A method for processing extended k-combination anti-skyline query

Abstract

The application discloses a method for processing extended k-combination reverse skyline query, and is designed to accurately identify a combination with a base number k from a query object data set, wherein the combination can reversely dominate the most objects compared with all other combinations with the same base number. In addition, the application also introduces a parameter of combination reverse skyline domination rate to expand the query result set corresponding to the combination, so as to enrich the query result. The application solves the problem that the existing reverse skyline query technology cannot efficiently process specific combinations, optimizes the ability of identifying the most advantageous combination in the data set, and can expand the result set corresponding to the combination, so as to provide more efficient and rich recommendations for decision makers.

Description

Technical Field

[0001] This invention relates to the field of data query technology, specifically an extended k-combination reverse Skyline query processing method. Background Technology

[0002] Skyline queries play a crucial role in multi-objective decision-making, recommender systems, the internet, big data analytics and mining, and IoT data querying. With the development of big data and cloud computing technologies, the application scope and efficiency of skyline queries have been significantly enhanced. Skyline queries aim to select data items from a dataset that are optimal in at least one dimension and no worse than other data items in other dimensions.

[0003] Reverse Skyline queries are a variation of Skyline queries. Unlike Skyline queries, reverse Skyline queries start from the query point and find all objects that can control that query point. If Skyline queries focus on the perspective of ordinary users, then reverse Skyline queries focus on the perspective of decision-makers. Therefore, since their initial introduction, reverse Skyline queries have been widely used in fields such as business decision-making, sensor networks, and medical diagnostics.

[0004] Currently, traditional reverse Skyline query algorithms only consider the influence of a certain point on other points, without paying attention to the common influence of a group of points. They cannot perform reverse Skyline queries on a given combination, nor can they solve the problem of the small size of the result set after the query, and they cannot recommend combinations that can reversely dominate the most objects in the query dataset for decision-makers. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an extended k-combination anti-Skyline query processing method. Its purpose is to extend the traditional anti-Skyline query from a single-point-based query to a multi-point combination-based query, find the combination that can dominate the most users in the query object dataset, and expand the query result set according to the combination anti-Skyline dominance rate.

[0006] The technical solution adopted by the present invention to solve its technical problem is: an extended k-combination inverse Skyline query processing method, which mainly includes the following steps.

[0007] Step 1: Define the user dataset C = {c1, c2, ..., c...} m} and the query object dataset Q={q1,q2,…,q n}

[0008] Step 2: Calculate the dynamic Skyline set for each data point in the user dataset C. By comparing the size of the dynamic Skyline result set for each data point with the combination cardinality k, select combinations that may meet the conditions and add them to the candidate combination set CGSet. At the same time, add the data points that may constitute the result set to the candidate result set RSet.

[0009] Step 3: Construct a QR-GMap tree index structure based on the candidate combination set CGSet obtained in Step 2, and construct an R-tree index based on the candidate result set RSet.

[0010] Step 4: Using the QR-GMap tree and R-tree index, apply the k-combination inverse Skyline query method, iterate through the candidate combination set CGSet and the candidate result set RSet, and obtain the query results GSet and SSet. G。

[0011] Step 5: Based on the k-combination anti-Skyline query results obtained in Step 4, expand the result set by combining anti-Skyline dominance rate, sequentially adding i-level expansion point sets, and adding larger sub-combination expansion point sets in each level expansion point set until the target combination anti-Skyline dominance rate is reached.

[0012] Preferably, step two, which involves constructing the QR-GMap index, includes the following steps:

[0013] (1) Divide the space into 2 d Given a subspace, construct an empty quadtree, where d represents the depth of the quadtree.

[0014] (2) For all candidate combinations G in the candidate combination set i Construct an R-tree Rt of candidate combinations i .

[0015] (3) Leaf nodes within the same group jointly maintain hash table G. i A Map whose key corresponds to the combination of nodes, and whose value stores the combined Skyline query results.

[0016] Preferably, the k-combination reverse Skyline query method described in step four includes:

[0017] (1) Add the root nodes of the QR-GMap tree and the R tree to the two heaps respectively, and take out the top elements h and e of the heap respectively, and determine whether they are data points or non-data points.

[0018] (2) If e is a data point, and if e is dominated by the global Skyline point of the candidate combination, then discard e; otherwise, proceed to step (4).

[0019] (3) If e is a non-data point, decompose e and return to step (2).

[0020] (4) If h is a non-data point and a candidate combination node, the relationship is judged by the dynamic anti-domination domain formed by the current data point element e and the dynamic anti-domination domain, and partial results are obtained.

[0021] (5) If h is a non-data point and not a candidate combination node, continue to decompose h and then return to step (4).

[0022] (6) If h and e are both data points, the relationship is determined by the dynamic inverse dominance region of e to obtain partial results.

[0023] (7) Repeat the above steps until the intermediate query result GMap is obtained.

[0024] (8) Traverse all GMaps to obtain the k-combined inverse Skyline query results Gset and S G .

[0025] Preferably, in step (4) of the k-combination anti-Skyline query method, the method for determining the relationship between the dynamic dominating region and the dynamic anti-dominating region involves monitoring the dynamic dominating region and the dynamic anti-dominating region of node e, as well as the boundary relationship of node h, to determine whether node e can constitute the k-combination anti-Skyline query result of node h. Specifically, if the minimum boundary rectangle formed by the candidate combination node h is completely located within the dynamic anti-dominating region DADR(e) of node e, then node e can be added to the query result with h as the candidate combination. Conversely, if the minimum boundary rectangle of the candidate combination node h is not located within the dynamic anti-dominating region DADR(e) of node e, then node e should not constitute the query result corresponding to node h.

[0026] Preferably, in step (6) of the k-combination anti-Skyline query method, the relationship judgment method of the dynamic anti-dominance region determines whether node e can constitute the k-combination anti-Skyline query result of the candidate combination where node h is located by monitoring the dynamic dominance domain of node e and the boundary relationship of node h. If data point h is located within the dynamic dominance region DDR(e) of node e, then node e cannot constitute a query result with the combination where data point h is located as the candidate combination.

[0027] Preferably, the combined anti-Skyline dominance rate proposed in step five adopts the following formula:

[0028] in This represents the expanded combined reverse Skyline query result. express All points c x With all points q in G y The resulting anti-Skyline dominance point pair.

[0029] Overall, the above-described technical solutions conceived by this invention can achieve the following beneficial effects compared with the prior art.

[0030] (1) The extended k-combination inverse Skyline query processing method adopted in this invention can efficiently find a combination with cardinality k from the query object dataset, which has the property of maximizing the dominating object points. This method not only improves the accuracy of the query, but also speeds up the processing.

[0031] (2) Traditional reverse skyline queries often fail to effectively handle data combination problems. The method used in this invention successfully solves this problem, enabling users to perform in-depth analysis and queries on specific data combinations.

[0032] (3) This invention utilizes combined inverse Skyline dominance to dynamically expand query results. This method not only reduces time overhead but also enriches the content of the result set. This feature enables decision-makers to make more effective decisions based on more comprehensive and in-depth data analysis. Attached Figure Description

[0033] Figure 1 This is a flowchart of an extended k-combination reverse Skyline query processing method according to the present invention.

[0034] Figure 2 This is an example diagram of the QR-GMap index structure in a specific embodiment of the present invention.

[0035] Figure 3 This is an example diagram illustrating the dynamic dominance domain and dynamic anti-dominance domain determination method in a specific embodiment of the present invention. Implementation

[0036] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described examples are only a part of the embodiments of the present invention, and not all of them. Furthermore, it should be understood that after reading this invention, those skilled in the art can make various modifications or alterations to the present invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0037] like Figure 1 As shown, the present invention provides an extended combined reverse Skyline query processing method, which includes the following steps.

[0038] Step 1: Define the user dataset C = {c1, c2, ..., c...} m} and the query object dataset Q={q1,q2,…,q n}

[0039] Step 2: Based on the dominance relationship between the user dataset C and the query object dataset Q, obtain the dynamic Skyline set of each point in C. By comparing the size of the dynamic Skyline result set of each point in C and the combination cardinality k, select combinations that may meet the conditions and add them to the candidate combination set CGSet. At the same time, add the data points that may constitute the result set to the candidate result set RSet.

[0040] Specifically, the candidate combination set CGSet is the set of all dynamic Skyline query results that satisfy |DSL(c)|=k selected from the dynamic Skyline dataset; the candidate result set RSet is the set of all points that satisfy |DSL(c)|>k. c The set consisting of points c, where DSL(c) represents the dynamic Skyline set of point c.

[0041] Step 3: Construct a QR-GMap tree index structure based on the candidate combination set CGSet obtained in Step 2, and construct an R-tree index based on the candidate result set RSet.

[0042] Specifically, the construction steps of the QR-GMap tree are as follows.

[0043] (1) Divide the space into 2 d Given a subspace, construct an empty quadtree, where d represents the depth of the quadtree.

[0044] (2) For all candidate combinations G in the candidate combination set i Construct an R-tree Rt of candidate combinations i。

[0045] (3) Leaf nodes within the same group jointly maintain hash table G. i A Map whose key corresponds to the combination of nodes, and whose value stores the combined Skyline query results.

[0046] Specifically, the QR-GMap tree index structure is as follows: Figure 2As shown, the QR-GMap tree consists of a quadtree of depth 2 and five R-trees Rt0 to Rt4, and is associated with the five subspaces S0 to S4. Candidate combination R-trees RG1 to RG6 correspond to candidate combinations G1 to G6. Each leaf node of an R-tree maintains a hash table for the combination it belongs to. Taking Rt1 as an example, leaf nodes q5, q6, q7, and q8 jointly maintain a hash table G2Map, which uses the combination G2 = {q5, q6, q7, q8} as the key and the reverse Skyline query result of G2 as the value.

[0047] Step 4: Using the QR-GMap tree and R-tree index, apply the k-combination inverse Skyline query method, iterate through the candidate combination set CGSet and the candidate result set RSet, and obtain the query results GSet and SSet. G .

[0048] Specifically, the k-combination reverse Skyline query method includes the following steps.

[0049] (1) Add the root nodes of the QR-GMap tree and the R tree to the two heaps respectively, and take out the top elements h and e of the heap respectively, and determine whether they are data points or non-data points.

[0050] (2) If e is a data point, and if e is dominated by the global Skyline point of the candidate combination, then discard e; otherwise, proceed to step (4).

[0051] (3) If e is a non-data point, decompose e and return to step (2).

[0052] (4) If h is a non-data point and a candidate combination node, the relationship is judged by the dynamic anti-domination domain formed by the current data point element e and the dynamic anti-domination domain, and partial results are obtained.

[0053] (5) If h is a non-data point and not a candidate combination node, continue to decompose h, and then return to step (4).

[0054] (6) If h and e are both data points, the relationship is determined by the dynamic inverse dominance region of e to obtain partial results.

[0055] (7) Repeat the above steps until the intermediate query result GMap is obtained.

[0056] (8) Traverse all GMaps to obtain the final result Gset and S of the k-combined inverse Skyline query. G。

[0057] Specifically, the method for determining relationships in step (4) using the dynamic dominance region and the dynamic anti-dominance region is as follows: Figure 3 As shown, the dynamic dominance domain DDR(e) and dynamic anti-dominance domain DADR(e) of data point e are distinguished by different shades in the figure, where candidate combination nodes h j The corresponding minimum bounding rectangle falls entirely within DADR(e), therefore node e can be added to the query results with h as a candidate combination. Candidate combination node h i The corresponding minimum bounding rectangle falls entirely within DDR(e), therefore node e cannot constitute the query result corresponding to node h.

[0058] Specifically, the method for determining relationships through the dynamic anti-dominated region in step (6) is as follows: Figure 3 As shown, candidate combination h k There exists a point within the minimum bounding rectangle that lies within DDR(e), therefore node e cannot form a candidate combination h. k In the combined query results.

[0059] Step 5: Based on the k-combination anti-Skyline query results obtained in Step 4, expand the result set by combining anti-Skyline dominance. Add i-level expansion point sets sequentially, and add larger sub-combination expansion point sets to each level expansion point set until the target combination anti-Skyline dominance is reached, to obtain the final expanded combination anti-Skyline query processing result. This result includes a recommended combination and the corresponding expanded query result set.

[0060] Specifically, the combined anti-Skyline dominance rate is calculated using the following formula:

[0061] in This represents the expanded combined reverse Skyline query result. express All points c x With all points q in G y The resulting anti-Skyline dominance point pair.

[0062] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. An extended k-combination inverse Skyline query processing method, characterized in that, Includes the following steps: Step 1: Set the user dataset C = {c 1, c2,…,c m } and the query object dataset Q={q1,q2,…,q n }; Step 2: Calculate the dynamic Skyline set for each data point in the user dataset C. By comparing the size of the dynamic Skyline result set for each user data point with the combination cardinality k, select the subsequent combinations that meet the conditions and add them to the candidate combination set CGSet. Add the data points that subsequently constitute the result set to the candidate result set RSet. Step 3: Construct a QR-GMap index structure based on the candidate combination set CGSet obtained in Step 2, and construct an R-tree index based on the candidate result set RSet; Step 4: Using the QR-GMap tree and R-tree index, apply the k-combination inverse Skyline query method to obtain the Gset and S set of the k-combination inverse Skyline query results. G ; The specific steps of the k-combination reverse Skyline query method include: (1) Add the root nodes of the QR-GMap tree and the R tree to the two heaps respectively, and take out the top elements h and e of the heap respectively, and determine whether they are data points or non-data points; (2) If e is a data point, and if e is dominated by the global Skyline point of the candidate combination, then discard e; otherwise, proceed to step (4). (3) If e is a non-data point, decompose e and return to step (2); (4) If h is a non-data point and a candidate combination node, the relationship is determined by the dynamic anti-domination domain formed by the current data point element e and the dynamic anti-domination domain, and partial results are obtained. (5) If h is a non-data point and not a candidate combination node, continue to decompose h, and then return to step (4); (6) If both h and e are data points, the relationship is determined by the dynamic dominance region of e to obtain partial results; (7) Repeat the above steps until the intermediate query result GMap is obtained; (8) Traverse all GMaps to obtain the result Gset and S of the k-combination inverse Skyline query. G ; Step 5: Based on the k-combination anti-Skyline query results obtained in Step 4, sequentially add the k-combination anti-Skyline query results of the sub-combinations to the unexpanded result set according to the combination anti-Skyline dominance rate, until the target combination anti-Skyline dominance rate is met, to obtain the final expanded combination anti-Skyline query processing result. This result includes a recommended combination and the corresponding expanded query result set. The combination anti-Skyline dominance rate is calculated using the following formula: ,in This represents the expanded combined reverse Skyline query result. express All points c x With all points q in G y The resulting anti-Skyline dominance point pair.

2. The extended k-combination reverse Skyline query processing method according to claim 1, characterized in that, The steps for constructing the QR-GMap index described in step three include: (1) Divide the space into 2 d Given a subspace, construct an empty quadtree, where d represents the depth of the quadtree; (2) For all candidate combinations G in the candidate combination set i Construct an R-tree Rt of candidate combinations i ; (3) Leaf nodes within the same group jointly maintain hash table G. i A Map whose keys correspond to the combination of nodes, and whose values ​​store the combined Skyline query results.

3. The k-combination reverse Skyline query processing method according to claim 1, wherein the method for determining the relationship in the dynamic reverse domination region described in step four (6) is characterized in that if the data point h is in the dynamic domination region DDR(e) of e, then the point e cannot constitute a GMap with the candidate combination where the data point h is located as the key, and should be discarded.

Images