CN105005627A - Shortest path key node query method based on Spark distributed system - Google Patents
Shortest path key node query method based on Spark distributed system Download PDFInfo
- Publication number
- CN105005627A CN105005627A CN201510478276.5A CN201510478276A CN105005627A CN 105005627 A CN105005627 A CN 105005627A CN 201510478276 A CN201510478276 A CN 201510478276A CN 105005627 A CN105005627 A CN 105005627A
- Authority
- CN
- China
- Prior art keywords
- node
- shortest path
- label
- spark
- method based
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2453—Query optimisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2471—Distributed queries
Abstract
The invention discloses a shortest path key node query method based on a Spark distributed system. Through carrying out pretreatment on a total node sequence formed by ranking all nodes in a graph according to the key degree on a Spark platform, a hierarchical label corresponding to each node is obtained, shortest path key node query between any node pair can be realized, and key nodes with the specified number on the shortest path are returned. Thus, a large amount of access work is completed during the pretreatment stage based on the Spark platform, the occupied space in the case of search can be narrowed, and certain extension is realized.
Description
Technical field
What the present invention relates to is a kind of technology of field of information processing, specifically a kind of shortest path key point querying method based on Spark distributed system.
Background technology
Have the algorithm of many shortest path aspects at present, namely node s is designated as SP<s to the shortest path of node t, t>, and corresponding bee-line is dist<s, t>.Disjktra is the classic algorithm of graph theory signal source shortest path problem, but its efficiency can not meet mainstream demand.In recent years many relevant highly effective algorithms are emerged, such as A*, a kind of didactic shortest route for point to point Direct search algorithm; Such as combine terrestrial reference (landmark) thought, the ALT algorithm of application triangle length of side inequality constrain; The hub labeling algorithm (hub labeling, HL) that the people such as such as Abraham propose; And breviary hierarchical algorithms (Contraction Hierarchies, CH), a kind of pre-service shortest path first containing index structure, improves search efficiency by adding shortcut in pre-service.
In level hub label (Hierarchicalhub labeling) algorithm that the people such as Abraham propose, all nodes are sorted according to certain standard, make the level (level) that each node correspondence one is different.Each node u comprises two labels: forward direction label Lf (u) and reverse label Lr (u).For forward direction label, in the label of each node, store a binary set { (v
1, d (u, v
1)), (v
2, d (u, v
2)) ..., (v
x, d (u, v
x)), each two tuples are by distance d (u, the v between node vi and u and this node
i) (in reverse label, be then d (v
i, u)) and composition.The label of node u contains the segment connection information of u.Level hub label meets covering attribute: for two arbitrary node s and t, must have one s ?some w on t shortest path, it belongs to Lf (s) and Lr (t) simultaneously, and it is the node that the upper level of shortest path SP<s, t> is the highest.
Breviary hierarchical algorithms CH is proposed in 2008 by people such as Geisberger, is a kind of to be widely used in the shortest path first on all kinds of figure efficiently, and its main thought is the efficiency that structure by constructing a belt secondary index improves Shortest Path Searching.CH needs vertex sequence, such as a r (v).The pre-service of CH is the process of a continuous reduction figure, comprises multiple iteration, and current figure is reduced a part by each iteration, and adds corresponding virtual shortcut.According to total order order from low to high, node is conducted interviews, and successively node is deleted from figure.If delete figure be followed successively by G1 ?>...Gn.In delete procedure, the shortest path of deleted Vertex cover may be affected.Suppose that the current point being about to delete is v, its adjoint point that enters integrates as Nin (v), going out adjoint point integrates as Nout (v), to every a pair u ∈ Nin (v), w ∈ Nout (v), the figure deleting v adopts witness search for and (namely Shortest Path Searching is carried out to the adjoint point deleting point, determine each adjoint point between bee-line) obtain u ?w minor increment be d ' (u, w), if d ' (u, w) >edge (u, v)+edge is (v, w), so illustrate delete v to u ?w bee-line can have an impact.Therefore, need, deleting in figure remaining after v, to add the virtual shortcut (u, w) that a length is edge (u, v)+edge (v, w), so as not in remaining figure u ?bee-line between w change because of the deletion of v.
But in most of the cases do not need detailed, complete shortest path, only need to obtain parton path, thus some path summary algorithms become study hotspot.The k that people such as Yufei Tao 2011 propose at " On k ?skip Shortest Paths " ?skip algorithm be also a kind of path summary algorithm, to query point pair, can provide k ?skip shortest path P*: on actual shortest path, every k continuous print point just has a point at least in P*, that is, P* samples to P by the probability of at least 1/k.
In daily life, often by virtue of experience can judge the importance of each node on road network, i.e. its crucial degree.Key degree represents the significance level of node in road network, and the crucial degree of whole node can be regarded as a total order of all nodes, and the crucial degree of node v is expressed as r (v).
In recent years, along with the popularization of large data and distributional concept, multiple stage machine utilizes distributed platform carry out implementation algorithm and become an attractive target, Spark is an efficient distributed computing system based on internal memory, and Spark is the cluster Computing Platform rising in Univ California-Berkeley AMPLab.To the data needing iterative computation, Spark provides cache to operate, and by data buffer storage in internal memory, can save the expense of a large amount of reading and writing of files; Can direct read/write HDFS file system; In addition, Spark has numerous distributed computing framework, can carry out various Distributed Calculation flexibly.
RDD (Risilient Distributed Dataset, elasticity distribution formula data set) is the core of Spark, is an abstract concept of distributed memory.For developer, RDD can be regarded as an object in Spark, run in internal memory.RDD is enough to represent polytype calculating, comprises Map Reduce algorithm and special iteration programming model (as Pregel).
GraphX is the figure computing module of the special disposal figure based on Spark platform, and it stores for figure and figure calculating provides abundant API, and can realize parallel figure and calculate, namely a figure splits into several subgraphs, carries out iteration calculating stage by stage respectively to subgraph.
Distributed C H algorithm is proposed at " Parallel time ?dependent contraction hierarchies " by people such as Vetter, but is only realized on MPI by algorithm, lacks distributed memory system, parallel task management and distribution mechanism as support.
Through finding the retrieval of prior art, Chinese patent literature CN104518965A, day for announcing 2015.4.15, disclose a kind of shortest path query method and device, comprise: determine and each node in stored data base respectively belonging to the zone of convergency of each rank, and the shortest path of each node respectively and between the Centroid of the zone of convergency of each affiliated rank, and the shortest path between the Centroid of the zone of convergency of different highest level; According to the zone of convergency stored and shortest path, determine the shortest path between two nodes in inquiry request.But this technology needs layering aggregation and stores data, for given point to only there being two central points, need travel through all nodes during inquiry, consuming time longer.
Summary of the invention
The present invention is directed to prior art above shortcomings, a kind of shortest path key point querying method based on Spark distributed system is proposed, carry out based on the Distributed C H pre-service on Spark platform and level label configurations to all nodes in road network in pre-service, utilize pretreated result during inquiry, return the key point of specifying number to user.
The present invention is achieved by the following technical solutions:
The present invention by carrying out pre-service to all nodes in figure according to the node total order of crucial degree sequence composition on distributed Spark platform, obtain the level label that each node is corresponding, realize the shortest path key point inquiry between arbitrfary point pair, and return the key point of shortest path being specified number.
Described pre-service comprises: Distributed C H pre-service, structure and distributed storage level label.
Described Distributed C H pre-service comprises: set up independent sets and add virtual shortcut.
Described figure refers to: the figure that the point given by several and connection any two points are formed, and the point in figure is node.
Described crucial degree refers to: any one node significance level in the drawings, and its measurement index includes but not limited to: node degree and covering power.
Described independent sets refers to: from current figure, can reduce several nodes according to random order, and not affect the point set of the correctness of the bee-line of figure in following iteration.
Described virtual shortcut (shortcut) refers to: for any one point in figure to <s, t>, if to <s, every bit v in t> bee-line except s, t, the crucial degree of v is all lower than the crucial degree of s and t, then exist virtual shortcut s ?>t, its weights are the bee-line between <s, t>.
Described level label is two tuple-sets, and in two tuple-sets of each node, two tuples are made up of destination node and the distance vector of this node to destination node.
Described level label is according to the direction of distance vector, be further divided into forward direction label and reverse label, and meet the Covering property of level label, namely any one is put <s, the common factor of the forward direction label of t>, starting point s and the reverse label of terminal t must comprise the node that on their shortest paths, crucial degree is the highest.
Described key point refers to: the node that crucial degree is the highest on shortest paths or several nodes front of crucial degree descending sort.
Technique effect
Compared with prior art, the present invention carries out pre-service by Distributed C H pre-service and level label configurations to ordering point set total order, Spark platform and HDFS document storage system is utilized to obtain the support of an efficient Distributed Computing Platform and distributed memory system by algorithm, extensive work is completed at the pretreatment stage based on Spark platform, the level label obtained by pre-service, the space taken when reducing search, and there is certain expansion.
Accompanying drawing explanation
Fig. 1 is schematic diagram of the present invention;
Fig. 2 is pretreatment time figure in road network NWs8324 in embodiment;
Fig. 3 is pretreatment time figure in road network BAYs15164 in embodiment;
Fig. 4 is pretreatment time figure in road network Floridas52781 in embodiment;
Fig. 5 is pretreatment time figure in road network NWs111729 in embodiment.
Embodiment
Elaborate to embodiments of the invention below, the present embodiment is implemented under premised on technical solution of the present invention, give detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.
Embodiment 1
The Spark platform release that the present embodiment relates to is Hadoop2.4.1, Spark1.3.0; The compilation tool version of Spark platform is sbt0.13.4, program language version is Java1.7.0, scala2.10.5.Spark platform comprises the cluster of 9 worker compositions, and each worker has 6 cores, internal memory 18GB; 9 worker amount to 54 cores, internal memory 164GB.
The figure G that the present embodiment example adopts is from 9
ththe full figure of road network NWs8324, BAYs15164, Floridas52781 or NWs111729 of DIMACS or subgraph.
As shown in Figure 1, the present embodiment comprises the following steps:
Step 1, to figure G in all node v sort according to crucial degree, formed total order o (v).
Described road network is the network architecture be made up of with certain density and form traffic major trunk roads, subsidiary road and size branch road.
Described road network is expressed as oriented band non-negative weights figure G=(V, E), and wherein: V is set of node, E is limit collection.
Major trunk roads in described road network, subsidiary road and size branch road are respectively a limit in figure G, and major trunk roads, road junction between subsidiary road and size branch road are the node in figure G.
Described road network adopts HDFS distributed storage file system to carry out data storage.
The crucial degree of the node v in described road network is expressed as r (v).
Step 2, on Spark platform, based on GraphX, described total order is carried out to the Distributed C H pre-service of multiple iteration, obtain the figure comprising all virtual shortcuts, concrete steps comprise:
Step 2.1) mode of being reduced by iteration sets up the residual graph G ' obtaining independent sets and remove independent sets from figure G.
Described independent sets refers to: from current figure, can reduce several nodes according to random order, and not affect the point set of the correctness of the bee-line of figure in following iteration, and the independent sets in the figure Gi of i-th iteration is expressed as Ii.
Described independent sets is set up in the following manner: for each node v in figure G and its adjoint point Neighbor (v), if to any w ∈ Neighbor (v), there is r (v) >r (w), then node v is put into independent sets, and in current iteration, v is reduced.
Step 2.2) virtual shortcut is added to the node in residual graph G ': for any point v in the independent sets Ii in an iteration, u ∈ Nin (v) and w ∈ Nout (v), set a definite value λ, in figure G ' to the adjoint point of each some v in independent sets to <u, w> carry out λ ?jumping figure (λ ?hop) shortest path Witness search for, if the u obtained ?w bee-line d ' (u, w) limit (u is greater than, v) with (v, w) length sum edge (u, v)+edge (v, w), then need interpolation u ?virtual shortcut between w, the length of virtual shortcut is edge (u, v)+edge (v, w).
Described Nin (v) enters adjoint point collection for some v's, and Nout (v) goes out adjoint point collection for a v's.
The Shortest Path Searching of described λ ?jumping figure is that the Bellman ?Ford limiting jumping figure searches for.
Described bee-line is determined by Witness algorithm search.
Step 3, the figure G* tectonic remnant basin label comprising all virtual shortcuts that obtains according to step 2 utilize HDFS distributed memory system to store: to each node contained in the figure of all virtual shortcuts, namely root node carries out Distributed C H sweep forward and reverse search respectively, in the forward direction label node searched being stored into respectively root node and reverse label, as the level label of root node.
Step 4, inquire about based on the shortest path key point of level label, concrete steps comprise:
Step 4.1) to figure G in arbitary inquiry point to <s, t>, given s is starting point, and t is terminal, and k is for returning key point number, obtain a four-tuple (G, s, t, k), can by step 1 ?3 obtain the point set { v that several nodes on shortest path SP (s, t) are formed
1, v
2..., v
x, according to the covering attribute of level label, point is concentrated has at least a node w to belong to forward direction label Lf (s) of s and reverse label Lr (t) of t simultaneously, and is the node that the upper level of SP (s, t) is the highest, and namely w is key point.
Step 4.2) with key point w, path is split as (s, w) and (w, t), recursive call step 4.1 on subpath), several key points will be obtained.Sort according to the crucial degree of these key points, return a k unit key point group kTop=((v
1, d (s, v
1), (v
2,d (s, v
2)) ..., (v
k,d (s, v
k))), and by these point form compressed path kPath=s ?>v
i1... ?>v
ik... ?>t, v
1~ v
kfor SP (s, t) goes forward k key point, i1 ~ ik is the sequence number sorted to these key points according to actual shortest path position.
Definition V (kPath) is the node set on kPath, and nodes is | V (kPath) |, if | V (kPath) | <k, so kPath (s, t)=SP (s, t).
As figure 2 ?shown in 5, test four road networks related in the present embodiment, in CH pre-service and two stages of level label configurations, along with the quantity participating in the core calculated increases, time loss amount declines, pre-service improved efficiency.
The pretreatment time of above-mentioned four road networks is as shown in table 1.
Table 1
| Road network | Point | Limit | CH pretreatment time (s) | Level label configurations time (s) |
| NWs8324 | 8324 | 17792 | 78 | 13 |
| BAYs15164 | 15164 | 42464 | 317 | 76 |
| Floridas52781 | 52781 | 127274 | 785 | 153 |
| NWs111729 | 111729 | 234224 | 2967 | 1229 |
The CH pre-processed results of above-mentioned four road networks is as shown in table 2.
Table 2
| Road network | Iterations | Virtual shortcut quantity | Average layer secondary label size |
| NWs8324 | 26 | 12953 | 10.2 |
| BAYs15164 | 63 | 69131 | 125 |
| Floridas52781 | 121 | 148121 | 55.5 |
| NWs111729 | 486 | 167939 | 72 |
Because CH inquiry is carried out on the figure that with the addition of all virtual shortcuts, be the Disjktra search of a beta pruning, and search is only from the low node of level to the high node searching of level, greatly reduces the space searched for and take.This method can find figure mid point to the key point on shortest path by the Covering property of level label fast.No matter truly the length of shortest path, all can return the node of specified quantity, the importance that these nodes are determined according to arbitrary index sorts, to ensure that these nodes are nodes of most important front specified quantity on whole piece shortest path.Therefore, the present invention concisely can obtain outbound path summary efficiently according to user's request, and has certain expansion.
Claims (8)
1. the shortest path key point querying method based on Spark distributed system, it is characterized in that, by carrying out pre-service to all nodes in figure according to the node total order of crucial degree sequence composition on Spark platform, obtain the level label that each node is corresponding, realize the shortest path key point inquiry between arbitrfary point pair, and return the key point of shortest path being specified number;
Described level label is two tuple-sets, and in two tuple-sets of each node, two tuples are made up of destination node and the distance vector of this node to destination node.
2. the shortest path key point querying method based on Spark distributed system according to claim 1, it is characterized in that, described pre-service comprises: Distributed C H pre-service, structure and distributed storage level label.
3. the shortest path key point querying method based on Spark distributed system according to claim 2, it is characterized in that, described level label is according to the direction of distance vector, be further divided into forward direction label and reverse label, and forward direction label and reverse label meet the Covering property of level label respectively, namely right to any one point, the common factor of the level label of two nodes that this point is internal must comprise the node that on their shortest paths, crucial degree is the highest.
4. the shortest path key point querying method based on Spark distributed system according to claim 2, it is characterized in that, described Distributed C H pre-service refers to: the Distributed C H pre-service described node total order being carried out to multiple iteration on Spark platform, obtains the figure comprising all virtual shortcuts.
5. the shortest path key point querying method based on Spark distributed system according to claim 2 or 4, it is characterized in that, described Distributed C H pre-service specifically comprises the following steps:
Step 2.1) mode of being reduced by iteration sets up the residual graph obtaining independent sets and remove independent sets from figure;
Step 2.2) virtual shortcut is added to the node in residual graph.
6. the shortest path key point querying method based on Spark distributed system according to claim 5, it is characterized in that, described step 2.1, be specially: for each node v in figure G and its adjoint point Neighbor (v), if to any w ∈ Neighbor (v), there is r (v) >r (w), then node v is put into independent sets, and in current iteration, v is reduced.
7. the shortest path key point querying method based on Spark distributed system according to claim 5, it is characterized in that, described step 2.2, be specially: for any point v in the independent sets Ii in an iteration, u ∈ Nin (v) and w ∈ Nout (v), set a definite value λ, in figure G ' to the adjoint point of each some v in independent sets to <u, w> carry out λ ?jumping figure (λ ?hop) shortest path Witness search for, if the u obtained ?w bee-line d ' (u, w) limit (u is greater than, v) with (v, w) length sum edge (u, v)+edge (v, w), then need interpolation u ?virtual shortcut between w, the length of virtual shortcut is edge (u, v)+edge (v, w).
8. the shortest path key point querying method based on Spark distributed system according to claim 2, it is characterized in that, described structure and memory hierarchy label refer to: to each node contained in the figure of all virtual shortcuts, namely root node carries out Distributed C H sweep forward and reverse search respectively, in the forward direction label node searched being stored into respectively root node and reverse label, as the level label of root node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510478276.5A CN105005627A (en) | 2015-08-07 | 2015-08-07 | Shortest path key node query method based on Spark distributed system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510478276.5A CN105005627A (en) | 2015-08-07 | 2015-08-07 | Shortest path key node query method based on Spark distributed system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105005627A true CN105005627A (en) | 2015-10-28 |
Family
ID=54378303
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510478276.5A Pending CN105005627A (en) | 2015-08-07 | 2015-08-07 | Shortest path key node query method based on Spark distributed system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105005627A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105928535A (en) * | 2016-06-15 | 2016-09-07 | 苏州清研捷运信息科技有限公司 | Vehicle routing planning method capable of avoiding road restrictions |
| CN106372127A (en) * | 2016-08-24 | 2017-02-01 | 云南大学 | Spark-based diversity graph sorting method for large-scale graph data |
| CN111314138A (en) * | 2020-02-19 | 2020-06-19 | 腾讯科技（深圳）有限公司 | Detection method of directed network, computer readable storage medium and related equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6990484B1 (en) * | 2002-08-09 | 2006-01-24 | Ncr Corporation | Determining the satisfiability and transitive closure of conditions in a query |
| US7321891B1 (en) * | 2004-02-19 | 2008-01-22 | Ncr Corp. | Processing database queries |
| CN101393562A (en) * | 2008-09-26 | 2009-03-25 | 复旦大学 | Combined web service data processing method |
| CN101788999A (en) * | 2009-12-30 | 2010-07-28 | 安徽大学 | Binary chop tracking method of shortest paths in network map |
| US20140280360A1 (en) * | 2013-03-15 | 2014-09-18 | James Webber | Graph database devices and methods for partitioning graphs |
| CN104516350A (en) * | 2013-09-26 | 2015-04-15 | 沈阳工业大学 | Mobile robot path planning method in complex environment |
| CN104518965A (en) * | 2013-09-30 | 2015-04-15 | 华为技术有限公司 | Method and device for querying shortest paths |
-
2015
- 2015-08-07 CN CN201510478276.5A patent/CN105005627A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6990484B1 (en) * | 2002-08-09 | 2006-01-24 | Ncr Corporation | Determining the satisfiability and transitive closure of conditions in a query |
| US7321891B1 (en) * | 2004-02-19 | 2008-01-22 | Ncr Corp. | Processing database queries |
| CN101393562A (en) * | 2008-09-26 | 2009-03-25 | 复旦大学 | Combined web service data processing method |
| CN101788999A (en) * | 2009-12-30 | 2010-07-28 | 安徽大学 | Binary chop tracking method of shortest paths in network map |
| US20140280360A1 (en) * | 2013-03-15 | 2014-09-18 | James Webber | Graph database devices and methods for partitioning graphs |
| CN104516350A (en) * | 2013-09-26 | 2015-04-15 | 沈阳工业大学 | Mobile robot path planning method in complex environment |
| CN104518965A (en) * | 2013-09-30 | 2015-04-15 | 华为技术有限公司 | Method and device for querying shortest paths |
Non-Patent Citations (1)
| Title |
|---|
| 寒冬中的新蕊: "GraphX最短路径之实践与心得", 《HTTPS://WENKU.BAIDU.COM/VIEW/9667AA4EAAEA998FCD220E2D.HTML》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105928535A (en) * | 2016-06-15 | 2016-09-07 | 苏州清研捷运信息科技有限公司 | Vehicle routing planning method capable of avoiding road restrictions |
| CN105928535B (en) * | 2016-06-15 | 2018-08-31 | 苏州清研捷运信息科技有限公司 | A kind of vehicle path planning method of road limitation |
| CN106372127A (en) * | 2016-08-24 | 2017-02-01 | 云南大学 | Spark-based diversity graph sorting method for large-scale graph data |
| CN106372127B (en) * | 2016-08-24 | 2019-05-03 | 云南大学 | The diversity figure sort method of large-scale graph data based on Spark |
| CN111314138A (en) * | 2020-02-19 | 2020-06-19 | 腾讯科技（深圳）有限公司 | Detection method of directed network, computer readable storage medium and related equipment |
| CN111314138B (en) * | 2020-02-19 | 2021-08-31 | 腾讯科技（深圳）有限公司 | Detection method of directed network, computer readable storage medium and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106528773B (en) | Map computing system and method based on Spark platform supporting spatial data management | |
| Zhang et al. | All-nearest-neighbors queries in spatial databases | |
| CN104281652B (en) | Strong point data partition method one by one in metric space | |
| CN103198151B (en) | The search index system and method for regional urban public traffic vehicles operation information | |
| Navarro et al. | Space-efficient data-analysis queries on grids | |
| CN102609490B (en) | Column-storage-oriented B+ tree index method for DWMS (data warehouse management system) | |
| CN109492060A (en) | A kind of map tile storage method based on MBTiles | |
| CN104239377A (en) | Platform-crossing data retrieval method and device | |
| Ouyang et al. | Progressive top-k nearest neighbors search in large road networks | |
| CN101324896A (en) | Method for storing and searching vector data and management system thereof | |
| US10545960B1 (en) | System and method for set overlap searching of data lakes | |
| CN105357247A (en) | Multi-dimensional cloud resource interval finding method based on hierarchical cloud peer-to-peer network | |
| Li et al. | Fastest path query answering using time-dependent hop-labeling in road network | |
| CN109540165B (en) | Heuristic search expressway network constraint routing algorithm | |
| Yang et al. | Pase: Postgresql ultra-high-dimensional approximate nearest neighbor search extension | |
| Lee et al. | Efficient index-based approaches for skyline queries in location-based applications | |
| CN105005627A (en) | Shortest path key node query method based on Spark distributed system | |
| CN112035586A (en) | Spatial range query method based on extensible learning index | |
| Ali et al. | The maximum trajectory coverage query in spatial databases | |
| Zheng et al. | Repose: distributed top-k trajectory similarity search with local reference point tries | |
| Cho et al. | A basis of spatial big data analysis with map-matching system | |
| CN104462095A (en) | Extraction method and device of common pars of query statements | |
| CN101753324A (en) | Rapid rate configuration and searching method | |
| CN113449052A (en) | Method for establishing spatial index, method and device for querying spatial region | |
| CN105005628A (en) | Shortest path key node query method based on centralized platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151028 |
|
| RJ01 | Rejection of invention patent application after publication |