Me, the search query Qu is associated with the meta information Me. For example, when inputting meta information Me, the user displays the search query Qu on the input display screen 102 and performs an association process on the meta information Me, thereby establishing a corresponding association between the search query Qu and the meta information Me. The user can arbitrarily establish a corresponding association between the search query Qu and the meta information Me.
[0105] In addition, even when the meta information Me is automatically acquired, the input display screen 102 can display the meta information Me under the control of the display control unit 26 , and the user can confirm the meta information Me.
[0106] When the execution button 108 of the input display screen 102 is operated, the process flow proceeds to step S4 of filtering the database 40. In addition, after step S4, the process flow can proceed to step S5 of acquiring the second search result.
[0107] <Steps to filter the database>
[0108] In the step of filtering the database (step S4 ), the filtering unit 54 performs filtering on the database 40 based on the meta-information Me.
[0109] The filter section 54 filters the database 40 according to the image data ID or the text data TD included in the meta information Me. The filter section 54 compares the text data TD or the image data ID of the meta information Me with the text data 42 or the image data 43 of the database 40 .
[0110] When the meta information Me is text data TD, the filter unit 54 can filter the database 40 by, for example, extracting only the text data 42 that is the same as the text data TD of the meta information Me or other text data.
[0111] Furthermore, the filter unit 54 can filter the database 40 based on the analysis result of the meta-information Me. For example, when the meta-information Me is the damage type, the filter unit 54 calculates the ratio of the damage type as the analysis result, and compares it with the statistical information of the damage type ratio in the database 40, thereby filtering the database 40. Furthermore, when the meta-information is the size of the damage type, the filter unit 54 calculates the ratio of the size of the damage type as the analysis result, and compares it with the statistical information of the size ratio of the damage type in the database 40, thereby filtering the database 40. These filtering methods are examples of filtering.
[0112] Furthermore, the filter unit 54 may perform step-by-step filtering on the database 40. For example, when the meta information Me includes a plurality of text data TD, the filter unit 54 may perform filtering by "elapsed years" and further by "component name." The step-by-step filtering is not particularly limited.
[0113] In addition, as a result of filtering, there are cases where the parent parameters of the database 40 are narrowed down or cases where they are not narrowed down.
[0114] The processing flow proceeds to step S5 of acquiring the second search result.
[0115] <Steps to obtain the second search result>
[0116] In the step of acquiring the second search result (step S5), the second search result acquisition unit 55 acquires the second search result of the filtered database 40 according to the search query. The second search result acquisition unit 55 acquires the inspection data 41 according to the search query from the filtered database 40, for example.
[0117] The difference between the acquisition of the second search result by the second search result acquisition unit 55 based on the search query and the acquisition of the first search result by the first search result acquisition unit 52 based on the search query is that the second search result acquisition unit 55 uses the filtered database 40 as an object, while the first search result acquisition unit 52 uses the database 40 before filtering as an object. Except for this point, the second search result acquisition unit 55 acquires the second search result in the same way as the first search result acquisition unit 52.
[0118] Fig. 9 This is an example of displaying the second search result on the result display screen 104. Fig. 9 As shown, the second search result obtained by the second search result acquisition unit 55 is displayed on the result display screen 104 under the control of the display control unit 26 . Fig. 9 An example of the second search result in the present invention is shown. Fig. 9 The second search result shown and Figure 7Compared with the first search result shown, the displayed results are different because the database 40 has been filtered.
[0119] Next, when the execution button 108 is operated, the process flow proceeds to step S6 of obtaining the reliability, and the reliability calculation unit 56 obtains the reliability of the first search result.
[0120] Fig. 9 Although the second search result is shown to be displayed on the display screen 100, the second search result may not be displayed on the display screen 100. That is, the processing flow may proceed to step S6 of executing reliability processing for determining the reliability of the first search result after step S5 of acquiring the second search result.
[0121] <Steps for determining the reliability of the first search result>
[0122] In the step of obtaining the reliability of the first search result (step S6), the reliability calculation unit 56 obtains the reliability of the first search result based on the second search result. By obtaining the reliability of the first search result, the accuracy of the first search result can be improved.
[0123] A preferred method for obtaining the reliability will be described below. The reliability calculation unit 56 determines the application conditions of the first search result and the application conditions of the second search result to obtain the reliability. The reliability calculation unit 56 can obtain the reliability from each application condition.
[0124] For example, the reliability calculation unit 56 determines the index of the database 40 for the first search result and the second search result (so-called similarity ranking) as an applicable condition for obtaining the reliability. Fig.10 This is a diagram conceptually illustrating a case where an index is used as an object. Fig.10 10A shows an example of the first search result, where four data A, B, C, and D are obtained from the database 40 as the first search result. The four data A to D are numerically represented by the ranking of similarity to the search query. The smaller the number, the more similar it is to the search query. That is, the first search result indicates that the order of D, A, B, and C is similar to the search query.
[0125] Fig.10 10B shows an example of the second search result. In the second search result, two data A and D are obtained from the filtered database 40. The two data A and D are numerically represented by the ranking of the similarity with respect to the search query. That is, it means that in the second search result, the order of D and A is similar with respect to the search query.
[0126] The reliability calculation unit 56 calculates the reliability of the first search result (including the index) based on the second search result (including the index). Fig.10In the example of , the order of D and A in the second search result is consistent with the order of D and A in the first search result, and the reliability calculation unit 56 calculates the reliability of the first search result as 100%.
[0127] Furthermore, as another viewpoint, the reliability calculation unit 56 can use the first N (N is a natural number) results of the first search result and the first N (N is a natural number) results of the second search result as applicable conditions. In this case, meta information attached to the search result can also be included. The reliability calculation unit 56 can calculate the reliability of the first search result (including the first N results) based on the second search result (including the first N results).
[0128] The reliability calculation unit 56 can calculate the similarity or distance in the process of obtaining the reliability. The reliability calculation unit 56 can obtain the similarity or distance by applying a method selected from the group consisting of cosine similarity, deviation pattern similarity, Jaccard coefficient, Dyss coefficient, Simpson coefficient, Pearson correlation coefficient, Spearman correlation coefficient, bulldozer distance, Euclidean distance, weighted Euclidean distance, Hamming distance, Mahalanobis distance, and Canberra distance. The method shown here is a known technique, so the description is omitted.
[0129] Furthermore, when the first search result (including the first N results) is used as an applicable condition based on the second search result (including the first N results), the reliability calculation unit 56 obtains statistical information for the first search result or the second search result based on the first search result or the second search result, and determines whether the statistical information is within the threshold range, thereby being able to calculate the reliability.
[0130] The reliability calculation unit 56 calculates as statistical information the ratio of specific information included in the first N positions of the second search result to that included in the first N positions of the first search result. The reliability calculation unit 56 determines whether the ratio is within a preset threshold value and finally obtains the reliability.
[0131] The reliability calculation unit 56 calculates as statistical information the ratio of specific information included in the first N search results of the first search results to that included in the first N search results of the second search results. The reliability calculation unit 56 determines whether the ratio is within a preset threshold value and finally obtains the reliability.
[0132] When the reliability is obtained by the reliability calculation unit 56, the processing flow proceeds to step S7 where the reliability is output.
[0133] <Steps for outputting reliability>
[0134] In the step of outputting the reliability (step S7 ), the output unit 57 outputs the reliability to the display device 30 via the display control unit 26 . Fig.11This is an example of the display screen after the reliability output. Fig.11 As shown, the reliability is displayed in a column of the result display screen 104. Fig.11 In the display device 30, the reliability and the first search result are displayed simultaneously. Since the reliability for the first search result is displayed, the user can refer to the reliability to determine whether the past inspection result can be used from the first search result.
[0135] The case where the output unit 57 displays the reliability on the display device 30 is illustrated, but the output unit 57 can output the reliability to a printing device and can output it as electronic data in various data formats.
[0136] When the output unit 57 outputs the reliability, the processing flow ends.
[0137] <Preferred Embodiment>
[0138] Hereinafter, preferred embodiments will be described.
[0139] (Implementation Method 1)
[0140] In the first embodiment, the reliability is compared with a preset threshold value X, the search query is changed, and the processing flow is executed until the reliability satisfies the threshold value X.
[0141] exist Fig.11 In the example of , the display screen 100 displays the reliability as 80%. The user can judge that the reliability of the first search result is high. On the other hand, it is also possible to consider a case where the reliability calculation unit 56 calculates the reliability of the first search result as 20%.
[0142] Therefore, the reliability calculation unit 56 determines whether the reliability is equal to or greater than a preset threshold value X, and can execute a processing flow corresponding to the reliability.
[0143] Fig.12 is a flowchart of implementation mode 1, Figure 5 The reliability determination step S6A is added to the process. Figure 5 The differences are explained below.
[0144] As described above, the CPU 20 of the information processing device 10 executes steps S1 to S6. When the reliability calculation unit 56 obtains the reliability, the processing flow enters the determination step (step S6A). In step S6A, it is determined whether the reliability is greater than a preset threshold value X.
[0145] When it is determined that the reliability is less than the threshold value X, that is, when it is "No" in step S6A, the process proceeds to the step of acquiring the search query (step S1). In the step of acquiring the search query (step S1), the search query for which the reliability of the first search result does not satisfy the threshold value X is changed. The changed search query is acquired by the search query acquisition unit 51, and steps S1 to S6A are repeated until the reliability is determined to be above the threshold value X.
[0146] When the search query is image data, the change of the search query can be exemplified by changing it to image data taken from a different angle. The change of the search query can be performed automatically or manually. However, the change of the search query is not limited to this example.
[0147] When the reliability is determined to be equal to or greater than the threshold value X, that is, when the answer is “Yes” in step S6A, the process proceeds to a step of outputting the reliability (step S7 ), and when the output unit 57 outputs the reliability, the process flow ends.
[0148] In the first embodiment, the first search result having a reliability greater than or equal to the threshold value X is output, and thus the user can use the past inspection results based on the first search result by referring to the reliability.
[0149] (Implementation Method 2)
[0150] In the second embodiment, by feeding back the reliability result to the acquisition of the first search result, the accuracy of the first search result can be improved.
[0151] Fig.13 This is a flowchart of the second embodiment, showing the process of adding step S6B of feeding back the reliability result to the acquisition of the first search result. Figure 5 The differences are explained below.
[0152] As described above, the CPU 20 of the information processing device 10 executes steps S1 to S6. When the reliability calculation unit 56 obtains the reliability, the processing flow proceeds to a step of feeding back the reliability result to the acquisition of the first search result (step S6B).
[0153] When the first search result acquisition unit 52 uses a machine learning model, the first search result and the second search result can also be used as teacher data to optimize the parameters of the machine learning model through machine learning. By using the optimized machine learning model, the accuracy of the first search result acquired by the first search result acquisition unit 52 is improved.
[0154] The method of feeding back the reliability result to the acquisition of the first search result is not limited to the above method.
[0155] <Other>
[0156] In the above-mentioned embodiment, the hardware structure of the processing unit that performs various processes is various processors as shown below. The various processors include general-purpose processors that execute software (programs) and function as various processing units, namely CPU (Central Processing Unit), FPGA (Field Programmable Gate Array) and other processors whose circuit structure can be changed after manufacturing, namely programmable logic devices (PLD), ASIC (Application Specific Integrated Circuit) and other processors with circuit structures specially designed for performing specified processes, namely dedicated circuits, etc.
[0157] One processing unit may be composed of one of these various processors, or may be composed of two or more processors of the same or different types (for example, a plurality of FPGAs or a combination of a CPU and an FPGA). Furthermore, multiple processing units may be composed of one processor. As an example of multiple processing units composed of one processor, first, as represented by computers such as clients and servers, there is a method in which a processor is composed of a combination of one or more CPUs and software, and the processor functions as multiple processing units. Second, as represented by a system on chip (SoC), there is a method in which a processor is used that implements the functions of the entire system including multiple processing units using one IC (Integrated Circuit) chip. In this way, various processing units are composed of one or more of the above-mentioned various processors as a hardware structure.
[0158] Furthermore, more specifically, the hardware structure of these various processors is a circuit (circuitry) formed by combining circuit elements such as semiconductor elements.
[0159] The above structures and functions can be appropriately implemented by any hardware, software or a combination of the two. For example, the present invention can also be applied to a program that causes a computer to execute the above processing steps (processing sequence), a computer-readable storage medium (non-temporary storage medium) storing such a program, or a computer capable of installing such a program.
[0160] As mentioned above, although the example of this invention was demonstrated, this invention is not limited to the said embodiment, It goes without saying that various deformation|transformation is possible within the range which does not deviate from the summary of this invention.
[0161] Explanation of symbols
[0162] 1-structure, 3-camera, 5-mobile terminal, 10-information processing device, 12-input / output interface, 16-storage unit, 18-operation unit, 20-CPU, 22-RAM, 24-ROM, 26-display control unit, 30-display device, 40-database, 41-inspection data, 42-text data, 43-image data, 51-search query acquisition unit, 52-first search result acquisition unit, 53-meta information acquisition unit, 54-filtering unit, 55-second search result acquisition unit, 56-reliability calculation unit, 57-output unit, 100-display screen, 102-input display screen, 104-result display screen, 106-category area, 107-category area, 108-execution button, ID-image data, TD-text data, Me-meta information, Qu-search query.
Claims
1. An information processing device comprising a processor, The processor performs the following processing: Get the search query; Obtaining a first search result from a database according to the search query; Obtaining meta information of the search query; obtaining a second search result based on the search query for the database filtered using the meta information; and Based on the second search result, the reliability of the first search result is obtained.
2. The information processing device according to claim 1, in, The processor obtains the second search result based on the search query on the database that is gradually filtered using the meta information.
3. The information processing device according to claim 1 or 2, in, The processor calculates similarity when obtaining the reliability.
4. The information processing device according to claim 3, in, When calculating the reliability, the processor calculates the similarity according to a method selected from the group consisting of cosine similarity, deviation pattern similarity, Jaccard coefficient, Dyss coefficient, Simpson coefficient, Pearson correlation coefficient, Spearman correlation coefficient, bulldozer distance, Euclidean distance, weighted Euclidean distance, Hamming distance, Mahalanobis distance or Canberra distance.
5. The information processing device according to claim 1 or 2, in, When obtaining the reliability, the processor obtains statistical information for the first search result or the second search result based on the first search result or the second search result, and determines whether the statistical information is within a threshold range.
6. The information processing device according to claim 1 or 2, in, The processor outputs the reliability.
7. The information processing device according to claim 1 or 2, in, The processor feeds back the reliability to the acquisition of the first search result.
8. The information processing device according to claim 1 or 2, in, The meta information includes information related to the structure.
9. The information processing device according to claim 8, in, The information related to the structure of the meta-information includes at least one selected from the group consisting of damage image, specification, damage information, repair information, surrounding information, weather information and statistical information.
10. The information processing device according to claim 1 or 2, in, The search query includes information related to a structure.
11. The information processing device according to claim 10, in, The information related to the structure of the search query includes at least one selected from the group consisting of damage images, specifications, damage information, repair information, surrounding information, meteorological information and statistical information.
12. An information processing method, which is an information processing method executed by a processor, the information processing method comprising the following steps: Get the search query; Obtaining a first search result from a database according to the search query; Obtaining meta information of the search query; obtaining a second search result based on the search query for the database filtered using the meta information; and Based on the second search result, the reliability of the first search result is obtained.
13. A program for executing an information processing method executed by a processor, the program causing the processor to execute the following steps: Get the search query; Obtaining a first search result from a database according to the search query; Obtaining meta information of the search query; obtaining a second search result based on the search query for the database filtered by the meta information; and Based on the second search result, the reliability of the first search result is obtained. 14 . A recording medium which is non-transitory and computer-readable and has the program according to claim 13 recorded thereon.
