Recommended evaluation methods, systems, devices and storage media for airworthiness cases

Abstract

The present invention provides an airworthiness relevance case recommendation and evaluation method, system, device and storage medium, which obtain the evaluation result of the airworthiness relevance case compliance vector recommendation algorithm by evaluating the compliance case recommendation results under the three dimensions of compliance object, compliance activity and compliance target, evaluating the compliance case recommendation results under the two dimensions of compliance object and compliance activity and evaluating the reasonable value range of the weights of the three dimensions in the recommendation algorithm, thereby ensuring the accuracy of the airworthiness relevance case recommendation.

Description

Technical Field

[0001] The present invention relates to the application of knowledge graphs in the field of airworthiness review, and specifically to an airworthiness case recommendation and evaluation method, system, device and storage medium. Background Art

[0002] The recommendation problem of airworthiness cases is a task in a specific context, which requires solving the problems of knowledge storage and recommendation. Due to the abstractness of airworthiness standards and the flexibility of compliance methods, it is difficult to use traditional rule-based reasoning methods to reuse knowledge centered on review cases. One of the key issues is how to measure the similarity among recommended objects or cases.

[0003] Any conformity review of airworthiness review and the conformity cases generated by it contain attributes of three dimensions: conformity object, conformity target and conformity activity. Therefore, by establishing associations based on the above three dimensions, effective organization of conformity cases in the ontology can be achieved. To construct a conformity case recommendation method based on conformity vector, it is necessary to evaluate the accuracy of its case recommendations in each dimension. Summary of the invention

[0004] The purpose of the present invention is to provide an airworthiness case recommendation evaluation method, system, device and storage medium to provide an evaluation of airworthiness case recommendation results.

[0005] In order to achieve the above object, the present invention provides, on one hand, an airworthiness case recommendation and evaluation method, characterized in that it comprises the following steps:

[0006] Obtain case recommendation results of the case recommendation algorithm based on the compliance vector for the compliance case data set, and extract the compliance activity record with the highest similarity in the recommendation results of each case;

[0007] Based on the comparison of the above recommendation results with other recommendation results, the accuracy of the recommendation is judged.

[0008] Furthermore, the method further comprises the following steps:

[0009] Obtain similarity measures for compliance objects, compliance activities, and compliance goals;

[0010] Based on the similarity measurement of three dimensions, the top-k recommendation algorithm is used to recommend the matching cases;

[0011] The recommendation results are compared with those of the edit distance recommendation method and the CP-CPI recommendation method, and their precision, recall and F1 value are calculated respectively.

[0012] Furthermore, the method further comprises the following steps:

[0013] The calculation formula of the accuracy is:

[0014]

[0015] Among them, P is the precision, TP is the number of accurate hit cases, and FP is the number of false hit cases;

[0016] The calculation formula of the recall rate is:

[0017]

[0018] Among them, R is the recall rate, FN is the missed case;

[0019] The calculation formula of the F1 value is:

[0020]

[0021] Furthermore, the method further comprises the following steps:

[0022] Obtaining similarity measures of compliance objects and compliance activities;

[0023] Recommendation of conforming cases based on the top-k recommendation algorithm;

[0024] Compare the clause information in the recommended compliance case with the sample clauses, and judge the accuracy of the case recommendation through the information distance between the erroneous clauses and the sample clauses.

[0025] Furthermore, it also includes:

[0026] Obtaining similarity measures of compliance objects and compliance activities;

[0027] The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wo of the compliance object and the weight wa of the compliance activity;

[0028] Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wo and wa by analyzing the relationship between the values ​​and the average similarity of the compliance targets;

[0029] Obtain similarity measures between compliance activities and compliance goals;

[0030] The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wa of the compliance activity and the weight wt of the compliance target;

[0031] Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wa and wt by analyzing the relationship between the values ​​and the average similarity of the compliance targets;

[0032] obtaining a similarity measure between the compliance object and the compliance target;

[0033] The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wo of the compliance object and the weight wt of the compliance target;

[0034] Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wo and wt by analyzing the relationship between the values ​​and the average similarity of the compliance targets;

[0035] Combined with the reasonable values ​​of the weights wo, wa and wt in pairs, the reasonable value range of the three weights is comprehensively determined.

[0036] On the other hand, the present invention also provides an airworthiness relevance case recommendation and evaluation system, comprising:

[0037] The recommendation module obtains the case recommendation results of the compliance case data set by the case recommendation algorithm based on the compliance vector, and extracts the compliance activity record with the highest similarity in the recommendation results of each case;

[0038] The comparison module compares the above recommendation results with other recommendation results to determine the accuracy of the recommendation.

[0039] Furthermore, the recommendation module includes:

[0040] A first recommendation module, wherein the first recommendation module obtains similarity metrics of compliance objects, compliance activities, and compliance goals, and obtains recommendation results of compliance vectors based on a top-k recommendation algorithm;

[0041] A second recommendation module, wherein the second recommendation module obtains a similarity measure between the compliance object and the compliance target, and obtains a recommendation result of the compliance vector based on a top-k recommendation algorithm;

[0042] The weighted recommendation module calculates the reasonable values ​​of the weights of any two combinations of compliance objects, compliance activities and compliance goals during the calculation process of the top-k recommendation algorithm.

[0043] Furthermore, the comparison module further includes:

[0044] The first comparison module compares the recommendation result of the first recommendation module with the recommendation results of the edit distance recommendation method and the CP-CPI recommendation method, and calculates their precision, recall and F1 value respectively.

[0045] The second comparison module compares the clause information in the recommended compliance case with the sample clause based on the recommendation result of the second recommendation module, and determines the accuracy of the case recommendation through the information distance between the wrong clause and the sample clause;

[0046] The weight comparison module comprehensively compares and calculates the reasonable value range of compliance objects, compliance activities and compliance targets based on the reasonable values ​​of weights in any two combinations of compliance objects, compliance activities and compliance targets.

[0047] On the other hand, the present invention also provides an airworthiness relevance case recommendation and evaluation system, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0048] On the other hand, the present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the steps of the above method.

[0049] The present invention provides an airworthiness relevance case recommendation and evaluation method, system, device and storage medium, which obtains airworthiness relevance case recommendation and evaluation results by evaluating compliance case recommendation results under three dimensions of compliance object, compliance activity and compliance target, evaluating compliance case recommendation results under two dimensions of compliance object and compliance activity and evaluating the reasonable value range of weights of the three dimensions in the recommendation algorithm, thereby ensuring the accuracy of airworthiness relevance case recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0051] Figure 1 The present invention is a flowchart of an airworthiness relevance case recommendation and evaluation method according to an embodiment of the present invention.

[0052] Figure 2 It is a flow chart of a method for case recommendation based on three dimensions of conformity vectors according to an embodiment of the present invention.

[0053] Figure 3 It is a flow chart of a method for recommending cases based on compliance vectors in two dimensions according to an embodiment of the present invention.

[0054] Figure 4 It is a flow chart of a method for evaluating weights of three dimensions according to an embodiment of the present invention.

[0055] Figure 5 and Figure 6 A schematic diagram of the weight combination comparison effect of an embodiment of the present invention.

[0056] Figure 7 A system framework diagram of an airworthiness relevance case recommendation and evaluation system according to an embodiment of the present invention.

[0057] Figure 8 It is a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0061] The purpose of the present invention is to provide an airworthiness relevance case recommendation evaluation method, system, device and storage medium. The similarity measurement and recommendation algorithm of airworthiness relevance cases are tested in combination with the actual data collected by the airworthiness collaborative certification platform during the certification process to obtain and optimize the accurate evaluation results of airworthiness relevance case recommendation.

[0062] Figure 1 FIG. 1 is a flow chart of a method for recommending and evaluating airworthiness relevance cases according to an embodiment of the present invention. Figure 1 As shown, the airworthiness relevance case recommendation evaluation method of this embodiment includes the following steps:

[0063] S101, obtaining case recommendation results of a case recommendation algorithm based on a compliance vector for a compliance case data set, and extracting the compliance activity record with the highest similarity in the recommendation results of each case.

[0064] S102, judging the accuracy of the recommendation based on comparing the above recommendation result with other recommendation results.

[0065] Specifically, the compliance case dataset uses compliance plan data and expert recommendation data introduced to compare the effectiveness of algorithms.

[0066] For example, in one embodiment, the compliance activity record of a certain model of domestic transport aircraft is used as test data. Each piece of test data contains the attribute information attached to the review document case record. The airworthiness data is obtained by extracting the compliance activity cases of a certain model within a period of time. The airworthiness data covers most of the compliance activity types that need to be carried out by transport aircraft in a typical transport type number review.

[0067] Among them, the compliance case dataset needs to be preprocessed through the following steps:

[0068] 1. Store the samples in the compliance case collection in the form of a knowledge graph, and establish a tag library through the "compliance target" tag in the compliance activity record to calculate and store the repetition in the tag.

[0069] 2. Randomly select multiple records from the above compliance case set to ensure a wide distribution of cases to approach the actual engineering situation.

[0070] like Figure 2 As shown, in one embodiment, the method recommends cases of compliance vectors based on three dimensions, including the following steps:

[0071] S201, obtaining similarity measures of compliance objects, compliance activities, and compliance goals.

[0072] S202, based on the similarity measurement of three dimensions, a top-k recommendation algorithm is used to recommend conforming cases.

[0073] S203, comparing the recommendation result with the recommendation results of the edit distance recommendation method and the CP-CPI recommendation method, and calculating their precision, recall and F1 value respectively.

[0074] Specifically, in the recommendation results of the compliance cases, 3 records are selected for each recommended case to be compared with the recommendation results of the edit distance recommendation method and the CP-CPI recommendation method.

[0075] The calculation formula for accuracy is:

[0076]

[0077] Among them, P is the precision, TP is the number of accurate hit cases, and FP is the number of false hit cases;

[0078] The calculation formula of the recall rate is:

[0079]

[0080] Among them, R is the recall rate, FN is the missed case;

[0081] The calculation formula of the F1 value is:

[0082]

[0083] The following table shows the recommendation results of the conformity vector recommendation algorithm, the edit distance recommendation method, and the CP-CPI recommendation method used in this embodiment.

[0084]

[0085] As can be seen from the above table, the recommendation results of the compliance vector recommendation algorithm used in this embodiment are relatively close to the results recommended by experts. In contrast, although the recommendation results based on the edit distance of the case name can also locate a small number of valid cases, the proportion of incorrect results is very high and basically has no practical value. Taking into account that the compliance plan preparation process already contains a certain amount of manual classification experience information, the accuracy of the relevant case recommendation results based on the compliance plan (CP-CPI) structure has been improved and is in line with the actual airworthiness compliance project. In summary, the case recommendation method based on the compliance vector can more effectively recommend compliance case information with engineering value.

[0086] like Figure 3 As shown, in another embodiment, the method also uses a compliance vector recommendation algorithm based on two dimensions to recommend cases, the purpose of which is to test whether the most likely "regulatory requirement" label can be inferred using the information of the other two dimensions in the absence of the "regulatory requirement" dimension. The method includes the following steps:

[0087] S301, obtaining similarity measures between compliance objects and compliance activities.

[0088] S302, recommending conforming cases based on a top-k recommendation algorithm.

[0089] S303, comparing the clause information in the recommended compliance case with the sample clause, and judging the accuracy of the case recommendation through the information distance between the erroneous clause and the sample clause.

[0090] Among them, the recommendation algorithm is run in the absence of the compliance target dimension. The weights of the object and activity dimensions of the F value in the recommendation process are 0.6 and 0.4 respectively, and the clause information contained in the recommended case is extracted and compared with the sample clauses. The results are shown in the following table:

[0091]

[0092] It can be concluded that in the absence of information on the "compliance target" dimension, information on the two dimensions of "compliance object" and "compliance activity" can more effectively recommend similar compliance cases and infer the terms and conditions information required for the current compliance work.

[0093] like Figure 4 As shown, in another embodiment of the present invention, the method evaluates the reasonable value space of the weight values ​​wo, wa, and wt of the three dimensions in the top-k algorithm recommendation of the compliance vector.

[0094] Specifically, the following steps are included:

[0095] Gets similarity measures for compliance objects and compliance activities.

[0096] The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wo of the compliance object and the weight wa of the compliance activity.

[0097] Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wo and wa by analyzing the relationship between the values ​​and the average similarity of the compliance targets.

[0098] Obtain similarity measures for compliance activities and compliance goals.

[0099] The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wa of the compliance activity and the weight wt of the compliance target.

[0100] Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wa and wt by analyzing the relationship between the values ​​and the average similarity of the compliance targets.

[0101] Gets the similarity measure between the compliance object and the compliance target.

[0102] The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wo of the compliance object and the weight wt of the compliance target.

[0103] Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wo and wt by analyzing the relationship between the values ​​and the average similarity of the compliance targets.

[0104] Combined with the reasonable values ​​of the weights wo, wa and wt in pairs, the reasonable value range of the three weights is comprehensively determined.

[0105] like Figure 5 and Figure 6 As shown in FIG. 1 , the weights Wo, Wa, and Wt of different dimensions are combined in pairs to recommend the most matching compliance cases under the 4-sample condition, and the influence of the weight change on the similarity result of the recommended result object can be seen.

[0106] It can be seen that any single weight of the compliance object, compliance goal, and compliance activity in the current compliance vector algorithm cannot accurately recommend compliance cases. Comparing the three components, the weight of the compliance goal should be higher than the other two weights, and the weight of the compliance object should also be higher than the weight of the compliance activity. Therefore, the intervals of 0.5, 0.3, and 0.2 for Wt, Wo, and Wa are reasonable.

[0107] Figure 7 FIG. 1 is a system framework diagram of an airworthiness relevance case recommendation and evaluation system according to an embodiment of the present invention. Figure 7 As shown, the airworthiness relevance case recommendation and evaluation system of this embodiment mainly includes the following modules:

[0108] The recommendation module 100 obtains the case recommendation results of the compliance case data set by the case recommendation algorithm based on the compliance vector, and extracts the compliance activity record with the highest similarity in the recommendation results of each case.

[0109] The comparison module 200 compares the above recommendation result with other recommendation results to determine the accuracy of the recommendation.

[0110] The recommendation module 100 includes:

[0111] A first recommendation module 101, which obtains similarity metrics of compliance objects, compliance activities and compliance goals, and obtains recommendation results of compliance vectors based on a top-k recommendation algorithm;

[0112] A second recommendation module 102, the second recommendation module obtains a similarity measure between the compliance object and the compliance target, and obtains a recommendation result of the compliance vector based on a top-k recommendation algorithm;

[0113] The weighted recommendation module 103 calculates the reasonable weight value of each dimension in any two-dimensional combination of compliance objects, compliance activities and compliance goals during the calculation process of the top-k recommendation algorithm.

[0114] The comparison module 200 comprises:

[0115] The first comparison module 201 compares the recommendation result of the first recommendation module with the recommendation results of the edit distance recommendation method and the CP-CPI recommendation method, and calculates their precision, recall and F1 value respectively.

[0116] The second comparison module 202 compares the clause information in the recommended compliance case with the sample clause based on the recommendation result of the second recommendation module, and determines the accuracy of the case recommendation through the information distance between the wrong clause and the sample clause;

[0117] The weight comparison module 203 comprehensively compares and calculates the reasonable value ranges of the compliance objects, compliance activities and compliance targets according to the reasonable values ​​of the respective weights in any two combinations of the compliance objects, compliance activities and compliance targets.

[0118] Figure 8 FIG. 1 is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Figure 8 As shown, an electronic device according to an embodiment of the present invention includes one or more input devices 1000 , one or more output devices 1000 , one or more processors 3000 , and a memory 4000 .

[0119] In one embodiment of the present invention, the processor 1000, the input device 2000, the output device 3000 and the memory 4000 may be connected via a bus or other means. The input device 2000 and the output device 3000 may be standard wired or wireless communication interfaces.

[0120] The processor 1000 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0121] The memory 4000 may be a high-speed RAM memory or a non-volatile memory such as a disk memory. The memory 4000 is used to store a set of computer programs, and the input device 2000, the output device 3000 and the processor 1000 may call the program codes stored in the memory 4000.

[0122] The computer program stored in the memory 4000 includes program instructions, which, when executed by a processor, enable the processor to perform the steps of the evaluation method described in the above embodiment.

[0123] One embodiment of the present invention also provides a computer-readable storage medium. The computer-readable storage medium may be a high-speed RAM memory or a non-volatile memory, such as a disk memory. The computer-readable storage medium may be connected via an external computing device or a network to read a set of computer programs stored in the computer-readable storage medium. The computer program stored in the computer-readable storage medium includes program instructions, which, when executed by a processor, cause the processor to perform the steps of the method described in the above embodiment.

[0124] Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. A person of ordinary skill in the art can change, modify, replace and modify the above embodiments within the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The recommended evaluation method for airworthiness cases is characterized by: The following steps are involved: Obtain case recommendation results of the case recommendation algorithm based on the compliance vector for the compliance case data set, and extract the compliance activity record with the highest similarity in the recommendation results of each case; Compare the above recommendation results with other recommendation results to determine the accuracy of the recommendation; Obtain similarity measures for compliance objects, compliance activities, and compliance goals; Based on the similarity measurement of three dimensions, the top-k recommendation algorithm is used to recommend the matching cases; The recommendation results are compared with those of the edit distance recommendation method and the CP-CPI recommendation method, and their precision, recall and F1 values ​​are calculated respectively; Gets similarity measures for compliance objects and compliance activities, where: The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wo of the compliance object and the weight wa of the compliance activity; Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wo and wa by analyzing the relationship between the values ​​and the average similarity of the compliance targets; Obtain similarity measures between compliance activities and compliance goals; The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wa of the compliance activity and the weight wt of the compliance target; Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wa and wt by analyzing the relationship between the values ​​and the average similarity of the compliance targets; obtaining a similarity measure between the compliance object and the compliance target; The top-k recommendation algorithm is used to obtain the recommendation results of different compliance cases by modifying the weight wo of the compliance object and the weight wt of the compliance target; Compare the similarities of the compliance targets in the compliance case recommendation results under different weights, and determine the reasonable values ​​of weights wo and wt by analyzing the relationship between the values ​​and the average similarity of the compliance targets; Combined with the reasonable values ​​of the weights wo, wa and wt in pairs, the reasonable value range of the three weights is comprehensively determined.

2. The airworthiness case recommendation and evaluation method according to claim 1, characterized in that: The following steps are also included: The calculation formula of the accuracy is: Among them, P is the precision, TP is the number of accurate hit cases, and FP is the number of false hit cases; The calculation formula of the recall rate is: Among them, R is the recall rate, FN is the missed case; The calculation formula of the F1 value is:

3. The airworthiness case recommendation and evaluation method according to claim 1, characterized in that: The following steps are also included: Obtaining similarity measures of compliance objects and compliance activities; Recommendation of conforming cases based on the top-k recommendation algorithm; Compare the clause information in the recommended compliance case with the sample clauses, and judge the accuracy of the case recommendation through the information distance between the erroneous clauses and the sample clauses.

4. An airworthiness relevance case recommendation and evaluation system, which is applicable to the method according to any one of claims 1 to 3, characterized in that: include: The recommendation module obtains the case recommendation results of the compliance case data set by the case recommendation algorithm based on the compliance vector, and extracts the compliance activity record with the highest similarity in the recommendation results of each case; The comparison module compares the above recommendation results with other recommendation results to determine the accuracy of the recommendation.

5. The airworthiness relevance case recommendation and evaluation system according to claim 4, characterized in that: The recommendation module includes: A first recommendation module, wherein the first recommendation module obtains similarity metrics of compliance objects, compliance activities, and compliance goals, and obtains recommendation results of compliance vectors based on a top-k recommendation algorithm; A second recommendation module, wherein the second recommendation module obtains a similarity measure between the compliance object and the compliance target, and obtains a recommendation result of the compliance vector based on a top-k recommendation algorithm; The weighted recommendation module calculates the reasonable weight value of each dimension in any two-dimensional combination of compliance objects, compliance activities and compliance goals during the calculation process of the top-k recommendation algorithm.

6. The airworthiness relevance case recommendation and evaluation system according to claim 5, characterized in that: The comparison module also includes: A first comparison module compares the recommendation result of the first recommendation module with the recommendation results of the edit distance recommendation method and the CP-CPI recommendation method, and calculates their precision, recall and F1 value respectively; The second comparison module compares the clause information in the recommended compliance case with the sample clause based on the recommendation result of the second recommendation module, and determines the accuracy of the case recommendation through the information distance between the wrong clause and the sample clause; The weight comparison module comprehensively compares and calculates the reasonable value range of compliance objects, compliance activities and compliance targets based on the reasonable values ​​of the respective weights in any two combinations of compliance objects, compliance activities and compliance targets.

7. The airworthiness relevance case recommendation and evaluation system is characterized by: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 3.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 3.

Images