A plankton specimen information interaction system based on an examination sample

Abstract

The present application relates to the technical field of data processing, and especially relates to a plankton specimen information interaction system based on examination samples. The present application proposes the following scheme: the examination sample comprises a plankton specimen main body and a distinguishing part separately arranged at different sample positions, and each sample position is associated with a distinguishing feature of a classification level; a classification result and a feature recognition result submitted by an examiner are acquired, the classification result is matched with a standard classification path, and a classification matching level is determined; for a classification node to be supported, a minimum support position group is determined according to the difference distinguishing features and the associated sample positions between the classification node and similar classification objects, and a node support state is determined according to the feature recognition results in the group; a sample score is determined according to the classification matching level and the node support state, and an examination score is generated. The present application can take into account the classification result and the morphological identification basis, and improve the accuracy and traceability of the evaluation of the plankton identification capability.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an information interaction system for zooplankton specimens based on assessment samples. Background Technology

[0002] Zooplankton species composition and community structure are crucial foundational data for marine ecological monitoring, aquatic environment assessment, and biological resource surveys. Identification typically relies on technicians' comprehensive judgment under microscopic conditions, examining the overall morphology, segmental composition, appendage structure, setae arrangement, and local proportions of the specimen. Unlike physicochemical tests that directly output numerical results, zooplankton identification is highly dependent on professional experience, and the morphological characteristics used at different taxonomic levels vary significantly. While some higher-level groups can be identified based on their overall appearance, genus- and species-level identification often requires further observation of smaller, easily obscured local structures such as the fifth thoracic leg, caudal fork, and antennae. This makes it difficult to accurately evaluate identification capabilities through conventional assessment methods.

[0003] Current zooplankton proficiency assessments typically involve the assessment organizer selecting target specimens from collected samples and distributing multiple specimens in centrifuge tubes or specimen bottles. Due to the minute size of zooplankton and the fragility of their tentacles, appendages, and setae, specimens are prone to damage, detachment, adhesion, or changes in posture during selection, mixing, transportation, and repeated observation. This results in variations in individual integrity, visibility of key features, and observation difficulty among different assessment samples. To ensure sample stability, current assessments often favor common species with distinct morphologies and large numbers, failing to adequately cover closely related taxa that rely on minute distinguishing features for differentiation. This limits the depth of the assessment content and the comparability between different laboratories.

[0004] Furthermore, existing scoring methods primarily rely on whether the final classification name submitted by the examiner matches the standard answer, failing to reflect the morphological identification process corresponding to the classification conclusion. Examiners may have correctly identified a higher-level classification but deviated in a lower-level classification due to the omission of key appendages; conversely, they may fail to identify necessary distinguishing features but submit the correct name based on experience or chance. These different ability levels may receive similar or identical evaluations when only comparing the final results, making it impossible to accurately pinpoint specific issues such as omitted specimen parts, misidentification of morphological features, or errors in classification relationships. Therefore, establishing a correspondence between the classification results submitted by examiners and specific identification parts and their morphological characteristics while maintaining the independence of the assessment, and forming a traceable and comparable ability evaluation accordingly, has become a pressing technical problem to be solved in zooplankton identification assessments.

[0005] To address the above issues, this application presents a zooplankton specimen information exchange system based on assessment samples. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to address the shortcomings of the prior art by providing a zooplankton specimen information exchange system based on assessment samples, thereby improving the accuracy, comparability and traceability of zooplankton identification and assessment.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A zooplankton specimen information interaction system based on assessment samples, wherein the assessment samples include a zooplankton specimen body and identification sites located at different sample locations, and each sample location is associated with identification features of a corresponding taxonomic level; the system includes:

[0009] The data upload module is used to obtain the classification results submitted by the assessors for the assessment samples and the feature recognition results corresponding to the positions of each sample.

[0010] The data processing module is used to match the classification result with the standard classification path corresponding to the assessment sample, determine the classification matching level, determine the corresponding minimum support position group, and determine the node support status based on the feature recognition results corresponding to the position of each sample in the minimum support position group. The minimum support position group includes multiple classification nodes to be supported, and the classification nodes to be supported are determined by filtering the assessment classification path corresponding to the classification result.

[0011] The score generation module is used to determine the sample score of the assessment sample based on the classification matching level and the node support status of each of the classification nodes to be supported, and to generate an assessment score based on the sample scores of multiple assessment samples.

[0012] The assessment sample is a standardized physical assessment sample generated by separately encapsulating the main body of a zooplankton specimen and at least one identification site in the same carrier. The preparation method of the assessment sample includes:

[0013] The system comprises a carrier substrate, a support plate, and an encapsulation layer connected in sequence. The support plate has multiple independent sample positions. The carrier substrate supports the support plate, and the encapsulation layer closes each sample position.

[0014] The main body and identification area of ​​the target zooplankton specimen are fixed to the support plate, and the test sample is generated by sample encapsulation.

[0015] Multiple sample locations are formed in multiple grooves on the support plate, and the preparation method of the test sample further includes:

[0016] Based on the shape and size of the specimen body and the identification part to be fixed, the outline shape and accommodating depth of each groove are determined, and position marks corresponding to each groove are set on the supporting substrate. The position marks include at least the test sample number and the sample position number.

[0017] Fixing the main body of the specimen and the identification site to different sample locations, including:

[0018] Based on the group attributes and structural stability of the specimen body and each target identification site, the corresponding preservation matrix is ​​determined, the specimen body and each target identification site are placed in the corresponding sample position and fixed by the preservation matrix, and each sample position is sealed by the transparent encapsulation layer.

[0019] The data processing module includes:

[0020] The classification path processing unit is used to parse the classification result, generate an assessment classification path formed by sequentially connecting classification nodes of different classification levels, match the assessment classification path with the standard classification path, determine the classification matching level, and determine at least one classification node to be supported from the assessment classification path.

[0021] The support location determination unit is used to generate a candidate support location group for each of the classification nodes to be supported, based on the difference identification features between the classification node to be supported and the sample locations associated with each difference identification feature, and to determine the target support location group from the candidate support location group.

[0022] The support status determination unit is used to determine the node support status of the corresponding classification node based on the feature recognition results corresponding to the sample positions within the target support location group.

[0023] The data processing module is configured with a classification evidence processing strategy, which includes path filtering logic, location filtering logic, and evidence verification logic. The path filtering logic is configured in the classification path processing unit, the location filtering logic is configured in the support location determination unit, and the evidence verification logic is configured in the support status determination unit.

[0024] The path filtering logic is used to filter classification nodes located below the preset evidence verification level from the assessment classification path according to the preset evidence verification level, so as to obtain the classification nodes to be supported.

[0025] The location filtering logic includes:

[0026] For each of the classification nodes to be supported, identify multiple similar classification objects that share the same parent classification node with the classification node to be supported, and determine the difference identification features between the classification node to be supported and each of the similar classification objects;

[0027] Based on the correlation between the differential identification features and the sample locations, the set of distinguishable objects corresponding to each sample location is determined, and the combination of sample locations that can cover all similar classification objects in the corresponding set of distinguishable objects is determined as the candidate support location group.

[0028] Based on the number of sample locations contained in each candidate support location group and the discriminative power of the corresponding differential features, a target support location group is determined from the multiple candidate support location groups, wherein the number of sample locations contained in the target support location group is no greater than the number of sample locations contained in other candidate support location groups.

[0029] The evidence verification logic includes:

[0030] The feature recognition results corresponding to each sample position within the target support position group are matched with the difference identification features of the classification node to be supported at the corresponding identification part to obtain the evidence chain matching result.

[0031] If all feature recognition results within the target support location group match the corresponding difference identification features, the node support status of the classification node to be supported is determined to be complete support.

[0032] If at least one feature recognition result matches the corresponding difference identification feature, and the remaining feature recognition results are unrecognized, the node support status is determined to be partially supported.

[0033] If any feature recognition result conflicts with the corresponding difference identification feature, or if the conditions for determining complete support and partial support are not met, the node support status will be determined as not supported.

[0034] The score generation module includes:

[0035] The classification score determination unit is used to determine the initial classification score from the preset starting classification level to the classification matching level based on the preset level score for each classification level; determine the corresponding support coefficient based on the node support status of each node to be supported; and adjust the initial classification score based on the support coefficient, wherein the support coefficient corresponding to complete support is greater than the support coefficient corresponding to partial support, and the support coefficient corresponding to no support is less than the support coefficient corresponding to partial support.

[0036] The feature score determination unit is used to determine the feature score based on the matching status of the feature recognition result and the corresponding identification feature corresponding to each sample position, and when the same sample position is associated with multiple classification nodes to be supported, the feature score corresponding to the sample position is included in one of the classification nodes to be supported.

[0037] A sample score generation unit is used to determine the sample score based on the adjusted classification score and the feature score.

[0038] The system also includes a teaching report generation module, which is used to generate a teaching report for the examiner based on the sample score.

[0039] The teaching report shall include at least:

[0040] The classification matching level, the node to be supported, the node support status, the classification score, the feature score, and the sample score corresponding to each assessment sample;

[0041] The target support location group corresponding to each of the nodes to be supported for classification, and the identification parts, difference identification features and evidence chain matching results in the target support location group;

[0042] The unidentified identification parts and conflict identification parts are determined based on the node support status of the unsupported or unsupported classification node, as well as the difference information between the unsupported classification node and similar classification objects.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] This invention establishes a correspondence between the specimen body, identification site, sample location, and classification level, enabling the classification results to be associated with specific morphological evidence. It also performs stratified evaluation based on classification matching level, minimum support location group, and node support status, thereby distinguishing different situations such as classification errors, feature omissions, and accidental hits. This improves the accuracy, comparability, and traceability of zooplankton identification assessment, while facilitating the identification of weak links and targeted teaching. Attached Figure Description

[0045] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0046] Figure 1 An exemplary application scenario diagram provided for an embodiment of this application;

[0047] Figure 2 This is another exemplary application scenario diagram provided in the embodiments of this application;

[0048] Figure 3A schematic diagram of a zooplankton specimen information interaction system based on assessment samples provided in this application embodiment;

[0049] Figure 4 A flowchart illustrating a method for information exchange of zooplankton specimens based on assessment samples, provided in an embodiment of this application;

[0050] Figure 5 This is a schematic diagram of the packaging structure of the test sample provided in the embodiments of this application. Detailed Implementation

[0051] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0052] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0053] In this application, the zooplankton specimen information exchange method is mainly applied to scenarios such as marine ecological monitoring, zooplankton morphological identification, personnel competence verification for testing and inspection institutions, and inter-laboratory proficiency testing. Especially in activities related to the accreditation of testing and inspection institutions, zooplankton identification results are often an important component of marine ecological survey data. Their accuracy depends not only on the instruments and testing procedures but also, to a large extent, on the ability of the identification personnel to observe, compare, and classify minute morphological structures. Therefore, how to evaluate the zooplankton identification capabilities of different personnel or laboratories in a repeatable, comparable, and traceable manner has long been a practical challenge in biological competence assessment.

[0054] Unlike physicochemical testing, zooplankton identification typically does not produce a single numerical result directly from the instrument. Technicians must determine the taxonomic level of a specimen under microscopic observation, based on its overall morphology, segmental composition, appendage structure, setae arrangement, terminal morphology, and the proportional relationships between different parts. Because the morphological differences between different zooplankton groups exhibit a clear hierarchical structure, some higher-level groups can be identified based on the overall outline of the specimen, while similar species within the same higher-level group may be distinguished only by a very small local structure.

[0055] For example, an examiner might be able to identify a specimen as a copepod based on its body shape and main structure, but to further differentiate its subclasses, they would need to observe the fifth thoracic leg, caudal fork, antennae, or other specific appendages. These identifying sites are usually small and may be located on the ventral side, distal end, or overlapping area of ​​the appendages. If the examiner misses these sites during observation, even if they correctly identify the major group to which the specimen belongs, they will find it difficult to accurately complete the genus or species identification. Therefore, the ability to identify zooplankton is not simply a matter of whether the final classification name is correct, but also reflects the examiner's ability to find key identifying sites, recognize the morphological characteristics of those sites, and use the identified characteristics to complete subsequent classification judgments.

[0056] Current methods for assessing zooplankton proficiency typically require experienced technicians to select target specimens from field collections. Multiple specimens are then mixed in centrifuge tubes or small specimen bottles and distributed as assessment samples to different personnel or laboratories. Due to the tiny size of zooplankton and the fragility of structures such as jellyfish tentacles and copepod appendages, specimens are prone to adhesion, damage, or loss during selection, mixing, transportation, and observation. Even if the assessment samples nominally contain the same species, differences may arise due to variations in individual specimens, structural integrity, and the visible condition of key parts, thus affecting the comparability of assessment results.

[0057] To avoid the aforementioned impacts, current assessments often tend to select common species that are numerous, morphologically obvious, and structurally relatively stable. While this reduces the difficulty of sample preparation and distribution, it correspondingly reduces the assessment of minute appendages, cryptic structures, and the ability to distinguish closely related species. This makes the assessment results more likely to reflect the assessor's ability to identify common broad categories, rather than accurately evaluating their actual mastery of key taxonomic features. For proficiency testing or interlaboratory comparisons that require demonstrating the laboratory's true identification capabilities, relying solely on common species and obvious features is insufficient to cover the complex taxa that may arise during actual monitoring.

[0058] On the other hand, existing scoring methods generally compare the species names submitted by examiners with preset results and determine the score based on whether the names match. This scoring method is difficult to distinguish between different identification processes. For example, two examiners may submit the same incorrect species name, but one has correctly identified it down to the family level and only misidentified it on a key appendage; the other may only make a guess based on the overall appearance. Conversely, examiners may not actually identify the key features that determine the lower-level classification, but submit the correct species name due to experience or chance. If the overall score is based solely on the final name, the above different ability states may result in the same or similar assessment results, making it difficult to truly reflect the examiners' morphological identification ability.

[0059] Those skilled in the art will understand that traditional mixed-sample assessment only provides the biological individuals to be observed, but fails to transform the key identifying features supporting classification judgments into independently locatable and recordable assessment objects. For assessment systems, they typically only obtain the classification names submitted by the assessors, without determining whether they observed the corresponding key features, nor can they stably correlate a particular characteristic judgment with the specific structures in the actual specimen. Therefore, there is an information gap between the observation process of the physical specimen and the digital scoring process, and the classification results lack morphological evidence that can be recorded and verified by the system.

[0060] Based on the above issues, this application proposes that zooplankton competence assessment should not only focus on the final classification name, but should organize assessment samples around the key morphological evidence needed to form classification conclusions. Specifically, while retaining the main body of the zooplankton specimen for overall morphological observation, the identifying parts used to distinguish similar classification objects are presented independently, so that the main body of the specimen and different identifying parts correspond to specific sample positions, and each sample position can establish a stable relationship with the corresponding identifying features and the classification level it supports.

[0061] Through the above method, examiners still need to independently complete specimen observation and classification without prompting. However, after the assessment, the system can obtain not only the submitted classification results but also the feature recognition results formed for different identification sites. Therefore, the examiner's classification path can be combined with key morphological evidence for evaluation, thus distinguishing between different situations such as only identifying the superior group, omitting key identification sites, misinterpreting local morphology, and obtaining correct classification results by chance despite lacking feature support.

[0062] It should be noted that the core concept of this application is not to change the taxonomic rules on which zooplankton morphology identification is based, nor to provide observation prompts to examiners during the assessment process. Rather, it is to adjust the organization of the main body and identification parts of the specimen in the physical assessment sample so that the originally difficult-to-record local observation results can be linked with the corresponding taxonomic level and participate in the determination of the assessment score after the assessment is completed.

[0063] The embodiments of this application are particularly applicable to zooplankton identification assessments under the following circumstances: significant differences in overall morphology between higher-level classifications, while differences between lower-level classifications are concentrated in local appendages; multiple similar classifications share some morphological features, requiring the combination of two or more identification sites to complete the distinction; key identification sites are small in size, easily obscured by the main body of the specimen, or damaged during mixed transportation; and application scenarios that require unified scoring and horizontal comparison of identification results from different laboratories or different assessment cycles.

[0064] It is worth noting that the assessment method disclosed in this application is not limited to copepods, but can also be applied to jellyfish, chaetognaths, cladocerans, and other zooplankton that rely on local morphological structures for classification. For different groups, appropriate identification sites can be selected according to their actual classification characteristics, so that the local structure presented by the assessment sample is consistent with the corresponding classification judgment process, thereby providing a basis for the standardization and traceable evaluation of zooplankton morphological identification capabilities.

[0065] refer to Figure 1 , Figure 1 This is an exemplary application scenario diagram provided for an embodiment of this application.

[0066] like Figure 1 As shown, the assessment organizer distributes pre-prepared and classified assessment samples to the assessment laboratory. Each assessment sample has a corresponding assessment sample number and includes the zooplankton specimen body for presenting its overall morphology and at least one identification site for presenting local identification features. After receiving the assessment samples, the assessment laboratory has assessors independently observe the specimen body and each identification site under preset assessment conditions. At the end of the assessment process, they submit a unified identification result for each assessment sample. It should be noted that the assessment system in this embodiment does not provide standard classification information, identification prompts, or result feedback before the assessors submit their results, to avoid system interaction interfering with the assessors' independent identification process.

[0067] The identification results uploaded by assessors can include classification results determined for the main body of the specimen, as well as feature recognition results for different identification sites within different sample locations. The classification results can include one or more classification levels determined by the assessors, and the feature recognition results represent the assessors' judgments on the morphological structure, quantity relationships, proportional relationships, segmentation status, attachment structures, or terminal morphology of the corresponding identification sites. When uploading identification results, the assessor's identifier, the assessment sample number, the sample location number, and the corresponding identification results are linked to ensure that data submitted by different laboratories can accurately correspond to the same standard assessment sample and its specific identification site.

[0068] After receiving the identification results, the assessment system retrieves the pre-bound standard classification path for the assessment sample based on the sample number, as well as the identification location, identification features, and classification level information corresponding to each sample position. The assessment system first determines the degree of classification matching between the classification result submitted by the assessor and the standard classification path. Then, combining this with the feature recognition results corresponding to each sample position, it judges whether the lower-level classification conclusion submitted by the assessor is supported by the corresponding morphological features, and determines the sample score for the assessment sample accordingly. For multiple assessment samples, the assessment system summarizes the scores of each sample to generate the assessor's assessment score and the corresponding result record.

[0069] thus, Figure 1 The application scenario shown forms a closed loop of "distribution of physical assessment samples - independent identification by assessment laboratories - unified uploading of assessment results - centralized processing by the assessment system". This scenario retains the professional assessment method of zooplankton morphological identification relying on actual specimen observation, while enabling the main body of the specimen, the local identification parts and their corresponding identification results to enter a unified information processing link after the assessment, thereby providing a data foundation for comparing results and evaluating capabilities among different assessors, different laboratories and different assessment cycles.

[0070] refer to Figure 2 , Figure 2 This is another exemplary application scenario diagram provided for the embodiments of this application.

[0071] exist Figure 1 On this basis, Figure 2 A further teaching system is provided, which establishes a data interaction relationship with the identification result uploading process. This system is used to obtain the classification matching results, node support status, sample scores, and related feature verification information generated by the assessment system after the examiners have completed the identification of all specimens and submitted the results. Based on this information, the system provides corresponding post-examination teaching content to the examiners. It should be noted that the teaching system does not provide standard classification information, key identification site prompts, or morphological feature descriptions to the examiners before the submission of the assessment results. This is to ensure that the teaching process and the assessment process are isolated in time, preventing the teaching information from affecting the authenticity of the assessment results.

[0072] Understandably, the assessment is not the ultimate goal. Because zooplankton identification involves multiple taxonomic levels, and the morphological criteria relied upon at each level differ, even if assessors receive low scores, it's difficult for them to pinpoint their specific weaknesses based solely on the final answer. For example, an assessor might correctly identify the major category based on the specimen's main body but fail to identify the key identifying features located at a particular sample location; they might have observed the corresponding identifying features but misjudged their segmentation relationships, terminal morphology, setae arrangement, or proportional characteristics; or they might correctly identify local features but fail to establish a proper correspondence between those features and the corresponding lower-level taxa. All of these situations may result in incorrect final classification results, but the corresponding skill deficiencies and subsequent training needs differ.

[0073] Traditional assessment results typically only include total score, accuracy rate, or incorrect species names. While these reflect whether assessors have met the pre-set assessment requirements, they fail to pinpoint whether errors stemmed from omitted specimen parts, misidentification of morphological features, or incorrect classification relationships. This is particularly problematic for zooplankton such as copepods, which rely on their fifth thoracic leg, caudal fork, antennae, or other minute appendages for sub-classification. Simply presenting assessors with standard species names often fails to enable them to accurately grasp the substantial differences between their current identification results and the standard results, and also hinders the development of targeted review strategies.

[0074] Based on this, in this embodiment, after the assessor uploads the assessment results, the assessment system not only generates the corresponding assessment score, but also retains the classification matching level, the classification node to be supported, the target support position group, and the feature matching status corresponding to each sample position for each assessment sample. Based on the above information, the teaching system determines the identification parts that the assessor did not identify or identified conflicting parts in each assessment sample, and retrieves the standard morphological information associated with the identification parts, the difference information of similar classification objects, and the identification description of the corresponding classification level, generating a teaching report for the assessor.

[0075] In one example, if the examiner correctly identifies the target specimen as belonging to the copepoda, but makes an error at the genus-level classification node, and the identification result of the fifth thoracic foot in the target support location group does not match the standard discriminant feature, the teaching system can identify the fifth thoracic foot as the identification site to be reviewed. After the assessment, the system will display the standard morphology of this identification site, its structural differences from similar taxa, and the role of this feature in the classification path. If the feature identification result of the fifth thoracic foot is correct, but the final genus-level classification result is incorrect, it further indicates that the examiner has obtained the corresponding morphological evidence, but has not correctly established the correspondence between morphological features and taxa.

[0076] refer to Figure 3 , Figure 3This application provides a schematic diagram of a zooplankton specimen information interaction system based on assessment samples. The assessment samples include the main body of the zooplankton specimen and identification sites located at different sample locations. Each sample location is associated with identification features of a corresponding taxonomic level. The system includes:

[0077] The data upload module is used to obtain the classification results submitted by the assessors for the assessment samples and the feature recognition results corresponding to the positions of each sample.

[0078] The data processing module is used to match the classification result with the standard classification path corresponding to the assessment sample, determine the classification matching level, determine the corresponding minimum support position group, and determine the node support status based on the feature recognition results corresponding to the position of each sample in the minimum support position group. The minimum support position group includes multiple classification nodes to be supported, and the classification nodes to be supported are determined by filtering the assessment classification path corresponding to the classification result.

[0079] The score generation module is used to determine the sample score of the assessment sample based on the classification matching level and the node support status of each of the classification nodes to be supported, and to generate an assessment score based on the sample scores of multiple assessment samples.

[0080] The data processing module includes:

[0081] The classification path processing unit is used to parse the classification result, generate an assessment classification path formed by sequentially connecting classification nodes of different classification levels, match the assessment classification path with the standard classification path, determine the classification matching level, and determine at least one classification node to be supported from the assessment classification path.

[0082] The support location determination unit is used to generate a candidate support location group for each of the classification nodes to be supported, based on the difference identification features between the classification node to be supported and the sample locations associated with each difference identification feature, and to determine the target support location group from the candidate support location group.

[0083] The support status determination unit is used to determine the node support status of the corresponding classification node based on the feature recognition results corresponding to the sample positions within the target support location group.

[0084] The data processing module is configured with a classification evidence processing strategy, which includes path filtering logic, location filtering logic, and evidence verification logic. The path filtering logic is configured in the classification path processing unit, the location filtering logic is configured in the support location determination unit, and the evidence verification logic is configured in the support status determination unit.

[0085] The path filtering logic is used to filter classification nodes located below the preset evidence verification level from the assessment classification path according to the preset evidence verification level, so as to obtain the classification nodes to be supported.

[0086] The score generation module includes:

[0087] The classification score determination unit is used to determine the initial classification score from the preset starting classification level to the classification matching level based on the preset level score for each classification level; determine the corresponding support coefficient based on the node support status of each node to be supported; and adjust the initial classification score based on the support coefficient, wherein the support coefficient corresponding to complete support is greater than the support coefficient corresponding to partial support, and the support coefficient corresponding to no support is less than the support coefficient corresponding to partial support.

[0088] The feature score determination unit is used to determine the feature score based on the matching status of the feature recognition result and the corresponding identification feature corresponding to each sample position, and when the same sample position is associated with multiple classification nodes to be supported, the feature score corresponding to the sample position is included in one of the classification nodes to be supported.

[0089] A sample score generation unit is used to determine the sample score based on the adjusted classification score and the feature score.

[0090] The system also includes a teaching report generation module, which is used to generate a teaching report for the examiner based on the sample score.

[0091] Next, with reference to the accompanying drawings, we will further elaborate on the method for information exchange of zooplankton specimens based on assessment samples provided in the embodiments of this application. It is understood that... Figure 4 The method shown can be configured as a program instruction within the aforementioned zooplankton specimen information interaction system based on assessment samples, and the method includes:

[0092] S1: Obtain the classification results submitted by the assessors for the assessment samples and the feature recognition results corresponding to each sample position;

[0093] In this embodiment, the examiners independently observe the test samples in the examination laboratory and submit the identification results in one go after completing all observations and classification judgments. The identification results include the classification results for the main body of the specimen and the feature identification results for one or more sample locations. The classification results can be filled in according to the preset classification hierarchy, such as group, order, family, genus and species in sequence. The feature identification results are submitted together with the sample location number. For example, location P1 corresponds to the fifth thoracic leg, and the feature identification results may include the number of segments, the number of terminal spines, the left-right structural relationship and local morphology; location P2 corresponds to the caudal fork, and the feature identification results may include the length-to-width ratio, the number of setae and their arrangement.

[0094] Those skilled in the art will understand that feature recognition results can be submitted in the form of options, numerical values, structured descriptions, or graphical symbols. The specific data format can be set according to the morphological characteristics of the identification part, as long as it can at least represent the examiner's observation and judgment on the corresponding identification part. This application does not impose any further limitations on this.

[0095] S2: Match the classification results with the standard classification path corresponding to the assessment sample to determine the classification matching level;

[0096] In this embodiment, a pre-registered standard classification path is retrieved based on the assessment sample number, and the classification results submitted by the assessors are parsed into assessment classification paths. For classification results represented by Chinese names, Latin scientific names, abbreviations, or historical synonyms, they can first be converted into unified classification nodes according to the name mapping table, and then matched level by level from high to low according to the classification hierarchy. The classification hierarchy corresponding to the lowest level classification node of consecutive matching is determined as the classification matching hierarchy.

[0097] For example, the standard classification path includes "crustaceans-copepods-target family-target genus-target species" in sequence. If the classification results submitted by the examiners are consistent with the standard classification path at the target family level and above, but the genus-level classification names are different, then the level corresponding to the target family will be determined as the classification matching level.

[0098] Furthermore, if the assessor only submits results up to a certain intermediate classification level, such as only confirming the family level and not continuing to submit genus and species levels, then the lowest-level classification node actually submitted can be taken as the path termination node, distinguishing it from the case of submitting incorrect lower-level classification nodes. The former indicates insufficient classification depth, while the latter indicates that a lower-level judgment has been made but the judgment result is biased.

[0099] S3: For the unsupported classification nodes in the assessment classification path corresponding to the classification results, determine the corresponding minimum support position group based on the identification features between each unsupported classification node and the similar classification objects and the associated sample positions;

[0100] In this embodiment, evidence verification levels can be pre-set, and classification nodes below the evidence verification level in the assessment classification path can be identified as classification nodes to be supported. For example, the overall morphology of the specimen body can usually support copepod or order-level classification, while genus-level and species-level classifications rely on more detailed appendage structures. Therefore, classification nodes can be required to have local identification evidence support starting from the genus or family level and below.

[0101] For each node to be supported for classification, similar class objects sharing the same parent class node are identified, and the discriminative features between the node to be supported and each similar class object are extracted. Then, based on the correlation between the discriminative features and the sample location, the similar class objects that can be distinguished at each sample location are determined. For example, location P1, where the fifth thoracic foot is located, can distinguish the target class node from similar objects B and C; location P2, where the tail fork is located, can distinguish the target class node from similar object D. Therefore, locations P1 and P2 together form a support location combination that excludes all similar class objects.

[0102] S4: Determine the node support status based on the feature recognition results corresponding to the positions of each sample within the minimum support position group;

[0103] In this embodiment, for each classification node to be supported, the feature recognition results corresponding to each sample position within the minimum support position group are matched with the differential identification features of the classification node at the corresponding identification site. If all feature recognition results within the group are consistent with the corresponding differential identification features, it indicates that the classification node submitted by the examiner has complete morphological evidence support, and the node support status is determined to be fully supported.

[0104] If there are no conflicting feature identification results, and only the feature identification results for some sample locations are consistent with the corresponding difference identification features, while the feature results for the remaining sample locations are not filled in or are explicitly marked as unidentifiable, then the node support status is determined to be partial support. Partial support indicates that the assessor has identified some key structures, but has not yet formed a complete chain of evidence that can exclude all similarly classified objects.

[0105] If the feature identification result of any sample location conflicts with the differential identification feature corresponding to the classification node to be supported, or if the feature identification result is consistent with the identification feature of a similar classification object, then the node support status is determined to be unsupported. For example, if the examiner submits the name of the target genus, but their description of the fifth thoracic foot distal structure only conforms to similar genera, then the target genus node does not have corresponding morphological support. By distinguishing between complete support, partial support, and no support, three capability states can be identified respectively: "key identification evidence has been mastered," "only partial identification evidence has been mastered," and "classification conclusion contradicts observational evidence."

[0106] It's important to clarify that "unidentified" and "identification conflict" have different meanings. "Unidentified" typically indicates that the assessor missed a corresponding part or couldn't determine the local morphology, and should not be directly equated with incorrect identification. "Identification conflict," on the other hand, means that the assessor has made a clear judgment, but the result supports other classifications. Treating these two separately avoids using the same evaluation method for incomplete observations and misinterpretations of morphological structures.

[0107] S5: Determine the sample score of the assessment sample based on the classification matching level and the node support status of each of the classification nodes to be supported, and generate an assessment score based on the sample scores of multiple assessment samples.

[0108] In this embodiment, corresponding level scores are pre-set for different classification levels. The initial classification score is determined based on the continuously matched classification nodes from the preset starting classification level to the classification matching level. For classification nodes below the evidence verification level that have already matched the standard classification path, a support coefficient can be configured based on the node's support status. Complete support corresponds to a higher support coefficient, partial support corresponds to a lower support coefficient, and no support corresponds to a support coefficient lower than that of partial support. Therefore, even if the lower-level classification name submitted by the evaluator is consistent with the standard result, but does not form complete identification evidence, the score for the relevant classification level will not be calculated based on the complete mastery status.

[0109] Feature scores are determined based on the matching status between the feature recognition results corresponding to each sample location and the standard discrimination features. For correctly identified discrimination parts, the corresponding feature score can be included; for unidentified or conflicting discrimination parts, no score is included or the corresponding score is reduced according to preset rules. When the discrimination feature carried by the same sample location is associated with multiple classification nodes, the corresponding feature score is only included in one of the classification nodes to prevent the same local structure from being scored repeatedly because it supports multiple classification levels.

[0110] For example, if an assessor correctly identifies the genus at the departmental level but incorrectly classifies it at the genus level, yet correctly identifies the fifth thoracic foot feature but fails to identify the tail fork feature, then the scores for the departmental level and above can be retained, and the score for the feature corresponding to the fifth thoracic foot can be included. If the genus-level classification does not receive the corresponding level score, the genus-level node can be determined as partially supported. If another assessor accidentally enters the correct genus name, but the identification results for the fifth thoracic foot and tail fork do not form complete support, then the genus-level score needs to be constrained by the support coefficient.

[0111] Next, we will further elaborate on the technical content of the method in this application regarding the test sample.

[0112] refer to Figure 5 , Figure 5 This is a schematic diagram of the packaging structure of the test sample provided in the embodiments of this application.

[0113] like Figure 5 As shown, the test sample packaging structure may include a substrate, a support layer, and a packaging layer arranged sequentially. The substrate provides an overall support base for the test sample, the support layer forms multiple independent sample locations, and the packaging layer covers the side of the support layer opposite to the substrate to seal and protect each sample location. The substrate, support layer, and packaging layer may be made of transparent materials or materials that are at least translucent in the sample observation area, allowing examiners to observe the internal specimen and identification areas using microscopic observation equipment without disassembling the packaging structure.

[0114] In this embodiment, the substrate is provided with a groove number printing area and a QR code printing area. The groove number printing area includes groove numbers corresponding to the positions of each sample, specifically groove number 1, groove number 2, groove number 3, and groove number 4. The QR code printing area is used to set information identifiers associated with the test samples. The information identifiers can be QR codes, barcodes, RFID tags, or other encoding forms that can be recognized by information terminals. The QR codes can be associated with the test sample number, preparation batch, standard classification path, specimen location in each groove, and corresponding identification features. However, during the assessment stage, according to access control, only the result upload portal is open to the assessors, and the standard classification information and standard identification information are not made available.

[0115] A support layer is disposed between the substrate and the encapsulation layer, and its surface has multiple mutually spaced grooves, each groove constituting a different sample location. Each groove can be used to accommodate a complete zooplankton specimen, a separated identification portion, or a control structure for auxiliary observation. For example, a complete copepod specimen can be placed in one groove, while the fifth thoracic leg, caudal fork, antennae, or other local structures used to distinguish similar taxa can be placed in other grooves. Thus, examiners can observe the overall morphology through the complete specimen, and can also independently observe smaller, concealed, or easily obscured identification portions within the corresponding grooves.

[0116] The outline shape and depth of each groove can be determined according to the shape and size of the specimen part to be fixed. For identification parts with slender structures, strip-shaped or elliptical grooves can be used; for specimens with a relatively compact overall outline, circular, elliptical, or irregularly shaped grooves that match the outline of the main body can be used. The depth of the groove should be sufficient to keep the specimen part relatively stable after sealing, while avoiding direct pressure from the sealing layer on the specimen, which could cause appendages to bend, overlap, or become distorted.

[0117] In this embodiment, the main body of the specimen and each identification part are fixed in their respective grooves using a preservation matrix adapted to their structural stability. For zooplankton with soft structures, high water content, or prone to morphological collapse, a gel-based preservation matrix can be used for support and fixation; for relatively stable parts such as crustacean appendages that need to maintain their local morphology for a long time, a transparent curing material can be used for embedding and fixation. The preservation matrix should not only meet the requirements for long-term specimen preservation but also maintain the observability of the corresponding identification features, avoiding obscuring key structures used for classification due to turbidity, air bubbles, or local shrinkage.

[0118] The encapsulation layer covers each groove on the support layer and forms a sealed connection with the support layer or substrate to isolate the specimen and preservation matrix from the influence of the external environment. The encapsulation layer can be made of transparent plastic sheet, glass plate, or other transparent sheet material, and its thickness and rigidity are determined according to the groove depth, preservation matrix type, and transportation conditions. After the encapsulation layer is connected to the support layer, each groove forms an independent closed preservation space, thereby preventing mixing, migration, or mutual obstruction between different identification sites, and reducing the possibility of specimen loss and damage during transportation and reuse.

[0119] It should be noted that, specifically, the method for determining the contour shape and accommodating depth of each groove based on the morphology and size of the specimen body and the target identification area to be fixed can refer to the size adaptation methods for microbial specimen embedding and micro-sample accommodating cavities. For example, the outer envelope size of the object to be fixed can be obtained through microscopic measurement, image contour extraction, or three-dimensional dimensional measurement, and a gap for the preservation matrix filling, encapsulation, and observation can be reserved based on the outer envelope size to determine the planar contour and depth of the corresponding groove. The groove only needs to be able to accommodate and restrict the corresponding specimen body or target identification area, and avoid excessive compression, movement, overlap, or obstruction of key identification features after encapsulation. The specific measurement method, gap value, and groove forming process can be selected according to the specimen type and encapsulation material, which will not be elaborated here.

[0120] In yet another example, the method further includes determining a corresponding preservation matrix based on the group attributes and structural stability of the specimen body and each target identification site, placing the specimen body and each target identification site in their respective sample positions and fixing them with the preservation matrix, and sealing each sample position with the transparent encapsulation layer.

[0121] It should be noted that the preservation matrix is ​​not only used to fix the specimen in the corresponding sample location, but should also take into account the tissue water content, surface structure, appendage fragility, and subsequent observation requirements of the zooplankton, while also considering morphological support, long-term preservation, and optical observation performance. Different zooplankton groups have significantly different tissue characteristics. If all specimens are preserved uniformly using liquid preservation or direct solidification and embedding, problems such as specimen shrinkage, local structural folding, turbidity of the embedding material, or obscuring of key identification features can easily occur. Therefore, based on the taxonomic attributes and structural stability of the main body of the specimen and the target identification site, aqueous gel-based preservation matrices or transparent resin-based preservation matrices can be selected respectively.

[0122] In this embodiment, for jellyfish and other specimens with soft tissues, high water content, and a tendency to collapse after being removed from a liquid environment, a gel-based preservation matrix containing a fixative can be used for fixation. For example, the fixed specimen is placed in a corresponding groove, and a preservation matrix containing agarose and formalin seawater is injected, allowing the gel to form a continuous support structure around the specimen. The agarose gel can restrict the displacement of tentacles, umbrella-shaped structures, and other soft structures while maintaining the aqueous environment around the specimen; the formalin seawater is used to maintain the fixed state, thereby reducing the possibility of folding, adhesion, or morphological collapse of the specimen during transportation and repeated observation.

[0123] For marine planktonic copepods and other crustacean specimens, an ethanol-replacement method combined with epoxy resin embedding can be used. These specimens are typically fixed in formalin first, and the fixed specimens still contain a significant amount of water and fixative components, often appearing milky white or opaque. If epoxy resin is directly inserted, the residual water can easily cause a turbid interface, microbubbles, or localized delamination between the resin and the specimen, potentially affecting resin curing and the clear presentation of local structures. Therefore, before epoxy resin embedding, ethanol can be used to gradually replace the residual water and formaldehyde solution inside and on the surface of the specimen, transforming it from an aqueous preservation state to a low-water-content state suitable for resin wetting.

[0124] Specifically, formalin-fixed specimens can be sequentially placed into ethanol solutions of increasing concentrations for gradient dehydration, such as 30%, 50%, 70%, 90%, and anhydrous ethanol, with the replacement time set according to specimen size, carapace thickness, and appendage integrity. Using progressively increasing ethanol concentrations prevents the specimen from shrinking violently, curling appendages, or becoming brittle due to rapid water loss. Once the specimen is dehydrated and fully impregnated with epoxy resin, it is placed in the corresponding groove, the orientation of the specimen or target identification area is adjusted, and then transparent epoxy resin is injected for embedding and curing.

[0125] Ethanol replacement and epoxy resin embedding can also improve the observation of marine planktonic copepods. The milky white appearance of formalin-fixed copepods is partly due to the optical differences between the internal aqueous medium and tissue structure of the specimen and the external observation medium. Incident light is scattered at different interfaces, making small segments and appendages difficult to identify. After dehydration with ethanol and thorough impregnation with transparent epoxy resin, the original aqueous components inside the specimen are gradually replaced, reducing the optical differences between the specimen and the embedding medium, and correspondingly reducing light scattering. This makes the originally milky white copepod specimens appear transparent or translucent, and their appearance is closer to their body color in the living state. As a result, segmental connections, appendage segmentation, setae arrangement, terminal spines, and other small distinguishing structures can be observed more clearly.

[0126] Understandably, epoxy resin in this embodiment serves both mechanical restraint and optical embedding functions. On one hand, the cured epoxy resin can prevent displacement, flipping, or loss of the main specimen and target identification sites such as the fifth thoracic leg, caudal fork, and antennae after dissection. On the other hand, the transparent resin can form a stable observation medium, reducing the impact of liquid sloshing and specimen posture changes on repeated observations. During embedding, the main observation surface of the target identification site can be aligned with the transparent encapsulation layer, and the resin filling height can be controlled to match the upper edge of the groove to avoid the encapsulation layer compressing the specimen or the resin layer being too thick, thus reducing the clarity of microscopic observation.

[0127] Those skilled in the art will understand that the ethanol concentration gradient, the replacement time at each stage, the type of epoxy resin, the curing conditions, and the embedding thickness can be adjusted according to the specimen's water content, size, shell hardness, and target identification characteristics. As long as the water and fixative are sufficiently replaced, the embedding material can be stably cured, and the morphological integrity and observability of the target identification site are maintained, this application does not impose any further limitations in this regard.

[0128] Next, we will further elaborate on the technical aspects of the classification matching hierarchy in this application.

[0129] In one example, the method for determining the classification matching level includes:

[0130] The classification results are analyzed to generate an assessment classification path formed by sequentially connecting classification nodes at different classification levels;

[0131] According to the classification hierarchy from high to low, the classification nodes in the assessment classification path are matched with the classification nodes in the standard classification path level by level;

[0132] The lowest-level classification node that is continuously matched is determined as the classification matching node, the classification level corresponding to the classification matching node is determined as the classification matching level, and the classification level located after the classification matching node in the standard classification path is determined as the classification level to be verified.

[0133] In this embodiment, the standard classification path represents the classification relationship of the assessment sample from the higher-level classification unit to the standard species, while the assessment classification path represents the classification relationship corresponding to the classification results submitted by the assessor. Classification nodes can be recorded using a unified classification identifier to avoid mismatches caused by differences in Chinese names, Latin names, synonyms, or abbreviations. During path matching, starting from the preset highest assessment level, the assessment classification nodes and standard classification nodes at the same classification level are compared sequentially, while simultaneously verifying the subordinate relationships between adjacent classification nodes.

[0134] If two category nodes at the current category level are identical and both belong to the previous matched category node, the current category level is considered a match, and the comparison continues to the next category level. If the current category nodes are different, the evaluation category path terminates at the current level, or the affiliation between the current category node and the previous category node is inconsistent, matching stops. The last consecutive identical category node before matching stops is the category matching node, and its level is the category matching level.

[0135] It is understandable that continuous matching is used instead of separate matching for each level because the classification path has a strict hierarchical dependency. If the evaluator has already selected an incorrect classification branch at the genus level, the species-level result following that incorrect branch cannot be independently determined as correct without considering its superior classification relationship.

[0136] After determining the classification matching level, the classification levels below the matching level in the standard classification path are identified as classification levels to be verified. For example, if the classification matching level is family level, then genus and species levels can be identified as classification levels to be verified. Instead of directly deeming these levels incorrect, the assessment team further evaluates whether the local feature recognition results submitted by the assessors support the corresponding classification nodes based on the discriminative features and sample locations corresponding to each classification level to be verified. This distinguishes between different situations such as lower-level classification errors, insufficient classification depth, and situations where local discriminative ability exists but a correct classification conclusion has not been reached.

[0137] The following is a specific example to aid understanding:

[0138] Suppose that the standard classification path corresponding to a certain assessment sample is as follows: Zooplankton - Arthropoda - Copepoda - Daphnia - Family A - Genus A - Species A. Here, the standard classification path represents the complete classification relationship corresponding to the assessment sample after pre-confirmation, with each classification node connected sequentially according to the classification hierarchy from highest to lowest.

[0139] After completing their observations, the examiners submitted the following classification result: Copepoda - Daphnia - Family A - Genus B - Species B. Parsing this classification result and combining it with pre-stored taxonomic relationships, a corresponding assessment classification path can be generated: Zooplankton - Arthropoda - Copepoda - Daphnia - Family A - Genus B - Species B.

[0140] Among them, the classification nodes corresponding to zooplankton, arthropoda, copepoda, daphnia, and family A are the same in sequence, and the subordinate relationship between adjacent classification nodes is consistent; when continuing to match genus-level classification nodes, the standard classification path corresponds to genus A, while the assessment classification path corresponds to genus B, and the two diverge, so the subsequent continuous matching is stopped.

[0141] In this case, the last consecutive identical classification node before the matching stops is family A. Therefore, family A is identified as the classification matching node, and the family level is identified as the classification matching level. This indicates that the assessors have correctly completed the continuous classification judgment from the preset starting classification level to the family level, but deviated from the standard classification path corresponding to the assessment sample starting from the genus level. The genus and species levels following family A in the standard classification path can be further identified as classification levels to be verified.

[0142] Furthermore, the classification level to be verified does not necessarily mean that all observations made by the assessor at the genus and species levels are incorrect. Rather, it indicates that their genus and species classification conclusions could not be confirmed through classification path matching. Subsequently, based on the distinguishing features between genera A and B, the corresponding identification sites and sample locations can be determined. Combined with the feature identification results submitted by the assessor for the corresponding sample location, it can be determined whether they have correctly identified some local morphological features. For example, if the main difference between genera A and B lies in the terminal structure of the fifth thoracic leg, the feature identification results for the sample location of the fifth thoracic leg can be further verified to distinguish whether the assessor failed to identify the key structure, incorrectly identified the key structure, or correctly identified the local features but incorrectly established the correspondence between the features and the classification objects.

[0143] For example, if the assessor only submits the classification to category A and does not continue to fill in the genus and species classification results, the resulting assessment classification path terminates at category A. In this case, category A is still identified as the classification matching node, and the category level is still identified as the classification matching level, but this means that the assessor correctly identified the category level and did not continue to make a lower-level classification judgment, rather than making an incorrect genus-level judgment.

[0144] It should be emphasized that the matching method in this application embodiment is applicable to situations where the assessor makes an error in the assessment classification path during the identification process. When the assessment classification path perfectly matches the standard classification path, but the feature recognition result has an error, since the classification nodes at each classification level have been continuously matched to the end of the standard classification path, there are no longer any classification levels to be verified after the classification matching nodes. Consequently, there are no unsupported classification nodes that require excluding similar classification objects using multiple sample positions. Therefore, there is no need to perform the process of determining the minimum support position group.

[0145] In this scenario, based on the sample location identifier, pre-associated discriminative features for each sample location can be retrieved, and the feature recognition results submitted by the examiner for the corresponding sample location can be directly matched with the discriminative features. If the feature recognition result matches the corresponding discriminative feature, the corresponding sample location is determined to be in a feature-matched state; if the examiner does not submit a valid recognition result or the submitted content is insufficient to determine the specific form, it is determined to be in a feature-unrecognized state; if the feature recognition result does not match the corresponding discriminative feature, it is determined to be in a feature-conflicted state. The matching status of each sample location can be further used to determine the feature score and generate post-exam teaching content, rather than being used to re-determine the classification matching level.

[0146] Next, we will further elaborate on the technical content of the method of this application regarding the minimum support position group.

[0147] It should be noted that the minimum support location group is used to characterize the minimum combination of sample locations that can form a complete morphological basis for the classification node to be supported. The subclassification of zooplankton is usually not directly determined by a single morphological feature; the same distinguishing feature may exist simultaneously in multiple closely related taxa. For example, the number of segments on the fifth thoracic leg can exclude some closely related genera, but two genera may still have the same number of segments; only by combining the morphology of the apical spine of the fifth thoracic leg or the length ratio of the caudal fork can the target genus be distinguished from the remaining closely related genera. Therefore, the distinguishing part carried by a single sample location may only have local distinguishing ability; only when the distinguishing features corresponding to multiple sample locations form a complementary relationship are they sufficient to support the corresponding classification node.

[0148] In this embodiment, for each classification node to be supported, firstly, closely related classification objects that share the same parent classification node and are easily confused during morphological identification are identified. These closely related classification objects can be different genera within the same family, different species within the same genus, or classification objects at different classification levels but with similar overall morphology. Subsequently, the distinguishing features between the classification node to be supported and each closely related classification object are determined, and each distinguishing feature is associated with the sample location carrying the corresponding identification site. Thus, each sample location not only represents an observation location in the specimen but also corresponds to a set of distinguishable objects, which represents closely related classification objects that can be excluded based on the distinguishing features at this sample location.

[0149] For example, a target genus of copepods may have three easily confused closely related genera. The location of the fifth thoracic leg can exclude the first and second closely related genera based on the morphology of the terminal spines; the location of the caudal fork can exclude the third closely related genera based on its length-to-width ratio; and the location of the first antennae can only exclude the second closely related genera again. In this case, the target genus cannot be identified solely based on the fifth thoracic leg, nor can it be identified solely based on the caudal fork. Using both the fifth thoracic leg and caudal fork locations together can cover all three closely related genera, thus forming a support location set supporting the target genus's taxonomic node.

[0150] Furthermore, not all sample locations related to the node to be supported need to be included in the support location group. If the fifth thoracic leg and the caudal fork have already eliminated all similar genera, adding the sample location of the first antenna, while increasing the number of available features, does not increase the actual distinguishing range for similar classification objects. If all relevant sample locations are used as classification criteria, not only will duplicate feature information be introduced, but a certain classification node may also be repeatedly supported due to multiple overlapping features, thereby amplifying the impact of local features on the assessment results. Based on this, from the sample location combination that can cover all similar classification objects, sample locations that do not play an independent distinguishing role are further removed to obtain the minimum support location group.

[0151] The term "minimum" in this embodiment does not simply mean a small number of sample locations, but rather that each sample location within the group plays an irreplaceable role in completing the classification. Specifically, after deleting any sample location from the minimum supporting location group, the set of distinguishable objects corresponding to the remaining sample locations will no longer be able to cover all closely related classification objects. For example, after deleting the location of the tail fork from the location group consisting of the fifth thoracic foot and the tail fork, the third closely related genus cannot be excluded; after deleting the location of the fifth thoracic foot, the first and second closely related genera cannot be excluded. Therefore, both sample locations are observation locations necessary for forming a complete classification basis.

[0152] When multiple location combinations satisfy the above conditions exist, further selection can be made based on the number of sample locations within the combination, the distinguishing stability of the discriminative features, the observability of the discriminative site, and the degree of information redundancy between different discriminative features. Combinations with fewer sample locations can be given priority; when the number of sample locations is the same, combinations that are more stable across different individuals, less affected by the specimen preservation state, and have a lower degree of redundancy with other features can be given priority.

[0153] Understandably, the minimum support location group essentially establishes a closed support relationship between the classification node and the necessary discriminative parts in the entity specimen. When subsequently determining the node support status, it is only necessary to obtain the feature recognition results corresponding to the sample locations within the group and determine whether these feature recognition results collectively meet the discriminative requirements of the classification node to be supported, without needing to perform indiscriminate processing on all discriminative parts in the assessment sample. This reduces the feature matching process unrelated to the current classification node, avoids the multiple inclusion of auxiliary and duplicate features, and ensures that the node support status reflects whether the assessor has formed complete morphological evidence sufficient to exclude similar classification objects.

[0154] It should be noted that a match between a feature recognition result and a standard discriminative feature does not necessarily indicate that the corresponding classification node has received complete support. Only when the discriminative features within the minimum support position group, which play different distinguishing roles, collectively meet the preset conditions, can it be said that the classification conclusion submitted by the assessor has complete morphological basis. Conversely, even if the final classification name is consistent with the standard classification result, if the necessary discriminative parts in the minimum support position group are not identified or feature conflicts occur, it still indicates that the corresponding classification conclusion has not yet formed complete support.

[0155] In one example, the method for determining the minimum support location group includes:

[0156] For each of the classification nodes to be supported, identify multiple similar classification objects that share the same parent classification node with the classification node to be supported, and determine the difference identification features between the classification node to be supported and each of the similar classification objects;

[0157] Based on the correlation between the differential identification features and the sample locations, the set of distinguishable objects corresponding to each sample location is determined, and the combination of sample locations that can cover all similar classification objects in the corresponding set of distinguishable objects is determined as the candidate support location group.

[0158] Based on the number of sample locations contained in each candidate support location group and the discriminative power of the corresponding differential features, a target support location group is determined from the multiple candidate support location groups, wherein the number of sample locations contained in the target support location group is no greater than the number of sample locations contained in other candidate support location groups.

[0159] In this embodiment, a discrimination matrix can first be established based on the distinguishing relationship between sample locations and similar classification objects. The matrix rows correspond to sample locations, and the matrix columns correspond to similar classification objects. When a differential feature associated with a sample location can distinguish the node to be supported from a similar classification object, the corresponding matrix element is recorded as a valid discrimination state. Based on the discrimination matrix, any of the following existing combined screening techniques can be used to determine the candidate support location group and the target support location group.

[0160] One implementation method is the combination enumeration method. Specifically, position combinations are generated sequentially, from fewest to most sample positions, consisting of one sample position, two sample positions, and then more sample positions. A union operation is then performed on the set of distinguishable objects corresponding to each position combination. When the union of a position combination covers all similarly classified objects, it is identified as a candidate support position group. After the first position combination that meets the coverage condition appears, the generation of combinations containing more sample positions can be stopped. Then, based on the discriminative power of the intra-group difference features, the target support position group is determined from an equal number of candidate support position groups. This method is suitable for situations with a small number of sample positions, directly obtaining position combinations containing the fewest possible positions.

[0161] Another implementation is the greedy set coverage method. During processing, the sample locations that can distinguish the most similar categories of objects are first selected. Then, from the remaining sample locations, the location that can add the most distinguishable objects is selected, and this selection operation is repeated until the set of distinguishable objects corresponding to the selected sample locations covers all similar categories of objects. For multiple sample locations with the same number of newly covered objects, priority can be ranked according to the discriminative power, recognition stability, or information redundancy of the corresponding difference features. This method does not require traversing all location combinations and is suitable for situations with a large number of sample locations or similar categories of objects, reducing the computational load of the combination selection process.

[0162] Another implementation method is the zero-one integer programming approach. A selection variable with a value of zero or one can be set for each sample location. The first optimization objective is to minimize the total number of selected sample locations, and the second optimization objective is to maximize the comprehensive discriminative power of the corresponding difference features. Simultaneously, a coverage constraint is set for each similar classification object, requiring it to be distinguished by at least one selected sample location. By solving the constraint relationship using an integer programming solver, branch and bound algorithm, or backtracking search algorithm, the location combinations that satisfy the full coverage condition can be obtained. For multiple feasible solutions containing the same number of sample locations, the discriminative power evaluation values ​​corresponding to each feasible solution can be further compared, and the location combination with the higher evaluation value can be determined as the target support location group.

[0163] Those skilled in the art will understand that the selection of candidate support location groups can also employ dynamic programming, heuristic search, or other set coverage solution methods. The specific algorithm can be selected based on the number of sample locations, the number of similar classification objects, and processing efficiency requirements. It is sufficient to select location combinations that cover all similar classification objects and contain a small number of locations from multiple sample locations. This application does not impose any further limitations on this.

[0164] Next, we will further elaborate on the technical aspects of the node support state in this application.

[0165] It should be noted that the node support status indicates whether a classification node submitted by the assessor has morphological identification criteria corresponding to its classification level. The node support status is not equivalent to whether the classification node is consistent with the standard classification node. Consistency only indicates that the assessor submitted the correct classification name, while the node support status further reflects whether the assessor has identified the identification features necessary to distinguish between similar classification objects and made a consistent judgment on the corresponding morphological features.

[0166] In this embodiment, each sample position in the minimum support position group plays a different distinguishing role. For example, the position of the fifth thoracic foot is used to distinguish the first group of similar classification objects, and the position of the tail fork is used to distinguish the remaining similar classification objects. Since removing any sample position from the minimum support position group will prevent the remaining sample positions from covering all similar classification objects, the feature recognition results corresponding to each sample position within the group are not interchangeable, and it is not advisable to use majority voting to determine the node support status. Even if the evaluator correctly identifies most of the distinguishing parts, if the necessary features for independent distinguishing are lacking, it cannot be considered that the corresponding classification node has formed complete morphological support.

[0167] In practice, the feature recognition results corresponding to each sample position within the minimum support position group can be matched with the corresponding differential identification features. Based on the matching results, each sample position can be divided into three position states: feature matching, feature not identified, and feature conflict. Feature matching indicates that the morphological judgment submitted by the assessor is consistent with the differential identification features of the corresponding classification node; feature not identified indicates that the assessor did not submit a corresponding judgment, selected "cannot be identified," or the submitted content is insufficient to determine the specific morphological state; feature conflict indicates that the assessor's explicit judgment is inconsistent with the corresponding differential identification features, or that the judgment result conforms to the identification features of a similar classification object.

[0168] When all sample positions within the minimum support position group are in a feature-matched state, the corresponding classification node to be supported can be determined as complete support. At this point, the distinguishing function of each sample position has been satisfied, and multiple differential features together cover all similar classification objects that are easily confused with the classification node to be supported. Therefore, a complete evidentiary relationship can be formed from the local morphological observation results to the corresponding classification node. Complete support does not mean that there are many features, but rather that all the necessary morphological evidence required to exclude similar classification objects is present.

[0169] When at least one sample position within the minimum support position group is in a feature-matched state, and the remaining sample positions are in a feature-unrecognized state, and there are no feature conflicts, the corresponding classification node to be supported can be determined as partially supported. Partial support means that the assessor has grasped some morphological information that can point to the corresponding classification node, but the existing feature recognition results cannot exclude all closely related classification objects. For example, the recognition result of the fifth thoracic foot can exclude two closely related genera, but the tail fork feature used to distinguish the third closely related genera has not been recognized. At this time, the existing observation results do not contradict the target genus, but are still insufficient to form the complete basis required for genus-level classification.

[0170] Feature non-identification and feature conflict have different meanings in morphological identification ability evaluation. Feature non-identification usually reflects that the examiner has not found the corresponding identification part, the observation conditions are insufficient, or the local structure cannot be confirmed yet. Other features submitted by the examiner may still correctly point to the classification node to be supported. Feature conflict, on the other hand, indicates that the examiner has made a clear judgment on the key morphology, but the judgment is inconsistent with the classification node to be supported. If the two are treated as one, it is easy to confuse "not fully observed" with "having formed a contradictory morphological judgment" into the same ability status, making it difficult to accurately reflect the stage at which the error occurred.

[0171] Therefore, when any sample location within the minimum support location group is in a feature conflict state, the corresponding classification node to be supported can be determined as unsupported. Even if the feature recognition results of other sample locations are consistent with the classification node to be supported, multiple matching features cannot offset the conflict of a necessary discriminative feature. The reason is that the conflicting sample location itself plays an independent role in excluding specific similar classification objects; when the morphological judgment submitted by the evaluator actually matches the similar classification object, a substantial contradiction has arisen between the current classification conclusion and its observed evidence, and continuing to recognize it as partially supported will not accurately reflect the relationship between the classification judgment and the morphological basis.

[0172] When all sample positions within the minimum support position group are in a feature-unrecognized state, the corresponding classification node to be supported can also be determined as unsupported. This situation does not mean that the classification name submitted by the assessor is necessarily wrong, but rather that the necessary morphological evidence to support the corresponding classification node cannot be found from the submitted feature recognition results. For example, if the assessor submits the correct genus name, but fails to identify genus-level distinguishing features such as the fifth thoracic leg and the caudal fork, then the consistency of the classification name cannot replace the lack of morphological evidence, and the relevant genus-level nodes do not have a verifiable supporting relationship.

[0173] By employing a tiered approach of full support, partial support, and no support, assessment results with the same final classification name but different observation processes can be further differentiated. Different node support states can be formed for correct classifications based on complete identification features, classifications based on only partial key structures, classifications where local morphological judgments contradict the classification name, and classifications lacking morphological basis but accidentally matching the standard name. Therefore, when determining sample scores subsequently, while retaining the classification depth reflected by the classification matching level, the completeness of the morphological basis of lower-level classification conclusions can be differentiated. This reduces the randomness generated when evaluating solely based on the final name and provides a data foundation for post-exam identification of unidentified parts, incorrect identification features, and classification association biases.

[0174] In one example, the sample score is determined in the following ways:

[0175] Based on the preset level scores for each category level, determine the initial category score from the preset starting category level to the category matching level;

[0176] The support coefficients are determined based on the node support status of each classification node, and the initial classification score is adjusted based on the support coefficients. The support coefficient for complete support is greater than that for partial support, and the support coefficient for no support is less than that for partial support.

[0177] The feature score is determined based on the matching status of the feature recognition result corresponding to each sample location and the corresponding identification feature. When the same sample location is associated with multiple classification nodes, the feature score corresponding to the sample location is included in one of the classification nodes.

[0178] The sample score is determined based on the adjusted classification score and the feature score.

[0179] Understandably, specific tiered scores can be pre-set during the assessment task setup, based on the identification difficulty of each tier, the complexity of the corresponding identification site, and the impact of misclassification on the monitoring results. Tiered scores represent the classification score an assessor can obtain by correctly progressing along the standard classification path to the corresponding tier, rather than simply setting a single score for the final classification name. For the same assessment task, all assessment samples use the same tiered score configuration to ensure the comparability of assessment results among different personnel.

[0180] In a specific example, the maximum score for a single assessment sample can be set to 100 points, with 70 points for classification and 30 points for characteristics. Taking the identification assessment of marine planktonic copepods as an example, the initial classification level is preset to subclass, and the scores for subclass, order, family, genus, and species are set to 10, 10, 15, 15, and 20 points respectively. Compared to higher-level classifications, genus and species levels typically require observation of finer appendage structures and exclusion of more closely related taxa, therefore, relatively higher scores can be assigned to these levels.

[0181] The scores for each classification level can be determined in advance. For example, a representative set of specimens that have already been classified and confirmed can be selected, and multiple personnel with experience in zooplankton identification can complete the identification of each classification level. The average time, error rate, and number of parts to be identified for each classification level can be statistically analyzed, and these indicators can be normalized to form a level difficulty value. If the level difficulty values ​​for subclass, order, family, genus, and species are 0.10, 0.12, 0.18, 0.23, and 0.37, respectively, then the scores can be proportionally converted while maintaining a total classification score of 70 points, and then adjusted to 10, 10, 15, 15, and 20 points respectively, taking into account the convenience of scoring. Those skilled in the art can adjust the specific scores according to the actual assessment objects, as long as the scores corresponding to the lower classification levels can reflect their identification depth and difficulty.

[0182] The initial classification score is the cumulative value of the scores at each level from the preset starting classification level to the classification matching level. For example, if the standard classification path includes subclass, order, family, genus, and species levels in sequence, and the classification results submitted by the examiner continuously match up to the genus level, but diverge from the standard classification path starting at the species level, then the classification matching level is genus level, and the initial classification score is 50 points. Therefore, even if the species-level result submitted by the examiner is incorrect, the correct classifications already completed from the subclass to the genus level are still retained, and the entire assessment sample will not be directly scored as zero due to an incorrect final classification result.

[0183] For classification nodes that require support from local discriminative features, the corresponding support coefficients are set according to the node's support status. In one example, the support coefficient for complete support can be set to 1.0, the support coefficient for partial support can be set to 0.6, and the support coefficient for no support can be set to 0.2. The support coefficients can be determined through calibrated questionnaires. For example, three types of questionnaires can be pre-constructed: those with all discriminative features correct, those lacking a necessary discriminative feature, and those with conflicting key features. Multiple evaluators can then assess the effective discriminative ability retained relative to the state of complete mastery. If the normalized average evaluation values ​​are close to 1.00, 0.62, and 0.18, respectively, then 1.0, 0.6, and 0.2 can be taken as exemplary support coefficients.

[0184] When adjusting classification scores, coefficients can be corrected separately for classification levels requiring morphological evidence verification, without applying the same support coefficient repeatedly to all higher-level classification scores. Specifically, the score of each classification level can be multiplied by its corresponding support coefficient and then summed; for higher-level classification nodes that do not require local discriminative feature support, the support coefficient can be set to 1.0.

[0185] For example, the aforementioned assessors were continuously matched down to the genus level. At the subclass and order levels, no local morphological evidence verification was performed, and the support coefficient was 1.0 for both. At the family level, the support was complete, with a support coefficient of 1.0; at the genus level, the support was partial, with a support coefficient of 0.6. The adjusted classification score was 44 points.

[0186] If the genus-level node is fully supported, all 15 points corresponding to the genus-level node are retained, and the adjusted classification score remains 50 points. If the genus-level node is not supported, the corresponding score for the genus-level node is adjusted to 3 points, and the adjusted classification score is 38 points. Therefore, the final assessment results, even with the same classification name but different levels of completeness in form, can result in different classification scores.

[0187] Feature scores can be pre-set according to the importance of each sample location. For example, the feature scores for the sample locations of the fifth thoracic leg, caudal fork, first antenna, and genital segment can be set to 10, 8, 6, and 6 points respectively, for a total of 30 points. The feature score for each sample location can be determined based on the number of closely related taxa it can distinguish, the associated taxonomic level, and its irreplaceable role in the identification process. Sample locations that can independently exclude multiple closely related taxa or directly determine genus- or species-level classifications can be assigned higher feature scores.

[0188] When a sample location corresponds to multiple specific discriminative features, the feature matching degree of the corresponding location can be determined based on the weight ratio of the matching discriminative features. Assuming a sample location is associated with five discriminative features, each with equal weight, four features are consistent with the standard discriminative features, and one feature is not identified, then the feature matching degree is 0.8. If there are no mutually exclusive key feature conflicts, a feature matching degree of 0.8 or higher can be defined as a matching state, 0.5 or higher but lower than 0.8 as a partial matching state, and lower than 0.5 as an insufficient matching state. If any exclusive key discriminative feature conflicts with the standard feature, it is directly defined as a conflict state.

[0189] The values ​​of 0.8 and 0.5 mentioned above can be obtained through pre-calibration. For example, in representative assessment data, the feature matching degree of the results submitted by personnel who correctly identified the corresponding positions is mainly distributed between 0.82 and 1.00, while the results submitted by personnel who failed to form stable identification results are mainly below 0.48. In this case, 0.8 and 0.5 can be used as the dividing values ​​between the matching state and the partial matching state, respectively. The specific threshold can be adjusted according to the identification location and the number of features, and this application is not limited to the values ​​mentioned above.

[0190] In a specific calculation example, the feature score for the location of the fifth thoracic leg is 10 points, with four of the five distinguishing features matching and one not identified, resulting in a feature matching degree of 0.8 and a corresponding feature score of 10 points. The feature score for the location of the caudal fork is 8 points, with three of the four distinguishing features matching and one not identified, resulting in a feature matching degree of 0.75 and a corresponding feature score that can be calculated as 8 × 0.75, or 6 points. The location of the first antenna has one exclusive key feature conflict, resulting in a corresponding feature score of 0 points. All distinguishing features for the location of the genital segment match, resulting in a corresponding feature score of 6 points. The feature score of the test sample is 22 points.

[0191] A single sample location may be associated with multiple classification nodes simultaneously. For example, the morphological features of the fifth thoracic foot can participate in the support determination of both genus-level and species-level nodes. To avoid duplicate scoring of the same observation, the corresponding feature score can be counted only in the classification node with the lowest association level and whose support status is not unsupported; if multiple classification nodes are at the same classification level, the score is counted in the classification node that has a direct distinguishing relationship with the corresponding discriminative feature. Regardless of how many classification nodes a sample location is associated with, its feature score is only accumulated once in a single evaluation sample.

[0192] Using the aforementioned values, the adjusted classification score is 44 points, the feature score is 22 points, and the sample score of the assessment sample is 66 points.

[0193] Using the above scoring method, the classification score indicates the depth to which the assessor correctly progressed along the classification path, the support coefficient reflects whether the corresponding classification node has the necessary morphological basis, and the feature score indicates the assessor's observation and identification of specific identification parts. Even if the final classification results are the same, assessment results that completely identify key identification parts, only identify some identification parts, or lack morphological evidence can still form distinguishable evaluation results through sample scores.

[0194] In yet another example, the teaching report includes at least:

[0195] The classification matching level, the node to be supported, the node support status, the classification score, the feature score, and the sample score corresponding to each assessment sample;

[0196] The target support location group corresponding to each of the nodes to be supported for classification, and the identification parts, difference identification features and evidence chain matching results in the target support location group;

[0197] The unidentified identification parts and conflict identification parts are determined based on the node support status of the unsupported or unsupported classification node, as well as the difference information between the unsupported classification node and similar classification objects.

[0198] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A zooplankton specimen information exchange system based on assessment samples, characterized in that, The assessment samples include zooplankton specimens and identification sites located at different sample locations. Each sample location is associated with identification features of a corresponding taxonomic level. The system includes: The data upload module is used to obtain the classification results submitted by the assessors for the assessment samples and the feature recognition results corresponding to the positions of each sample. The data processing module is used to match the classification result with the standard classification path corresponding to the assessment sample, determine the classification matching level, determine the corresponding minimum support position group, and determine the node support status based on the feature recognition results corresponding to the position of each sample in the minimum support position group. The minimum support position group includes multiple classification nodes to be supported, and the classification nodes to be supported are determined by filtering the assessment classification path corresponding to the classification result. The score generation module is used to determine the sample score of the assessment sample based on the classification matching level and the node support status of each of the classification nodes to be supported, and to generate an assessment score based on the sample scores of multiple assessment samples.

2. The zooplankton specimen information interaction system based on assessment samples according to claim 1, characterized in that, The assessment sample is a standardized physical assessment sample generated by separately encapsulating the main body of a zooplankton specimen and at least one identification site in the same carrier. The preparation method of the assessment sample includes: The system comprises a carrier substrate, a support plate, and an encapsulation layer connected in sequence. The support plate has multiple independent sample positions. The carrier substrate supports the support plate, and the encapsulation layer closes each sample position. The main body and identification area of ​​the target zooplankton specimen are fixed to the support plate, and the test sample is generated by sample encapsulation.

3. The zooplankton specimen information interaction system based on assessment samples according to claim 2, characterized in that, Multiple sample locations are formed in multiple grooves on the support plate, and the preparation method of the test sample further includes: Based on the shape and size of the specimen body and the identification part to be fixed, the outline shape and accommodating depth of each groove are determined, and position marks corresponding to each groove are set on the supporting substrate. The position marks include at least the test sample number and the sample position number.

4. The zooplankton specimen information interaction system based on assessment samples according to claim 2, characterized in that, Fixing the main body of the specimen and the identification site to different sample locations, including: Based on the group attributes and structural stability of the specimen body and each target identification site, the corresponding preservation matrix is ​​determined, the specimen body and each target identification site are placed in the corresponding sample position and fixed by the preservation matrix, and each sample position is sealed by the transparent encapsulation layer.

5. The zooplankton specimen information interaction system based on assessment samples according to claim 1, characterized in that, The data processing module includes: The classification path processing unit is used to parse the classification result, generate an assessment classification path formed by sequentially connecting classification nodes of different classification levels, match the assessment classification path with the standard classification path, determine the classification matching level, and determine at least one classification node to be supported from the assessment classification path. The support location determination unit is used to generate a candidate support location group for each of the classification nodes to be supported, based on the difference identification features between the classification node to be supported and the sample locations associated with each difference identification feature, and to determine the target support location group from the candidate support location group. The support status determination unit is used to determine the node support status of the corresponding classification node based on the feature recognition results corresponding to the sample positions within the target support location group. The data processing module is configured with a classification evidence processing strategy, which includes path filtering logic, location filtering logic, and evidence verification logic. The path filtering logic is configured in the classification path processing unit, the location filtering logic is configured in the support location determination unit, and the evidence verification logic is configured in the support status determination unit. The path filtering logic is used to filter classification nodes located below the preset evidence verification level from the assessment classification path according to the preset evidence verification level, so as to obtain the classification nodes to be supported.

6. The zooplankton specimen information interaction system based on assessment samples according to claim 5, characterized in that, The location filtering logic includes: For each of the classification nodes to be supported, identify multiple similar classification objects that share the same parent classification node with the classification node to be supported, and determine the difference identification features between the classification node to be supported and each of the similar classification objects; Based on the correlation between the differential identification features and the sample locations, the set of distinguishable objects corresponding to each sample location is determined, and the combination of sample locations that can cover all similar classification objects in the corresponding set of distinguishable objects is determined as the candidate support location group. Based on the number of sample locations contained in each candidate support location group and the discriminative power of the corresponding differential features, a target support location group is determined from the multiple candidate support location groups, wherein the number of sample locations contained in the target support location group is no greater than the number of sample locations contained in other candidate support location groups.

7. The zooplankton specimen information interaction system based on assessment samples according to claim 5, characterized in that, The evidence verification logic includes: The feature recognition results corresponding to each sample position within the target support position group are matched with the difference identification features of the classification node to be supported at the corresponding identification part to obtain the evidence chain matching result. If all feature recognition results within the target support location group match the corresponding difference identification features, the node support status of the classification node to be supported is determined to be complete support. If at least one feature recognition result matches the corresponding difference identification feature, and the remaining feature recognition results are unrecognized, the node support status is determined to be partially supported. If any feature recognition result conflicts with the corresponding difference identification feature, or if the conditions for determining complete support and partial support are not met, the node support status will be determined as not supported.

8. The zooplankton specimen information interaction system based on assessment samples according to claim 7, characterized in that, The score generation module includes: The classification score determination unit is used to determine the initial classification score from the preset starting classification level to the classification matching level based on the preset level score for each classification level; determine the corresponding support coefficient based on the node support status of each node to be supported; and adjust the initial classification score based on the support coefficient, wherein the support coefficient corresponding to complete support is greater than the support coefficient corresponding to partial support, and the support coefficient corresponding to no support is less than the support coefficient corresponding to partial support. The feature score determination unit is used to determine the feature score based on the matching status of the feature recognition result and the corresponding identification feature corresponding to each sample position, and when the same sample position is associated with multiple classification nodes to be supported, the feature score corresponding to the sample position is included in one of the classification nodes to be supported. A sample score generation unit is used to determine the sample score based on the adjusted classification score and the feature score.

9. The zooplankton specimen information interaction system based on assessment samples according to claim 8, characterized in that, The system also includes a teaching report generation module, which is used to generate a teaching report for the examiner based on the sample score.

10. The zooplankton specimen information interaction system based on assessment samples according to claim 9, characterized in that, The teaching report shall include at least: The classification matching level, the node to be supported, the node support status, the classification score, the feature score, and the sample score corresponding to each assessment sample; The target support location group corresponding to each of the nodes to be supported for classification, and the identification parts, difference identification features and evidence chain matching results in the target support location group; The unidentified identification parts and conflict identification parts are determined based on the node support status of the unsupported or unsupported classification node, as well as the difference information between the unsupported classification node and similar classification objects.

Images