A query method and apparatus for cross-block protection based on scoring
By configuring reference identifiers and scoring symbols for query blocks, generating an execution plan tree, and verifying slot consistency, the problem of inaccurate scoring in nested queries is solved, ensuring the accuracy of retrieval results.
Patent Information
- Application Number
- CN202611425328.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-15
- Publication Date
- 2026-11-17
AI Technical Summary
Traditional query methods often result in inaccurate scoring when queries are nested, leading to incorrect search results.
Configure a reference identifier and a score symbol for each query block. Generate an execution plan tree and statically bind the reference identifier and score symbol to the score-related operation nodes. Verify slot consistency to prevent cross-block calls and ensure accurate access to scores.
It improves the accuracy of query block scoring in nested query scenarios, prevents crosstalk between inner and outer scoring layers, and ensures the accuracy of retrieval results.
Smart Images

Figure CN123115222A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of retrieval and scoring technology, specifically relating to a query method and apparatus based on cross-block protection of scoring. Background Technology
[0002] With the continuous expansion of data scale in the era of big data and the increasing demand for real-time retrieval and personalized ranking from various applications, the scoring and ranking mechanism in database query engines has become a core element affecting the quality of system response.
[0003] In database full-text search and relevance scoring scenarios, when the query statement contains nested subqueries, such as relevance subqueries, LATERAL joins, CTE common table expressions, materialized nodes, etc., the current scoring-based query method has a serious cross-block layering defect.
[0004] Specifically, traditional query methods use a scoring function. After the scanning node completes its calculation, subsequent score-consuming nodes typically deduce the attribution of the current query block and the currently hit row from the execution context, such as expression context, tuple table slots, or system column identifiers, to obtain the score for the currently hit row. However, when queries are nested, the execution context may point to the wrong query block. This can cause an inner query block without a scoring predicate to read the outer score, or an outer query block without a scoring predicate to be contaminated by the inner score, resulting in an incorrect score. Consequently, incorrect search results are returned, leading to poor search performance. Summary of the Invention
[0005] This application provides a query method and apparatus based on score-based cross-block protection, which can solve the problem that traditional query methods cannot accurately obtain scores in cross-level query scenarios where the query statement is nested, resulting in poor query performance.
[0006] In a first aspect, embodiments of this application provide a query method based on scoring-based cross-block protection, the method comprising: Obtain the query statement and parse the query statement to obtain multiple query blocks; Generate a reference identifier and a score symbol for each query block; wherein the score symbol includes the weight, predicate information, and read boundary of the query block, and the read boundary is used to indicate the slot where the score of the query block is stored; An execution plan tree is generated based on the query statement, and the reference identifier and score symbol are statically bound to the operation nodes related to the score. The execution plan tree includes multiple operation nodes, and the operation nodes related to the score include scanning nodes and score consumption nodes. Based on the reference identifier and the scoring symbol, a retrieval query is performed according to the execution plan tree. During the execution of the scanning node, the scores generated are stored in the slot of the query block corresponding to the scanning node. When a score consumer node reads a score, it verifies whether the slot to which the read request belongs and the slot storing the score are the same object. Reading the score is only allowed if they are the same object; otherwise, it is considered a cross-block call, read is refused, and a default score value is returned. This ensures that inner query blocks without scoring predicates cannot read the scores of outer query blocks, and that the scores of outer query blocks without scoring predicates are not contaminated by the scores of inner query blocks.
[0007] In one possible implementation of the first aspect, generating an execution plan tree based on the query statement includes: The scoring call function in the query statement is rewritten as an explicit scoring reference statement based on the reference identifier, resulting in the rewritten query statement; The execution plan tree is generated based on the rewritten query statement.
[0008] In one possible implementation of the first aspect, performing the scan node based on the reference identifier and the scoring symbol includes: Determine whether a scoring predicate exists among the predicates bound to the scan node; If the scoring predicate exists, calculate the score of the current hit row and store the correspondence between the hit row and the query block corresponding to the scan node; If the scoring predicate is not present, the default score will be used as the score for the current hit row.
[0009] In one possible implementation of the first aspect, executing the score consumption node based on the reference identifier and the score symbol includes: Based on the predicate information recorded in the rating symbol of the query block corresponding to the rating consumption node, determine whether to return the default rating value; If the default rating value is not required, the rating is read from the slot of the query block corresponding to the rating consumption node according to the read boundary bound to the rating consumption node.
[0010] In one possible implementation of the first aspect, only the slot is read when the scoring consumption node is executed.
[0011] Secondly, embodiments of this application provide a query apparatus based on scoring-based cross-block protection, including a compiler and an executor; The compiler is used for: Obtain the query statement and parse the query statement to obtain multiple query blocks; Generate a reference identifier and a score symbol for each query block; wherein the score symbol includes the weight, predicate information, and read boundary of the query block, and the read boundary is used to indicate the slot where the score of the query block is stored; An execution plan tree is generated based on the query statement, and the reference identifier and score symbol are statically bound to the operation nodes related to the score. The execution plan tree includes multiple operation nodes, and the operation nodes related to the score include scanning nodes and score consumption nodes. The executor is used to perform a retrieval query based on the reference identifier and the scoring symbol, according to the execution plan tree. The scores generated during the execution of the scan node are stored in the slot of the query block corresponding to the scan node. When reading a score, the executor verifies whether the slot to which the read request belongs and the slot storing the score are the same object. Reading the score is only allowed if they are the same object; otherwise, it is considered a cross-block call, and reading is refused, returning a default score value. This ensures that inner query blocks without scoring predicates cannot read the scores of outer query blocks, and that the scores of outer query blocks without scoring predicates are not contaminated by the scores of inner query blocks.
[0012] The beneficial effects of this application's embodiments compared to existing technologies are as follows: This application configures a reference identifier and a scoring symbol for each query block, with the read boundary in the scoring symbol specifying the slot for storing the score, allowing each query block's score to be stored in its designated independent slot. After statically binding the reference identifier and scoring symbol to the scanning node and the scoring consumer node, the score generated during scan execution is directly written to the corresponding query block's slot based on the binding relationship. The consumer node also reads the score from the corresponding slot based on the same binding relationship after slot identity verification. This establishes a deterministic mapping between query blocks and scores, eliminating the need to infer scores from context. Even with nested subqueries, the access to scores for each query block remains isolated, preventing crosstalk between inner and outer scoring layers and ensuring accurate retrieval results. Attached Figure Description
[0013] Figure 1 A flowchart illustrating a query method based on scoring-based cross-block protection provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a query system based on scoring cross-block protection provided in an embodiment of this application. Detailed Implementation
[0014] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.
[0015] It should be understood that, when used in this application specification, the term "comprising" indicates the presence of the described feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof.
[0016] It should also be understood that the term “and / or” as used in this application specification means any combination of one or more of the associated listed items, as well as all possible combinations, and includes such combinations.
[0017] As used in this application specification, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [the described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [the described condition or event] is detected," or "in response to detection of [the described condition or event]."
[0018] Furthermore, in the description of this application, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0019] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.
[0020] The present application will be described in further detail below with reference to specific embodiments, but the implementation of the present application is not limited thereto.
[0021] Example 1 The query method based on scoring-based cross-block protection provided in this application can be applied to electronic devices such as mobile terminals, personal laptops, and supercomputers equipped with query systems. This application does not impose any restrictions on the specific type of electronic device.
[0022] Figure 1 This is a flowchart illustrating a query method for cross-block protection based on scoring, provided in an embodiment of this application. As an example and not a limitation, the method may include steps S101-S105, which are described below.
[0023] S101, retrieve the query statement.
[0024] For example, a user can enter a query term in the search box, such as "Apple phone", and then the query system will generate a query statement based on the query term, such as Structured Query Language (SQL).
[0025] S102, parse the query statement to obtain multiple query blocks.
[0026] For example, in an SQL statement, one SELECT statement corresponds to one query block.
[0027] For example, taking the SQL query statement involving a related subquery as shown in Table 1 below, two query blocks can be parsed from the SELECT statement. Specifically, the business intent of this query statement is to find documents that contain the "query term" and whose relevance score (i.e., rating) is higher than the score of its inner category.
[0028] Table 1 Query Statements
[0029] S103, generate the reference identifier and score symbol for each query block.
[0030] For example, the reference identifier corresponds one-to-one with the query block and is a unique identifier used to index the query block.
[0031] Optionally, if the query statement contains multiple table JOINs, multiple source LATERALs, UNIONs with outer score adjustments, or other queries with multiple source identities, an error can be reported directly to avoid confusion with the return of default score values.
[0032] For example, the scoring symbols include the weight of the query block, predicate information, and read boundaries.
[0033] Specifically, the weight of a query block refers to the total weight denominator used when calculating the score. A query block may contain multiple scoring conditions, such as title matching, body matching, and time decay. Each scoring condition has its own weight. When calculating the total score, the scores of each scoring condition need to be weighted according to their respective weights, divided by the total weight denominator, and then normalized.
[0034] Specifically, predicate information is used to indicate the scoring criteria of a query block, as well as the membership relationship between the scoring criteria and the query block.
[0035] Specifically, the read boundary is used to indicate the slot where the score of the query block is stored. This slot is a container object that stores a row of data when the executor is running, rather than an address range. Cross-block protection is implemented by verifying whether the slot to which the read request belongs and the slot where the score is stored are the same object.
[0036] S104, Generate an execution plan tree based on the query statement, and statically bind the reference identifier and score symbol to the operation node related to the score.
[0037] For example, the execution plan tree may include multiple operation nodes and indicate the execution order of these operation nodes.
[0038] For example, the operation nodes related to scoring include at least a scanning node and a scoring consumer node. The scanning node may need to calculate the score during execution, while the scoring consumer node needs to read the score during execution. For instance, the scoring consumer node may include a projection node, a connection node, a sorting node, etc.
[0039] In one example, the scoring function call in the query statement can be rewritten as an explicit scoring reference statement based on the reference identifier, resulting in the rewritten query statement; then, an execution plan tree can be generated based on the rewritten query statement.
[0040] Traditional methods directly call the scoring function and infer which query block's score to read based on the execution context. This can easily lead to calculation errors in nested, cross-level query scenarios. By changing the call operation to an explicit score reference, the score can be retrieved directly based on the reference identifier and score symbol during reading, eliminating the need for calculation and accurately locating the score, thus avoiding the acquisition of incorrect scores.
[0041] In one example, after obtaining the execution plan tree, the reference identifier and score symbol of the query block corresponding to each operation node can be statically bound to that operation node.
[0042] The above steps S101-S104 are executed during the compilation phase. By binding metadata such as reference identifiers and scoring symbols to the execution plan tree during the compilation phase, the metadata can be restored based on the execution plan tree in the subsequent execution phase, thus achieving lossless transfer of metadata such as reference identifiers and scoring symbols from compilation to execution.
[0043] S105, perform a retrieval query based on the execution plan tree according to the reference identifier and scoring symbol.
[0044] In one possible implementation, for a scan node, it can be first determined whether a scoring predicate exists in the scoring symbols bound to the scan node. If it exists, when executing the scan node, the score of the currently hit row can be calculated, and the correspondence between the identifier of the currently hit row and the reference identifier bound to the scan node can be stored, thereby realizing the storage of the correspondence between the score of the hit row and the query block. If no scoring predicate exists, a default score value, such as 0, can be returned as the score of the currently hit row.
[0045] For example, the current hit row refers to a data row in a candidate document matched from the database.
[0046] For example, a query block that includes a scoring predicate needs to calculate a score during execution.
[0047] For example, referring to the query statements shown in Table 1 above, if the outer query block has a scoring condition (i.e., a scoring predicate) like MATCH "query term", then this query block needs to be scored. If the inner query block only has the simple equality condition "d2.cat_id = d.cat_id" and no scoring predicate, then the nodes related to the inner query block do not need to be scored.
[0048] Specifically, scores can be calculated only when scanning nodes, and scores are not calculated for other nodes.
[0049] Specifically, for a scan node that requires a score, the calculated score can be stored in the slot of the query block corresponding to that scan node.
[0050] Optionally, for query blocks that do not include scoring predicates, a "zero-score boundary flag" can be created, causing the scoring read entry to short-circuit and return the default score value as a semantically correct result when the flag is detected.
[0051] In one possible implementation, when executing a score consumption node, it can first determine whether to return a default score value based on the predicate information recorded in the score symbol of the query block corresponding to the score consumption node. If it is not necessary to return a default score value, the score is read from the slot of the query block corresponding to the score consumption node.
[0052] For example, the query block corresponding to the operation node here refers to the query block to which the reference identifier bound to the operation node belongs.
[0053] Specifically, the query block corresponding to the scoring consumer node can be determined based on the reference identifier bound to the scoring consumer node, and then the predicate information recorded by the scoring symbol of the query block can be used to determine whether to return the default scoring value.
[0054] In one example, the zero-score boundary flag of the query block corresponding to the score-consuming node can be obtained first. If the flag is obtained, the default score value is returned directly, and a success report is given. If the flag is not obtained, the score is then read.
[0055] For example, when executing the query statement shown in Table 1, if the query block corresponding to a rating consumer node is an inner query block, it can directly return "0" as the rating value. If the query block corresponding to a rating consumer node is an outer query block, the rating can be read from that slot if the slot to which the read request belongs and the slot storing the rating are the same object.
[0056] Specifically, when returning the default score value, different codewords can be returned for the case of no score predicate and the exception case of uninitialized score storage. The former is returned as a semantically correct result and a success status is returned, while the latter is returned as a failure status and logged, so as to distinguish between the two cases.
[0057] In one example, the slot to which the read request belongs can be matched with the slot storing the score, and the score of the current hit row can be read from the slot storing the score only if the two are the same object.
[0058] For example, referring to the pseudocode shown in Table 2 below, the score value stored in the slot is returned only if the slot to which the read request belongs and the slot storing the score are the same object (i.e., the pointers of the two slots are exactly equal). If they are not the same object, it is determined to be a cross-block call, the read of the score is rejected, and the default score value is returned.
[0059] Table 2 Pseudocode
[0060] In another possible implementation, to address the situation where the scoring entry point cannot read the scores in the slots due to intermediate results being sorted and then processed by custom scan nodes or other planned configurations, the total score of the current hit row can be written to a hidden column after the scan node generates the score for each currently hit row. This hidden column is then bound to the hit row. Subsequent sorting only changes the scoring order but not the numerical score. When subsequent scoring consumer nodes execute, they can read the hidden column based on the column name to obtain the score of the current hit row.
[0061] This application configures a reference identifier and a score symbol for each query block, with the read boundary in the score symbol specifying the slot for storing the score, ensuring that the score for each query block is stored in its own designated independent slot. After statically binding the reference identifier and score symbol to the scanning node and the score consumption node, the score generated during scan execution is directly written to the corresponding query block's slot according to the binding relationship. The consumption node also reads the score from the corresponding slot after verifying slot identity based on the same binding relationship. This establishes a deterministic mapping between query blocks and scores, eliminating the need to infer scores from context. Even with nested subqueries, the access to scores for each query block remains isolated, preventing crosstalk between inner and outer score layers and ensuring accurate retrieval results.
[0062] Example 2 Figure 2 This is a schematic diagram of the structure of a query system based on score-based cross-block protection provided in an embodiment of this application. As an example and not a limitation, system 200 may include compiler 210 and executor 220.
[0063] In some embodiments, compiler 210 can obtain a query statement and parse it to obtain multiple query blocks. It then generates a reference identifier and a score symbol for each query block, generates an execution plan tree based on the query statement, and binds the reference identifier and score symbol to operation nodes related to the score. Executor 220 can perform a retrieval query based on the reference identifier and score symbol, using the execution plan tree.
[0064] For example, the scoring symbols include the weight of the query block, predicate information, and read boundaries.
[0065] Specifically, read boundaries are used to indicate the slots where the scores of the stored query blocks are located.
[0066] For example, the execution plan tree may include multiple operation nodes. Operation nodes related to scoring may include scan nodes and score consumption nodes.
[0067] In one possible implementation, compiler 210 can identify the scoring call function of the query statement and rewrite it into an explicit scoring reference statement based on the reference identifier to obtain the rewritten query statement, and then generate an execution plan tree based on the rewritten query statement.
[0068] In one example, compiler 210 can first identify the scoring call function in the query statement. If the query statement does not include a scoring call function, it can directly generate an execution plan tree based on the query statement, without needing to consider the calculation and reading of the score in subsequent processes.
[0069] In another example, if the query statement includes a scoring call function, the scoring call function can be rewritten, and all metadata, such as the list of reference identifiers for query blocks, the scoring symbol mapping table, and the scoring reference binding relationships, can be serialized and bound to the execution plan tree generated based on the rewritten query statement for executor 220 to recover.
[0070] In one possible implementation, when executing a scan node, executor 220 can first determine whether a scoring predicate exists in the predicate bound to the scan node. If it exists, it calculates the score of the currently hit row and stores the correspondence between the hit row and the query block corresponding to the scan node. If it does not exist, it directly uses the default score value as the score of the currently hit row.
[0071] For example, score storage can be used to ensure deduplication of score calculations, so that the same combination of hit row identifier, predicate, and execution round is only calculated once.
[0072] In one example, executor 220 can first obtain the identifier of the currently hit row, and write the score into the slot of the query block corresponding to the scan node through the processor's score callback or batch interface.
[0073] Optionally, for index-only paths, executor 220 can directly read from the index and calculate the final score, writing it to the materialized column and the reader's row score cache.
[0074] Optionally, for custom scan nodes, executor 220 can call the populate score column function to write the calculated total score into a hidden column of the result slot when calculating the score of the current hit row.
[0075] For example, column type validation can be performed before writing the total score in the hidden column to prevent it from pointing to a non-score column due to incorrect attribute number calculation.
[0076] In one possible implementation, when the executor 220 executes the score consumption node, it can first determine whether to return the default score value based on the predicate information recorded by the score symbol of the query block corresponding to the score consumption node; if the default score value is not returned, the score is read from the slot of the corresponding query block according to the read boundary it is bound to.
[0077] In one example, during the initialization phase, executor 220 can recover all metadata from the execution plan tree.
[0078] Optionally, since the start callback of the scan node is triggered internally when the executor 220 starts, if the reference identifier, scoring symbol, and the aforementioned correspondence of the query block are not obtained before starting the scan node, subsequent processing will be impossible. Therefore, the metadata can be stored in a global location before executing the scan node for use in the scan node's callback.
[0079] In one example, executor 220 can use a reader to read the score.
[0080] For example, for a query block without a scoring predicate, the reader can detect the zero-score boundary marker at the first step of the entry and short-circuit to return the default score value.
[0081] For example, for a query block with a scoring predicate, the scores can be read according to the priority chain.
[0082] Specifically, except at the scan node where the score needs to be calculated, the executor 220 only performs read operations on the slots where the scores are stored at other operation nodes.
[0083] This application configures a reference identifier and a score symbol for each query block, with the read boundary in the score symbol specifying the slot for storing the score, ensuring that the score for each query block is stored in its own designated independent slot. After statically binding the reference identifier and score symbol to the scanning node and the score consumption node, the score generated during scan execution is directly written to the corresponding query block's slot according to the binding relationship. The consumption node also reads the score from the corresponding slot after verifying slot identity based on the same binding relationship. This establishes a deterministic mapping between query blocks and scores, eliminating the need to infer scores from context. Even with nested subqueries, the access to scores for each query block remains isolated, preventing crosstalk between inner and outer score layers and ensuring accurate retrieval results.
[0084] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.
Claims
1. A query method based on score-based cross-block protection, characterized in that, include: Obtain the query statement and parse the query statement to obtain multiple query blocks; Generate a reference identifier and a score symbol for each query block; wherein the score symbol includes the weight, predicate information, and read boundary of the query block, and the read boundary is used to indicate the slot where the score of the query block is stored; An execution plan tree is generated based on the query statement, and the reference identifier and score symbol are statically bound to the operation nodes related to the score. The execution plan tree includes multiple operation nodes, and the operation nodes related to the score include scanning nodes and score consumption nodes. Based on the reference identifier and the rating symbol, a retrieval query is performed on the execution plan tree. During the execution of the scanning node, the rating generated is stored in the slot of the query block corresponding to the scanning node. During the execution of the rating consumption node, the rating is returned only when the slot of the read request matches the slot of the stored rating.
2. The method according to claim 1, characterized in that, The step of generating an execution plan tree based on the query statement includes: The scoring call function in the query statement is rewritten as an explicit scoring reference statement based on the reference identifier, resulting in the rewritten query statement; The execution plan tree is generated based on the rewritten query statement.
3. The method according to claim 1, characterized in that, Executing the scan node based on the reference identifier and the scoring symbol includes: Determine whether a scoring predicate exists among the predicates bound to the scan node; If the scoring predicate exists, calculate the score of the current hit row and store the correspondence between the hit row and the query block corresponding to the scan node; If the scoring predicate is not present, the default score will be used as the score for the current hit row.
4. The method according to claim 3, characterized in that, Executing the rating consumption node based on the reference identifier and the rating symbol includes: Based on the predicate information recorded in the rating symbol of the query block corresponding to the rating consumption node, determine whether to return the default rating value; If the default rating value is not required, the rating is read from the slot of the query block corresponding to the rating consumption node according to the read boundary bound to the rating consumption node.
5. The method according to claim 4, characterized in that, When executing the rating consumption node, only the slot is read.
6. A query device based on scoring-based cross-block protection, characterized in that, Includes compilers and executors; The compiler is used for: Obtain the query statement and parse the query statement to obtain multiple query blocks; Generate a reference identifier and a score symbol for each query block; wherein the score symbol includes the weight, predicate information, and read boundary of the query block, and the read boundary is used to indicate the slot where the score of the query block is stored; An execution plan tree is generated based on the query statement, and the reference identifier and score symbol are statically bound to the operation nodes related to the score. The execution plan tree includes multiple operation nodes, and the operation nodes related to the score include scanning nodes and score consumption nodes. The executor is used to perform a retrieval query based on the reference identifier and the rating symbol, according to the execution plan tree. During the execution of the rating consumption node, the rating is returned only when the slot of the read request matches the slot of the stored rating.
7. The apparatus according to claim 6, characterized in that, In the process of generating an execution plan tree based on the query statement, the compiler is specifically used for: The scoring call function in the query statement is rewritten as an explicit scoring reference statement based on the reference identifier, resulting in the rewritten query statement; The execution plan tree is generated based on the rewritten query statement.
8. The apparatus according to claim 6, characterized in that, During the execution of the scan node based on the reference identifier and the scoring symbol, the executor is specifically used for: Determine whether a scoring predicate exists among the predicates bound to the scan node; If the scoring predicate exists, calculate the score of the current hit row and store the correspondence between the hit row and the query block corresponding to the scan node; If the scoring predicate is not present, the default score will be used as the score for the current hit row.
9. The apparatus according to claim 8, characterized in that, During the execution of the scoring consumption node based on the reference identifier and the scoring symbol, the executor is specifically used for: Based on the predicate information recorded in the rating symbol of the query block corresponding to the rating consumption node, determine whether to return the default rating value; If the default rating value is not required, the rating is read from the slot of the query block corresponding to the rating consumption node according to the read boundary bound to the rating consumption node.
10. The apparatus according to claim 9, characterized in that, The actuator performs read operations only on the slots during the execution of the scoring consumption node.
