Tag quality optimization method and apparatus

Abstract

The application discloses a label quality optimization method and device, comprising: determining a label to be optimized; for each label to be optimized, the following operations are performed: constructing a set of business parameter combinations of the label to be optimized according to all original business parameters of the label to be optimized and all general business parameters corresponding to a case type corresponding to the label to be optimized; constructing a new label according to different business parameter combinations in the set respectively, calculating a quality evaluation index of the new label, and when the quality evaluation index of any new label reaches a preset range value, updating the original business parameters of the label to be optimized according to the business parameter combination of the label. The judgment label quality optimization method and device can automatically optimize the label, effectively improve the label quality, reduce the labor input, and improve the portrait accuracy of security objects such as personnel and vehicles.

Description

Technical Field

[0001] This application relates to the field of security technology, and more particularly to a method and apparatus for optimizing label quality. Background Technology

[0002] With the widespread adoption of smart security communities, an increasing number of images and sounds of people and vehicles are being collected and stored by image and sound acquisition devices at pedestrian and vehicle barriers, roads, and elevators. Massive amounts of images, videos, and audio data, after structured analysis, aggregation, and tagging, can generate profiles of individuals, vehicles, and other security targets. Currently, this approach focuses solely on individuals, comparing and analyzing their profiles with related target data to generate correlated data, which is then displayed across multiple screens. However, this method does not verify the effectiveness of the generated target profiles and correlated data, nor does it provide effective optimization suggestions. This hinders the continuous improvement of the data's utilization value and reduces the chances of increasing the crime-solving rate. Summary of the Invention

[0003] This application provides a label quality optimization method and apparatus that can automatically optimize labels, effectively improve label quality, reduce manpower input, and improve the accuracy of profiling security targets such as personnel and vehicles.

[0004] This application provides a label quality optimization method, including:

[0005] Identify the tags to be optimized;

[0006] For each tag to be optimized, perform the following operations:

[0007] Construct a set of business parameter combinations for the tag to be optimized based on all the original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized;

[0008] New tags are constructed based on different combinations of business parameters in the set, and the quality evaluation index of the new tags is calculated. When the quality evaluation index of any new tag reaches a preset range value, the original business parameters of the tag to be optimized are updated according to the combination of business parameters of the tag.

[0009] In one exemplary embodiment, a set of business parameter combinations for the tag to be optimized is constructed based on all the original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized, including:

[0010] Deduplication is performed on all business parameters of the tag to be optimized and all general business parameters corresponding to the case type of the tag to be optimized.

[0011] Based on the deduplicated business parameters and general business parameters, various combinations of business parameters are constructed to obtain a set of business parameter combinations.

[0012] In one exemplary embodiment, before calculating the quality evaluation index of the new label, the method further includes:

[0013] Select a new label using a backtracking algorithm;

[0014] And set the backtracking point to be greater than the preset threshold.

[0015] In one exemplary embodiment, updating the original service parameterization of the tag to be optimized based on the combination of service parameters of the tag includes:

[0016] Replace all the original business parameters of the tag to be optimized with the current combination of business parameters.

[0017] In one exemplary embodiment, determining the label to be optimized includes:

[0018] The tags to be optimized are determined based on the quality evaluation index of the tags, which includes the original business parameters.

[0019] When the quality evaluation index of a tag containing original business parameters reaches a preset range value, the tag containing original business parameters will be identified as a tag to be optimized.

[0020] or,

[0021] When the quality evaluation index of the new label constructed based on each combination of business parameters in the set of business parameter combinations is not greater than the preset threshold, the label to be optimized is determined as the label to be optimized again.

[0022] In one exemplary embodiment, the quality evaluation index of the label is calculated as follows:

[0023] The quality evaluation index of the label is calculated based on the first probability, the second probability, and the third probability.

[0024] The first probability refers to the probability of the tag appearing in the first set. The first set includes all cases that successfully collide with all objects and cases that occurred in the case pool within the first preset time period.

[0025] The second probability refers to the probability of the label appearing in the second set, which includes all cases in the first set that have the same case type as the label;

[0026] The third probability refers to the probability that the tag does not appear within a preset time period;

[0027] The object refers to an entity that contains one or more tags.

[0028] In one exemplary embodiment, calculating the quality evaluation index of the label based on a first probability, a second probability, and a third probability includes:

[0029] The quality evaluation index of the label is calculated based on the first probability, the second probability, the third probability, the preset weight of the first probability, the preset weight of the second probability, and the preset weight of the third probability.

[0030] In one exemplary embodiment, the probability of the tag appearing in the first set is calculated based on the number of times the tag appears in the first set and the total number of cases in the first set;

[0031] The probability of the tag appearing in the second set is calculated based on the number of times the tag appears in the second set and the total number of cases in the second set;

[0032] The probability that a tag does not appear within a preset time period is calculated as a weighted average of the probabilities of the tag not appearing within different preset time periods.

[0033] In one exemplary embodiment, when the label to be optimized is determined to be optimized again, the method further includes:

[0034] Add an optimization count indicator to the tag to be optimized;

[0035] Determine whether the number of times the optimization count indicator has been added to the tag to be optimized exceeds the preset number. If so, mark the tag to be optimized as to be deleted.

[0036] This application provides a tag quality optimization device, including a determination module, a business parameter combination construction module, and an optimization module;

[0037] The determining module is configured to determine the label to be optimized.

[0038] The business parameter combination construction module is configured to construct a set of business parameter combinations for each tag to be optimized based on all the original business parameters of each tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized.

[0039] The optimization module is configured to construct new tags based on different combinations of business parameters in the set, calculate the quality evaluation index of the new tags, and update the original business parameters of the tag to be optimized based on the combination of business parameters of the tag when the quality evaluation index of any new tag reaches a preset range value.

[0040] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. Other advantages of this application can be realized and obtained by means of the solutions described in the description and the accompanying drawings. Attached Figure Description

[0041] The accompanying drawings are used to provide an understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0042] Figure 1 This is a flowchart of a label quality optimization method according to an embodiment of this application;

[0043] Figure 2 This is a flowchart of the security object tag optimization method according to an embodiment of this application;

[0044] Figure 3 This is an example of a combination of business parameters in an embodiment of this application;

[0045] Figure 4 This is a schematic diagram of the label optimization process according to an embodiment of this application;

[0046] Figure 5 This is a schematic diagram of a label quality optimization device according to an embodiment of this application. Detailed Implementation

[0047] Figure 1 This is a flowchart of the label quality optimization method according to an embodiment of this application, such as... Figure 1 As shown, the label quality optimization method of this embodiment includes steps S11-S12:

[0048] S11. Identify the tags to be optimized;

[0049] S12. For each tag to be optimized, perform the following operations:

[0050] Construct a set of business parameter combinations for the tag to be optimized based on all the original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized;

[0051] New tags are constructed based on different combinations of business parameters in the set, and the quality evaluation index of the new tags is calculated. When the quality evaluation index of any new tag reaches a preset range value, the original business parameters of the tag to be optimized are updated according to the combination of business parameters of the tag.

[0052] In one exemplary embodiment, the preset range value can be [0.1-0.6], where [] represents a closed interval. In other embodiments, the user can also set the preset range value to other values ​​according to actual needs.

[0053] In one exemplary embodiment, a set of business parameter combinations for the tag to be optimized is constructed based on all the original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized, including:

[0054] Deduplication is performed on all business parameters of the tag to be optimized and all general business parameters corresponding to the case type of the tag to be optimized.

[0055] Based on the deduplicated business parameters and general business parameters, various combinations of business parameters are constructed to obtain a set of business parameter combinations.

[0056] For example, all the original business parameters of the tag to be optimized include parameter 1, parameter 2, and parameter 3. All the general business parameters corresponding to the case type of the tag to be optimized include parameter 4 and parameter 5. Since parameter 2 and parameter 5 are the same, either parameter 2 or parameter 5 is deleted. Assuming parameter 2 is deleted, the remaining original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized are parameter 1, parameter 3, 4, and 5. These four parameters are combined in different ways to construct parameter combinations. All parameter combinations form a set of parameter combinations.

[0057] The parameter combinations constructed from parameters 1, 3, 4, and 5 include (parameter 1, parameter 3), (parameter 1, parameter 4), (parameter 1, parameter 5), (parameter 3, parameter 4), (parameter 3, parameter 5), (parameter 4, parameter 5), (parameter 1, parameter 3, parameter 4), (parameter 1, parameter 3, parameter 5), (parameter 3, parameter 4, parameter 5), (parameter 1, parameter 3, parameter 4, parameter 5).

[0058] In one exemplary embodiment, before calculating the quality evaluation index of the new label, the method further includes:

[0059] Select a new label using a backtracking algorithm;

[0060] And set the backtracking point to be greater than the preset threshold.

[0061] The preset threshold can be 0.6, and users can set this preset threshold according to their actual needs.

[0062] In some other exemplary embodiments, instead of using a backtracking algorithm to select and calculate the quality evaluation index of new labels, the quality evaluation index of all new labels can be calculated. However, using a backtracking algorithm, once the quality evaluation index of the selected new label is greater than a preset threshold, the quality evaluation index of the remaining new labels does not need to be calculated, thereby improving efficiency and reducing computational load.

[0063] In one exemplary embodiment, updating the original business parameters of the tag to be optimized according to the combination of business parameters of the tag includes: replacing all the original business parameters of the tag to be optimized with the combination of business parameters of the tag.

[0064] In one exemplary embodiment, determining the label to be optimized includes:

[0065] The tags to be optimized are determined based on the quality evaluation index of the tags, which includes the original business parameters.

[0066] When the quality evaluation index of a tag containing original business parameters reaches a preset range value, the tag containing original business parameters will be identified as a tag to be optimized.

[0067] or,

[0068] When the quality evaluation index of the new label constructed based on each combination of business parameters in the set of business parameter combinations is not greater than the preset threshold, the label to be optimized is determined as the label to be optimized again.

[0069] In one exemplary embodiment, the quality evaluation index of the label is calculated as follows:

[0070] The quality evaluation index of the label is calculated based on the first probability, the second probability, and the third probability.

[0071] The first probability refers to the probability of the tag appearing in the first set. The first set includes all cases that successfully collide with all objects and cases that occurred in the case pool within the first preset time period.

[0072] The second probability refers to the probability of the label appearing in the second set, which includes all cases in the first set that have the same case type as the label;

[0073] The third probability refers to the probability that the tag does not appear within a preset time period;

[0074] The "object" refers to an entity that contains one or more tags. Tags are highly refined feature identifiers obtained through information analysis of the entity.

[0075] The object mentioned can be a suspect in the case or a suspicious vehicle, etc.

[0076] In one exemplary embodiment, calculating the quality evaluation index of the label based on a first probability, a second probability, and a third probability includes:

[0077] The quality evaluation index of the label is calculated based on the first probability, the second probability, the third probability, the preset weight of the first probability, the preset weight of the second probability, and the preset weight of the third probability.

[0078] In one exemplary embodiment, the probability of the tag appearing in the first set is calculated based on the number of times the tag appears in the first set and the total number of cases in the first set;

[0079] The probability of the tag appearing in the second set is calculated based on the number of times the tag appears in the second set and the total number of cases in the second set;

[0080] The probability that a tag does not appear within a preset time period is calculated as a weighted average of the probabilities of the tag not appearing within different preset time periods.

[0081] For example, the probability of each tag occurring in a case (i.e., the first probability) = the number of times each tag appears in a case (within 1 year) / the total number of cases (within 1 year) where the collision is successful;

[0082] The probability of each label occurring in the same type of case (incident) (i.e., the second probability) = the number of times each label appears in the total number of cases (incidents) of the same type / the total number of cases (incidents) of the same type that successfully collide;

[0083] The probability of each label not appearing for a long period of time (i.e., the third probability) = 5% * probability of not appearing in the last 7 days + 10% * probability of not appearing in the last 30 days + 25% * probability of not appearing in the last 180 days + 60% * probability of not appearing in the last 360 days;

[0084] Tag quality evaluation index = 10% * probability of occurrence of each tag case (incident) + 80% * probability of occurrence of each tag corresponding type of case (incident) + 10% * probability of each tag not appearing for a long time.

[0085] In one exemplary embodiment, when the label to be optimized is determined to be optimized again, the method further includes:

[0086] Add an optimization count indicator to the tag to be optimized;

[0087] Determine whether the number of times the optimization count indicator has been added to the tag to be optimized exceeds the preset number. If so, mark the tag to be optimized as to be deleted.

[0088] The optimization frequency indicator can be added to the upper right corner of the tag, such as an asterisk (*). Tags marked as pending deletion can be deleted periodically.

[0089] The tag optimization method provided in this application can automatically optimize tags, effectively improve tag quality, reduce manpower input, and improve the accuracy of portraits of security objects such as personnel and vehicles.

[0090] Figure 2 This is a flowchart of the security object tag optimization method according to an embodiment of this application, such as... Figure 2 As shown, steps S21-S25 are included:

[0091] S21. Identify the tags to be optimized;

[0092] S22. Extract key business parameters according to case or event type, and construct a general label optimization model for each case or event type;

[0093] S23. Merge and deduplicate the business parameters of the tag to be optimized and the business parameters in the general tag optimization model for the case or event type corresponding to the tag to be optimized.

[0094] S24. Use a backtracking algorithm to traverse and merge the deduplicated business parameter combinations, and calculate the quality index of the multiple tags formed by each business parameter combination.

[0095] S25. When the quality index of a label meets the preset backtracking point condition (i.e., the first preset condition mentioned above), the label to be optimized and the general optimization model are processed according to the preset strategy, and the processed label to be optimized is used as the optimized label; when the quality index of a label does not meet the preset backtracking point condition, the label is determined as the label to be optimized.

[0096] In step S21, determining the label to be optimized includes the following sub-steps:

[0097] S211. Perform data collision between all security objects and cases and / or events that occur within a preset event segment in the case database and / or event database, count the total number of cases and / or events that successfully collide, and count the total number of cases or events of each case type or event type in the cases and / or events that successfully collide.

[0098] S212. For each tag of each security object, determine the case type or event type corresponding to that tag;

[0099] S213. Calculate the first probability based on the number of times the label appears in the cases and / or events where the collision is successful, and the total number of cases and / or events where the collision is successful.

[0100] S214. Calculate the second probability based on the number of times the label appears in the case type and / or event type corresponding to the label in the cases and / or events where the collision is successful, and the total number of case types and / or event types corresponding to the label in the cases and / or events where the collision is successful;

[0101] S215. Calculate the third probability based on the probability of the label not appearing in different preset time periods and the corresponding weights.

[0102] S216. Calculate the quality index of the label based on the first probability, the second probability, the third probability, the weight corresponding to the first probability, the weight corresponding to the second probability, and the weight corresponding to the third probability.

[0103] S217. When the quality index of the label meets the preset conditions, the label is determined to be a label to be optimized.

[0104] For example, a specific example of the label quality evaluation index is as follows:

[0105] Assume there are 1000 cases in the current case pool from the past year (360 days), including 500 thefts, 100 robberies, and 400 frauds. In the last 7 days, there were 3 thefts, 1 robbery, and 5 frauds; in the last 30 days, there were 35 thefts, 7 robberies, and 30 frauds; and in the last 180 days, there were 315 thefts, 80 robberies, and 165 frauds.

[0106] Security targets A and B successfully matched a total of 60 cases with the case pool, including 30 thefts, 10 robberies, and 20 frauds. Specifically, security target A had 20 successful cases matched, including 10 thefts (i.e., the "frequently passes through the parking lot at night" tag occurred 10 times), 5 robberies, and 5 frauds; security target B had 40 successful cases matched, including 20 thefts, 5 robberies, and 15 frauds.

[0107] With a time frame of 1 year, the probability of each tag occurring in a case (i.e., the first probability) = the number of times each tag appears in a case (within 1 year) / the total number of cases (within 1 year) where the collision is successful;

[0108] The probability of each label occurring in the same type of case (i.e., the second probability) = the number of times each label appears in the total number of cases (incidents) of the same type (within 1 year) / the total number of cases (incidents) of the same type that successfully collide (within 1 year);

[0109] The probability of each label not appearing for a long period of time (i.e., the third probability) = 5% * probability of not appearing in the last 7 days + 10% * probability of not appearing in the last 30 days + 25% * probability of not appearing in the last 180 days + 60% * probability of not appearing in the last 360 days;

[0110] Wherein, the probability of each label not appearing = 1 - the probability of the occurrence of the case (incident) of the corresponding type for each label; for example...

[0111] The probability that a tag has not appeared in the past 7 days = 1 - the probability of the occurrence of the type of case (incident) corresponding to the tag (within the past 7 days).

[0112] Tag quality evaluation index = 10% * probability of occurrence of each tag case (incident) + 80% * probability of occurrence of each tag corresponding type of case (incident) + 10% * probability of each tag not appearing for a long time.

[0113] So,

[0114] The probability of a case tagged "frequently passes through the parking lot at night" (360 days) is approximately 10 / 60 ≈ 0.17.

[0115] The probability of theft cases corresponding to the tag "frequently passes through the parking lot at night" (360 days) = 10 / 30 ≈ 0.33;

[0116] The probability of theft cases corresponding to the tag "frequently passes through the parking lot at night" (180 days) = 7 / 18 ≈ 0.39;

[0117] The probability of theft cases corresponding to the tag "frequently passes through the parking lot at night" (30 days) = 2 / 4 = 0.50;

[0118] The probability of theft cases corresponding to the tag "frequently passes through the parking lot at night" (7 days) = 1 / 1 = 1.00;

[0119] The probability that the label "frequently passes through the parking lot at night" does not appear (360 days) = 1 - 10 / 30 ≈ 0.67;

[0120] The probability that the "frequently passes through the parking lot at night" label does not appear (180 days) = 1 - 7 / 18 ≈ 0.61;

[0121] The probability that the "frequently passes through the parking lot at night" label does not appear (30 days) = 1 - 2 / 4 = 0.50;

[0122] The probability that the "frequently passes through the parking lot at night" label does not appear (7 days) = 1 - 1 / 1 = 0.00;

[0123] The probability that the "frequently passes through the parking lot at night" label will not appear for a long time = 5% * 0 + 10% * 0.50 + 25% * 0.61 + 60% * 0.67 ≈ 0.60;

[0124] The quality evaluation index for the label "Frequently pass through the parking lot at night" = 10% * 0.17 + 80% * 0.33 + 10% * 0.60 = 0.017 + 0.264 + 0.06 = 0.341.

[0125] Tags are marked based on a tag quality evaluation index. For example, when the tag evaluation index is less than 0.1, the tag is marked as to be deleted; when the tag evaluation index is between 0.1 and 0.6, the tag is marked as to be optimized; and when the tag quality evaluation index is greater than 0.6, the tag is marked as an optimized tag.

[0126] In step S211, existing technology is used to perform a collision comparison using structured data and image features. A successful collision occurs when the unique identifiers in the structured data are identical, or when the image feature similarity is greater than or equal to a certain threshold (e.g., 95%). For example, for a person, the unique identifier in their structured data could be an "evidence number," and the image features could be "face" and "human image." If the similarity between the person's identification number and a case or event in the case database or event database reaches a certain threshold, a successful collision is considered.

[0127] Cases or events in the case or event database can be categorized by type, such as theft, robbery, fraud, etc. Each security object can have multiple tags. Each tag can also have multiple business parameters. Each tag also has a corresponding case (event) type. For example, the case (event) type corresponding to the tag "frequently passes through the parking lot at night" could be theft.

[0128] As the business parameters of the tags change continuously with the optimization of the tags, the cases or events that successfully collide with the security objects in the case database or event database also change. As a result, the total number of cases and / or events that successfully collide, the total number of cases or events of each case type or event type in the cases and / or events that successfully collide, the first probability, the second probability, and the probability of not appearing will all change.

[0129] In step S22, key business parameters, such as the time and location of the crime, are included. These key business parameters are incorporated into the general label optimization model for each case or event type.

[0130] In step S24, the backtracking algorithm used here is similar to an enumeration-based search process. It primarily searches for a solution during the search process, and when it finds that the solution conditions are no longer met, it "backtracks" and tries other paths. Backtracking is an optimization search method that searches forward according to optimization conditions to reach the goal. However, when it reaches a certain step and finds that the original choice was not optimal or could not reach the goal, it backtracks and chooses again. This technique of retreating when a path is blocked is called backtracking, and a point in a state that meets the backtracking conditions is called a "backtracking point." Many complex and large-scale problems can use backtracking.

[0131] The backtracking point condition in this step can be "the label quality index is greater than a preset threshold". For example, a label quality evaluation index greater than 0.6 can be used as the backtracking point condition. The backtracking point condition here is the same as the evaluation criteria for optimizing labels mentioned above.

[0132] The combined business parameters after merging and deduplication can be as follows: Figure 3 As shown, this diagram is only an example of combinations and does not include all combinations.

[0133] In step S25, when the backtracking point condition is met, the preset strategy can be to compare the business parameter combination that meets the backtracking point with the business parameters of the tag to be optimized, delete the business parameters of the tag to be optimized that are not in the business parameter combination that meets the backtracking point, add the parameters that are in the general optimized tag model when the backtracking point is met but are not in the business parameters of the tag to be optimized, and update the parameter values ​​that are in both the general optimized tag model and the business parameters of the tag to be optimized when the backtracking point is met to the parameter values ​​in the general optimized model. The modified tag business parameters are used as the parameters of the optimized tag. Alternatively, the modified tag can be written to the optimized tag list in the tag pool.

[0134] You can also add an optimization count indicator to the tag, and delete the tag after a certain number of optimizations have been achieved.

[0135] The tag optimization method provided in this application can effectively improve tag quality, reduce manpower input, and improve the accuracy of portraits of security objects such as personnel and vehicles.

[0136] Figure 4 This is a schematic diagram of the label optimization process according to an embodiment of this application, as shown below. Figure 4As shown, assuming the tag to be optimized is "frequently passes through the parking lot at night", this tag includes three parameters: "occurred 5 or more days in the last 30 days", "occurred 10 or more times in the last 30 days", and "occurred between 9 pm and 5 am the next day". The case type corresponding to the tag to be optimized is determined to be a motor vehicle theft case. The general business model for motor vehicle theft cases includes two parameters: "occurred 5 or more days in the last 30 days" and "occurred between 11 pm and 3 am the next day". These five parameters are merged and deduplicated, resulting in four parameters: "occurred 5 or more days in the last 30 days", "occurred 10 or more times in the last 30 days", "occurred between 9 PM and 5 AM the next day", and "occurred between 11 PM and 3 AM the next day". These four parameters are then combined in different ways. For example, some parameter combinations include: {["occurred 5 or more days in the last 30 days", "occurred between 11 PM and 3 AM the next day"], ["occurred 10 or more times in the last 30 days", "occurred between 11 PM and 3 AM the next day"], ["occurred between 9 PM and 5 AM the next day", "occurred between 11 PM and 3 AM the next day"], ["occurred 5 or more days in the last 30 days", "occurred 10 or more times in the last 30 days", "occurred between 11 PM and 3 AM the next day"], ["occurred 5 or more days in the last 30 days", "occurred 10 or more times in the last 30 days", "occurred between 11 PM and 3 AM the next day"]]. For each parameter combination, a new tag is constructed, and the quality index of the new tag is calculated (calculated as described above): 0.35, 0.38, 0.23, 0.69, 0.29, 0.41, and 0.31, respectively. Parameters with a quality index greater than 0.6 are selected as the optimized tag business parameters. The optimized "Frequently pass through parking lots at night" label includes three parameters: "occurred 5 or more days in the last 30 days", "occurred 10 or more times in the last 30 days", and "occurred between 11 pm and 3 am the next day".

[0137] Figure 5 This is a schematic diagram of the label optimization device according to an embodiment of this application, such as... Figure 5 As shown, a tag quality optimization device according to this embodiment includes a determination module, a business parameter combination construction module, and an optimization module;

[0138] The determining module is configured to determine the label to be optimized.

[0139] The business parameter combination construction module is configured to construct a set of business parameter combinations for each tag to be optimized based on all the original business parameters of each tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized.

[0140] The optimization module is configured to construct new tags based on different combinations of business parameters in the set, calculate the quality evaluation index of the new tags, and update the original business parameters of the tag to be optimized based on the combination of business parameters of the tag when the quality evaluation index of any new tag reaches a preset range value.

[0141] The label optimization device provided in this application embodiment can effectively improve label quality, reduce manpower input, and improve the accuracy of portraits of security objects such as personnel and vehicles.

[0142] This application describes several embodiments, but these descriptions are exemplary and not restrictive, and it will be apparent to those skilled in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with any other feature or element of any other embodiment, or may substitute for any other feature or element of any other embodiment.

[0143] This application includes and contemplates combinations of features and elements known to those skilled in the art. The embodiments, features, and elements disclosed in this application may also be combined with any conventional features or elements to form a unique inventive scheme as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive schemes to form another unique inventive scheme as defined by the claims. Therefore, it should be understood that any feature shown and / or discussed in this application may be implemented individually or in any suitable combination. Therefore, the embodiments are not limited except by the limitations imposed by the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

[0144] Furthermore, in describing representative embodiments, the specification may have presented methods and / or processes as a specific sequence of steps. However, the method or process should not be limited to the specific order of steps described herein, to the extent that it does not depend on such a specific order. As will be understood by those skilled in the art, other sequences of steps are also possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation of the claims. Moreover, the claims concerning the method and / or process should not be limited to the steps performed in the written order, and those skilled in the art will readily understand that these orders can be varied and still remain within the spirit and scope of the embodiments of this application.

[0145] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A method for optimizing label quality, characterized in that, Identify the tags to be optimized; For each tag to be optimized, perform the following operations: Construct a set of business parameter combinations for the tag to be optimized based on all the original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized; New tags are constructed based on different combinations of business parameters in the set, and the quality evaluation index of the new tags is calculated. When the quality evaluation index of any new tag reaches a preset range value, the original business parameters of the tag to be optimized are updated according to the combination of business parameters of the tag. The quality evaluation index of the label is calculated using the following method: The quality evaluation index of the label is calculated based on the first probability, the second probability, and the third probability. The first probability refers to the probability of the tag appearing in the first set. The first set includes all cases that successfully collide with all objects and cases that occurred in the case pool within the first preset time period. The second probability refers to the probability of the label appearing in the second set, which includes all cases in the first set that have the same case type as the label; The third probability refers to the probability that the tag does not appear within a preset time period; The object refers to an entity that contains one or more tags.

2. The method as described in claim 1, characterized in that, A set of business parameter combinations for the tag to be optimized is constructed based on all the original business parameters of the tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized, including: Deduplication is performed on all business parameters of the tag to be optimized and all general business parameters corresponding to the case type of the tag to be optimized. Based on the deduplicated business parameters and general business parameters, various combinations of business parameters are constructed to obtain a set of business parameter combinations.

3. The method as described in claim 1, characterized in that, Before calculating the quality evaluation index for the new label, the following steps are also included: Select a new label using a backtracking algorithm; And set the backtracking point to be greater than the preset threshold.

4. The method as described in claim 1, characterized in that, Update the original business parameterization of the tag to be optimized based on the combination of business parameters of the tag, including: Replace all the original business parameters of the tag to be optimized with the current combination of business parameters.

5. The method as described in claim 1, characterized in that, The process of determining the label to be optimized includes: The tags to be optimized are determined based on the quality evaluation index of the tags, which includes the original business parameters. When the quality evaluation index of a tag containing original business parameters reaches a preset range value, the tag containing original business parameters will be identified as a tag to be optimized. or, When the quality evaluation index of the new label constructed based on each combination of business parameters in the set of business parameter combinations is not greater than the preset threshold, the label to be optimized is determined as the label to be optimized again.

6. The method as described in claim 1, characterized in that, The quality evaluation index of the label is calculated based on the first probability, the second probability, and the third probability, including: The quality evaluation index of the label is calculated based on the first probability, the second probability, the third probability, the preset weight of the first probability, the preset weight of the second probability, and the preset weight of the third probability.

7. The method as described in claim 6, characterized in that, The probability of the tag appearing in the first set is calculated based on the number of times the tag appears in the first set and the total number of cases in the first set; The probability of the tag appearing in the second set is calculated based on the number of times the tag appears in the second set and the total number of cases in the second set; The probability that a tag does not appear within a preset time period is calculated as a weighted average of the probabilities of the tag not appearing within different preset time periods.

8. The method as described in claim 5, characterized in that, When the label to be optimized is identified as a label to be optimized again, it also includes: Add an optimization count indicator to the tag to be optimized; Determine whether the number of times the optimization count indicator has been added to the tag to be optimized exceeds the preset number. If so, mark the tag to be optimized as to be deleted.

9. A label quality optimization device, characterized in that: This includes a determination module, a business parameter combination construction module, and an optimization module; The determining module is configured to determine the label to be optimized. The business parameter combination construction module is configured to construct a set of business parameter combinations for each tag to be optimized based on all the original business parameters of each tag to be optimized and all the general business parameters corresponding to the case type of the tag to be optimized. The optimization module is configured to construct new tags based on different combinations of business parameters in the set, calculate the quality evaluation index of the new tags, and update the original business parameters of the tag to be optimized based on the combination of business parameters of the tag when the quality evaluation index of any new tag reaches a preset range value. The quality evaluation index of the label is calculated using the following method: The quality evaluation index of the label is calculated based on the first probability, the second probability, and the third probability. The first probability refers to the probability of the tag appearing in the first set. The first set includes all cases that successfully collide with all objects and cases that occurred in the case pool within the first preset time period. The second probability refers to the probability of the label appearing in the second set, which includes all cases in the first set that have the same case type as the label; The third probability refers to the probability that the tag does not appear within a preset time period; The object refers to an entity that contains one or more tags.

Images