CN114896480A - Top-K space keyword query method based on road network index - Google Patents

Abstract

The invention discloses a Top-K space keyword query method based on road network index, which comprises the following steps: 1) defining the related concepts of the road network; 2) performing weighting calculation by using the shortest passing time and a TF-IDF model to obtain a spatial text score function capable of measuring the spatial proximity and text similarity of the object; 3) constructing a road network index TKG-tree by using side and vertex information contained in a road network, wherein the road network index comprises a transit time matrix recorded with spatial index information and an inverted document recorded with text index information; 4) taking the spatial text score as an index of Top-K result sorting, and finding a spatial keyword object for ranking Top-K by using the constructed road network index TKG-tree; according to the invention, the road network index is constructed in a way of dividing subgraphs, the problem of high index storage cost is solved, the spatial text score of the node is taken as an upper boundary pruning object, and the speed of searching the Top-K spatial keyword object can be improved.

Description

Top-K space keyword query method based on road network index

Technical Field

The invention relates to the technical field of road network index and Top-K query, in particular to a Top-K space keyword query method based on road network index.

Background

The Top-k space keyword query problem is a big research hotspot of a spatial database, and aims to query a plurality of objects which are close to a user position and meet the query preference of the user, and the query intention is expressed by keywords. The Top-k space keyword query problem not only considers the matching degree of the object and the keyword, but also considers the space distance between the user and the query object, and can be divided into two different distances, namely a Euclidean space and a road network space, according to the difference of the measurement modes of the space distance. The Euclidean space represents that the distance between two points is measured by the straight line distance, and the distance between two points in the road network space is measured by the shortest distance on the road network.

However, it is very difficult to support efficient Top-K queries over the internet. The main bottleneck is that the cost of calculating the point-to-point shortest path is particularly high, unlike the method of calculating the Euclidean distance between two points. In addition, even if the shortest road distance can be obtained quickly, it is time consuming to calculate the Top-K result if there is no efficient pruning algorithm and strategy. Although there is a lot of work to study the Top-K query processing problem on the road network. However, none of the existing methods support large-scale road networks. The main problem of these methods is that indexing usually results in too high storage cost or too high preprocessing time cost.

Disclosure of Invention

The invention aims to overcome the defects and shortcomings of the prior art, provides a Top-K space keyword query method based on a road network index, constructs the road network index in a subgraph dividing mode, relieves the problem of high index storage cost, divides the space text of nodes into upper bound pruned objects, and can improve the speed of searching the Top-K space keyword objects.

In order to achieve the purpose, the technical scheme provided by the invention is as follows: a Top-K space keyword query method based on a road network index comprises the following steps:

1) defining the related concept of road network, including space keyword object o, road network graph G and shortest transit time function tau between two vertexes ^* (t)；

2) Using the shortest transit time function tau obtained in step 1) ^* (t) measuring the spatial proximity of the spatial keyword object o to obtain a spatial score function, which is marked as f _s (o); measuring the text similarity of the spatial keyword object o by using a TF-IDF model to obtain a text score function, and marking as f _d (o); using textScore function f _d (o) and a spatial score function f _s (o) weighting to obtain a spatial text score function, which is recorded as

3) Constructing a road network index TKG-tree by using the edge and vertex information contained in the road network defined in the step 1); the TKG-tree is a balanced tree structure cut from a graph, and a root node of the tree structure corresponds to the whole road network graph G and has the following properties: each descendant node n of the root node _i All correspond to a sub-graph G _i The lowest level of nodes is called leaf nodes; the TKG-tree has two hyper-parameters f and mu which respectively represent the branch number of non-leaf nodes and the top point number upper limit of leaf nodes; each node n _i All contain a transit time matrix M _i And an inverted document D in which object keyword information o.key is recorded _i ；

4) Using the spatial text score function obtained in step 2)

From query point v as an indicator of Top-K result ranking _q Starting, finding the spatial key object o of the ranking Top-K by using the road network index TKG-tree constructed in the step 3).

Further, in step 1), the following concepts are defined:

a. in the edges of the road network, the length of the edge is expressed by the passing time, and the passing time of the edge changes along with the difference of time t, so the passing time function of the edge is recorded as omega (t); among the road network vertices, some vertices carry text information in the form of a keyword, and a vertex having spatial position information and text keyword information is referred to as a spatial keyword object o, and is expressed as:

o＝(v _o ,o.key)

in the formula, v _o Representing the vertex position of the object, and representing the key information of the object by o.key;

b. modeling a road network as an undirected graph, represented as:

in the formula, G is a road network graph and represents an undirected road network structure formed by crossing a plurality of edges; v. of ₁ Denotes the 1 st vertex, v _n Denotes the nth vertex, V ═ V ₁ ,v ₂ ,...v _n Represents the set of all the vertexes in the road network; e.g. of the type ₁ Denotes the 1 st edge, e _n Denotes the nth side, E ═ E ₁ ,e ₂ ,...e _n Representing a set of edges in a road network; and omega ₁ (t) represents the transit time function of the 1 st side, ω _n (t) represents a transit time function of the nth side, W ═ ω ₁ (t),ω ₂ (t),...ω _n (t) then represents the set of transit time functions associated with the corresponding edge;

c. in a road network, a path ρ from a starting point to an end point is represented by a series of adjacent connected vertex sequences<v _i ,...v _j >Wherein v is _i Denotes the ith vertex, v _j Represents the jth vertex, and P is used to represent the set of all paths from the starting point to the end point; noting the transit time function of the path ρ as τ (t), the shortest transit time function τ from the start point to the end point will be ^* (t) is defined as follows:

τ ^* (t)＝min{τ(t)|ρ∈P}

in the formula, τ (t) represents a transit time function of the route ρ from the time t; tau is ^* (t) represents the shortest transit time function of all routes from the start point to the end point, starting at time t.

Further, the step 2) comprises the following steps:

2.1) defining a query parameter query of a Top-K space keyword based on a road network space:

query＝<v _q ,q.key,t,k>

in the formula, v _q Representing query points, q.key representing a keyword phrase of the query, t representing the searching time, and k representing the number of returned spatial keyword objects o;

2.2) Using the shortest transit time function τ ^* (t) to measure the spatial key object o and the query point v _q Spatial proximity of (a), defining a spatial scoring function f _s (o) the following:

in the formula (I), the compound is shown in the specification,

representing the vertex v at which the spatial key object o is located _o To a query point v _q The shortest transit time in between;

2.3) using TF-IDF model to measure the text similarity between the spatial keyword object o and the keyword group q.key of the query, defining a text score function f _d (o) the following:

in the formula, key and q.key respectively represent keyword and keyword phrase of query, tf (key) represents the frequency of occurrence of the keyword key in the spatial keyword object o, idf (key) represents the frequency reciprocal of occurrence of the keyword key in all the spatial keyword objects o; the importance of a key increases in direct proportion to the number of times it appears in a spatial key object o, but at the same time decreases in inverse proportion to the frequency with which it appears in all spatial key objects o;

2.4) Using the text score function f _d (o) and a spatial score function f _s (o) obtaining a spatial text scoring function

The calculation formula is as follows:

in the formula, alphaWhich represents a weight factor, is given by the weight factor,

representing a spatial text scoring function, which is defined by a spatial scoring function f _s (o) and a text scoring function f _d And (o) performing weighted calculation.

Further, in step 3), the road network index TKG-tree is constructed by a graph cutting method, first taking the entire road network graph G as a root node, then cutting G into f subgraphs with the same size as the root node, and taking the subgraphs as child nodes of the root node, and then continuing to cut the subgraphs recursively until the number of vertices included in the last subgraph does not exceed μ; running Floyd algorithm to obtain node n according to topological relation of road network _i The transit time matrix M _i Matrix M _i Records the node n _i Corresponding subgraph G _i Shortest transit time function tau between vertices within ^* (t); for the node n _i Corresponding subgraph G _i Key information o.key of all spatial key objects o in the node n is obtained _i In the reverse arrangement document D _i 。

Further, in step 4), after inputting the query parameters, the spatial text score function is used

Searching as an index of Top-K result sorting; defining a maximum priority queue Q and a result set R, from a query point v _q Starting searching on the sub-graph, adding all spatial keyword objects o in the sub-graph into a priority queue Q and according to a spatial text score function

Sorting; using n _x Recording the highest level node currently visited on the TKG-tree, and marking the current search range as the subgraph G corresponding to the node _x At the same time, defining the spatial text score of the node

Pruning as upper bound:

in the formula (f) _s (node) representing a query point v _q To the current sub-graph G _x The shortest passing time of the network is the passing time matrix M of the road network index TKG-tree _i Calculating to obtain;

the inverted document D which represents the maximum text score that the spatial keyword object o can reach and uses the road network index TKG-tree _i Calculating to obtain; sequentially dequeuing the spatial keyword objects o in the queue Q, and screening out the upper bound of the spatial text score ratio

Adding a high spatial keyword object o into a result set R; if queue Q is empty and result set R is less than k, n will be _x Updating the node to the parent node of the original node, and expanding the current search range to n _x After the nodes are updated, the corresponding subgraphs are simultaneously updated

To n _x The space text after the node is updated scores, the operation is repeated until k results exist in the result set R, and the algorithm is automatically ended; maximum priority queue Q and upper bound

Ensure the slave query point v _q The result is globally optimal by the kth, so that the Top-K space keyword object o can be correctly found.

Compared with the prior art, the invention has the following advantages and beneficial effects:

1. the method considers the influence of time factors on the distance of the Top-K road network, and has more practical application value.

2. Compared with the Top-K query method of other road network spaces, the method has the advantages of shorter preprocessing time and lower storage cost.

3. The invention has millisecond-level query response time on data sets of different scales and has good universality and expansibility.

Drawings

FIG. 1 is a schematic logic flow diagram of the method of the present invention.

FIG. 2 is a diagram of a road network index TKG-tree used in the present invention.

Detailed Description

The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.

Referring to fig. 1, the embodiment provides a Top-K space keyword query method based on a road network index, which specifically includes the following steps:

1) the space keyword object, the road network graph and the shortest transit time are defined:

a. in the edges of the road network, the length of the edge is expressed by the passing time, and the passing time of the edge changes along with the difference of time t, so the passing time function of the edge is recorded as omega (t); among the road network vertices, some vertices carry text information in the form of a keyword, and a vertex having spatial position information and text keyword information is referred to as a spatial keyword object o, and is expressed as:

o＝(v _o ,o.key)

in the formula, v _o Representing the vertex position of the object, and representing the keyword information of the object by o.key;

b. modeling a road network as an undirected graph, represented as:

in the formula, G is a road network graph and represents an undirected road network structure formed by crossing a plurality of edges; v. of ₁ Denotes the 1 st vertex, v _n Denotes the nth vertex, V ═ V ₁ ,v ₂ ,...v _n Represents the set of all the vertexes in the road network; e.g. of the type ₁ Denotes item 1Edge, e _n Denotes the nth side, E ═ E ₁ ,e ₂ ,...e _n Representing a set of edges in a road network; and omega ₁ (t) represents the transit time function of the 1 st side, ω _n (t) represents a transit time function of the nth side, W ═ ω ₁ (t),ω ₂ (t),...ω _n (t) then represents the set of transit time functions associated with the corresponding edge;

c. in a road network, a path ρ from a starting point to an end point can be represented by a series of adjacent connected vertex sequences<v _i ,...v _j >Wherein v is _i Denotes the ith vertex, v _j Representing the jth vertex, and P can be used to represent the set of all paths from the start point to the end point. Noting the transit time function of the path ρ as τ (t), the shortest transit time function τ from the start point to the end point will be ^* (t) is defined as follows:

τ ^* (t)＝min{τ(t)|ρ∈P}

in the formula, τ (t) represents a transit time function of the route ρ from the time t; tau is ^* (t) represents the shortest transit time function of all routes from the start point to the end point, starting at time t.

2) Using the shortest transit time function tau obtained in step 1) ^* (t) measuring the spatial proximity of the spatial key object o; measuring the text similarity of the spatial keyword object o by using a TF-IDF model; the space text score function is obtained by weighting calculation by using a text score function and a space score function, and the method comprises the following steps:

2.1) defining a query parameter query of a Top-K space keyword based on a road network space:

query＝<v _q ,q.key,t,k>

in the formula, v _q Representing query points, q.key representing a keyword phrase of the query, t representing the searching time, and k representing the number of returned spatial keyword objects o;

2.2) Using the shortest transit time function τ ^* (t) to measure the spatial key object o and the query point v _q Spatial proximity of (a), defining a spatial scoring function f _s (o) the following:

in the formula (I), the compound is shown in the specification,

representing the vertex v at which the spatial key object o is located _o To a query point v _q The shortest transit time in between.

2.3) using TF-IDF model to measure the text similarity between the spatial keyword object o and the keyword group q.key of the query, defining a text score function f _d (o) the following:

in the formula, key and q.key respectively represent keyword and keyword phrase of query, tf (key) represents the frequency of occurrence of the keyword key in the spatial keyword object o, idf (key) represents the reciprocal of the frequency of occurrence of the keyword key in all the spatial keyword objects o; the importance of a key increases in direct proportion to the number of times it appears in a spatial key object o, but at the same time decreases in inverse proportion to the frequency with which it appears in all spatial key objects o.

2.4) Using the text score function f _d (o) and a spatial score function f _s (o) obtaining a spatial text score function

The calculation formula is as follows:

in the formula, α represents a weight factor,

representing a spatial text scoring function, which is defined by a spatial scoring function f _s (o) and text scoreFunction f _d And (o) obtaining a weighted calculation.

3) Constructing a road network index TKG-tree by using the side and vertex information contained in the road network defined in the step 1); the road network index TKG-tree is constructed by a graph cutting method, and as shown in FIG. 2, the number f of branches of a non-leaf node is set to be 2, and the maximum vertex number mu of a leaf node is set to be 4; firstly, the whole road network graph G is taken as a root node, then the G is cut into f sub-graphs with the same size, the sub-graphs are taken as child nodes of the root node, and then the sub-graphs are cut continuously and recursively until the number of vertexes contained in the last sub-graph does not exceed mu; running Floyd algorithm to obtain node n according to topological relation of road network _i The transit time matrix M _i Matrix M _i Records the node n _i Corresponding subgraph G _i Shortest transit time function tau between vertices within ^* (t); for the node n _i Corresponding subgraph G _i Key information o.key of all spatial key objects o in the node n is obtained _i In the reverse arrangement document D _i 。

4) Using the spatial text score function obtained in step 2)

From query point v as an indicator of Top-K result ranking _q Starting, finding a spatial key word object of the ranking Top-K by using the road network index TKG-tree constructed in the step 3); defining a maximum priority queue Q and a result set R, from a query point v _q Starting searching on the sub-graph, adding all spatial keyword objects o in the sub-graph into a priority queue Q and according to a spatial text score function

Sorting; using n _x Recording the highest level node currently visited on the TKG-tree, and marking the current search range as the subgraph G corresponding to the node _x At the same time, the spatial text score of the node is defined

Cut as upper boundBranching:

in the formula (f) _s (node) representing a query point v _q To the current sub-graph G _x The shortest passing time of the network is the passing time matrix M of the road network index TKG-tree _i Calculating to obtain;

the inverted document D which represents the maximum text score that the spatial keyword object o can reach and uses the road network index TKG-tree _i Calculating to obtain; sequentially dequeuing the spatial keyword objects o in the queue Q, and screening out the upper bound of the spatial text score ratio

Adding a high spatial keyword object o into a result set R; if queue Q is empty and result set R is less than k, n will be _x Updating the node to the parent node of the original node, and expanding the current search range to n _x After the nodes are updated, the corresponding subgraphs are simultaneously updated

To n _x And (4) scoring the updated space text of the node, and repeating the operation until the Top-K space keyword object o is correctly found.

The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (5)

1. The Top-K space keyword query method based on the road network index is characterized by comprising the following steps of:

1) defining the related concepts of road network, including spatial key word object o, road network graph G andshortest transit time function tau between two vertices ^* (t)；

2) Using the shortest transit time function tau obtained in step 1) ^* (t) measuring the spatial proximity of the spatial keyword object o to obtain a spatial score function, which is marked as f _s (o); measuring the text similarity of the spatial keyword object o by using a TF-IDF model to obtain a text score function, and marking as f _d (o); using a text scoring function f _d (o) and a spatial score function f _s (o) weighting to obtain a spatial text score function, which is recorded as

3) Constructing a road network index TKG-tree by using the edge and vertex information contained in the road network defined in the step 1); the TKG-tree is a balanced tree structure cut from a graph, and a root node of the tree structure corresponds to the whole road network graph G and has the following properties: each descendant node n of the root node _i All correspond to a sub-graph G _i The lowest level of nodes is called leaf nodes; the TKG-tree has two hyper-parameters f and mu which respectively represent the branch number of non-leaf nodes and the top point number upper limit of leaf nodes; each node n _i All contain a transit time matrix M _i And an inverted document D in which object keyword information o.key is recorded _i ；

4) Using the spatial text score function obtained in step 2)

From query point v as an indicator of Top-K result ranking _q Starting, finding the spatial key object o of the ranking Top-K by using the road network index TKG-tree constructed in the step 3).

2. The method for querying Top-K spatial keywords based on the road network index as claimed in claim 1, wherein in step 1), the following concepts are defined:

a. in the edges of the road network, the length of the edge is expressed by the passing time, and the passing time of the edge changes along with the difference of time t, so the passing time function of the edge is recorded as omega (t); among the road network vertices, some vertices carry text information in the form of a keyword, and a vertex having spatial position information and text keyword information is referred to as a spatial keyword object o, and is expressed as:

o＝(v _o ,o.key)

in the formula, v _o Representing the vertex position of the object, and representing the key information of the object by o.key;

b. modeling a road network as an undirected graph, represented as:

in the formula, G is a road network graph and represents an undirected road network structure formed by crossing a plurality of edges; v. of ₁ Denotes the 1 st vertex, v _n Denotes the nth vertex, V ═ V ₁ ,v ₂ ,...v _n Represents the set of all the vertexes in the road network; e.g. of the type ₁ Denotes the 1 st edge, e _n Denotes the nth side, E ═ E ₁ ,e ₂ ,...e _n Representing a set of edges in a road network; and omega ₁ (t) represents the transit time function of the 1 st side, ω _n (t) represents a transit time function of the nth side, W ═ ω ₁ (t),ω ₂ (t),...ω _n (t) then represents the set of transit time functions associated with the corresponding edge;

c. in a road network, a path ρ from a starting point to an end point is represented by a series of adjacent connected vertex sequences<v _i ,...v _j >Wherein v is _i Denotes the ith vertex, v _j Represents the jth vertex, and P is used to represent all path sets from the starting point to the end point; noting the transit time function of the path ρ as τ (t), the shortest transit time function τ from the start point to the end point will be ^* (t) is defined as follows:

τ ^* (t)＝min{τ(t)|ρ∈P}

wherein τ (t) represents the transit time function of the route ρ, starting at time t; tau is ^* (t) thenWhich represents the shortest transit time function of all the routes from the start point to the end point, starting at time t.

3. The method for querying Top-K spatial keywords based on the road network index according to claim 1, wherein the step 2) comprises the following steps:

2.1) defining a query parameter query of a Top-K space keyword based on a road network space:

query＝<v _q ,q.key,t,k＞

in the formula, v _q Representing query points, q.key representing a keyword phrase of the query, t representing the searching time, and k representing the number of returned spatial keyword objects o;

2.2) Using the shortest transit time function τ ^* (t) to measure the spatial key object o and the query point v _q Spatial proximity of (a), defining a spatial scoring function f _s (o) the following:

in the formula (I), the compound is shown in the specification,

representing the vertex v at which the spatial key object o is located _o To a query point v _q The shortest transit time in between;

2.3) using TF-IDF model to measure the text similarity between the spatial keyword object o and the keyword group q.key of the query, defining a text score function f _d (o) the following:

in the formula, key and q.key respectively represent keyword and keyword phrase of query, tf (key) represents the frequency of occurrence of the keyword key in the spatial keyword object o, idf (key) represents the reciprocal of the frequency of occurrence of the keyword key in all the spatial keyword objects o; the importance of a key increases in direct proportion to the number of times it appears in a spatial key object o, but at the same time decreases in inverse proportion to the frequency with which it appears in all spatial key objects o;

2.4) Using the text score function f _d (o) and a spatial score function f _s (o) obtaining a spatial text score function

The calculation formula is as follows:

in the formula, α represents a weight factor,

representing a spatial text scoring function, which is defined by a spatial scoring function f _s (o) and text score function f _d And (o) performing weighted calculation.

4. The method for searching for Top-K spatial keywords based on the road network index as claimed in claim 1, wherein: in step 3), the road network index TKG-tree is constructed by a graph cutting method, first taking the entire road network graph G as a root node, then cutting G into f subgraphs with the same size as the root node, and taking the subgraphs as child nodes of the root node, and then continuing to cut the subgraphs recursively until the number of vertices contained in the last subgraph does not exceed μ; running Floyd algorithm to obtain node n according to topological relation of road network _i The transit time matrix M _i Matrix M _i Records the node n _i Corresponding subgraph G _i Shortest transit time function tau between vertices within ^* (t); for the node n _i Corresponding subgraph G _i Key information o.key of all spatial key objects o in the node n is obtained _i Inverted document D _i 。

5. The method for searching for Top-K spatial keywords based on the road network index as claimed in claim 1, wherein: in step 4), after inputting query parameters, a spatial text score function is used

Searching as an index of Top-K result sorting; defining a maximum priority queue Q and a result set R, from a query point v _q Starting searching on the sub-graph, adding all spatial keyword objects o in the sub-graph into a priority queue Q and according to a spatial text score function

Sorting; using n _x Recording the highest level node currently visited on the TKG-tree, and marking the current search range as the subgraph G corresponding to the node _x At the same time, defining the spatial text score of the node

Pruning as upper bound:

in the formula (f) _s (node) representing a query point v _q To the current sub-graph G _x The shortest passing time of the network is the passing time matrix M of the road network index TKG-tree _i Calculating to obtain;

the inverted document D which represents the maximum text score that the spatial keyword object o can reach and uses the road network index TKG-tree _i Calculating to obtain; sequentially dequeuing the spatial keyword objects o in the queue Q, and screening out the upper bound of the spatial text score ratio

High spatial key object o, joinA result set R; if queue Q is empty and result set R is less than k, n will be _x Updating the node to the parent node of the original node, and expanding the current search range to n _x After the nodes are updated, the corresponding subgraphs are simultaneously updated

To n _x The space text after the node is updated scores, the operation is repeated until k results exist in the result set R, and the algorithm is automatically ended; maximum priority queue Q and upper bound

Ensure the slave query point v _q The result is globally optimal by the kth, so that the Top-K space keyword object o can be correctly found.

Images