CN108280181A - The immediate processing method of network data - Google Patents
The immediate processing method of network data Download PDFInfo
- Publication number
- CN108280181A CN108280181A CN201810064016.7A CN201810064016A CN108280181A CN 108280181 A CN108280181 A CN 108280181A CN 201810064016 A CN201810064016 A CN 201810064016A CN 108280181 A CN108280181 A CN 108280181A
- Authority
- CN
- China
- Prior art keywords
- user
- route
- navigation
- information
- route information
- 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
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
- G01C21/3484—Personalized, e.g. from learned user behaviour or user-defined profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Social Psychology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Automation & Control Theory (AREA)
- Probability & Statistics with Applications (AREA)
- Navigation (AREA)
Abstract
The present invention provides a kind of immediate processing method of network data, this method includes：The present invention proposes a kind of immediate processing method of network data, by dividing user preference, and according to user preference local similarity by it is global it is neighbouring be mutually fitted with the neighbouring recommendation results in part, to improve recommendation precision, improve the convenience that user uses navigation feature.User terminal can obtain recommended route information according to predefined recommendation rules, meet the different demands of user.
Description
Technical field
The present invention relates to big data, more particularly to a kind of immediate processing method of network data.
Background technology
With being gradually increased for living standard, the scope of activities of people is also gradually expanded, and is no longer only limitted to the daily of oneself
Living area.Traditional navigator route planning is based substantially on algorithm, although by calculating road or the online method for calculating road offline
Relative to single calculation road mode provide it is alternative calculate road mode, in no network or in the case of without offline navigation data, navigation
Online road mode of calculating can be switched to alternative offline calculation road mode by system automatically.The planning of urban road cannot be covered to road
The vehicle flowrate of road situation, different time points, if having the considerations of practical path, the complex situations such as section comfort level, pushed away sometimes
The route recommended may block up seriously or the used time is longer etc. so that existing navigation feature cannot be satisfied the demand of user, influence
The comfort and convenience of user.And in calculating road handoff procedure, alternative road mode of calculating can only recommend uniline to user,
With the further expansion of navigation system scale, number of users and geodata sharply increase, and lead to the pole of user's score data
Hold sparsity, and obtained by global similarity calculation it is neighbouring inaccurate, so as to cause route recommendation recommendation quality drastically
Decline.
Invention content
To solve the problems of above-mentioned prior art, the present invention proposes a kind of quick processing side of network data
Method, including：
In response to the inquiry instruction for the user that navigates, the route for including origin information and destination information is sent to Cloud Server
Inquiry request；
Cloud Server generates a plurality of route information, selectes optimal recommended route information and feeds back to navigation user.
Preferably, the selected optimal recommended route information feeds back to navigation user, further comprises：
Using the adopted times of route as recommendation rules；
The Cloud Server obtains a plurality of route information from origin information to destination information, and obtains every route letter
Breath history adopts number；
History is adopted into the most route information of number and feeds back to navigation user as recommended route information.
Preferably, the selected optimal recommended route information feeds back to navigation user, further comprises：
User personality is obtained from the historical action data of navigation user；
According to the user personality that navigation history acts, selectes optimal recommended route information and feed back to navigation user.
Preferably, the preference route probability function under multiple preferences is obtained according to the action of the navigation history of user, according to inclined
Good route probability function generates route recommendation list, and optimal recommended route information is selected from the list and feeds back to navigation use
Family.
The present invention compared with prior art, has the following advantages：
The present invention proposes a kind of immediate processing method of network data, by dividing user preference, and it is inclined according to user
Good local similarity by it is global it is neighbouring be mutually fitted with the neighbouring recommendation results in part, to improve recommendation precision, improve user and use
The convenience of navigation feature.User terminal can obtain recommended route information according to predefined recommendation rules, meet user not
Same demand.
Description of the drawings
Fig. 1 is the flow chart of the immediate processing method of network data according to the ... of the embodiment of the present invention.
Specific implementation mode
Retouching in detail to one or more embodiment of the invention is hereafter provided together with the attached drawing of the diagram principle of the invention
It states.The present invention is described in conjunction with such embodiment, but the present invention is not limited to any embodiments.The scope of the present invention is only by right
Claim limits, and the present invention covers many replacements, modification and equivalent.Illustrate in the following description many details with
Just it provides a thorough understanding of the present invention.These details are provided for exemplary purposes, and without in these details
Some or all details can also realize the present invention according to claims.
Fig. 1 is flow chart according to the method for the embodiment of the present invention.In response to the inquiry instruction for the user that navigates, to Cloud Server
Send the route inquiry request comprising origin information and destination information.Such as by taking automobile data recorder as an example, the automobile data recorder
In the driving process of automobile, network and Cloud Server real-time Communication for Power can be passed through.When navigation user response in effect on it
Inquiry instruction, and by the way that route inquiry request can be sent to the remote interface of Cloud Server transmission data.
Obtain the inquiry instruction for including origin information and destination information.In order to ensure drive safety, it is preferred to use language
Sound instruction is as the inquiry instruction, wherein including purposefully to believe in the natural language for the phonetic order that navigation user sends out
Breath.The a plurality of route information from origin information to destination information is generated, optimal push away is selected according still further to predefined recommendation rules
It recommends route information and feeds back to navigation user.
The predefined recommendation rules include the adopted times of route corresponding to the recommended route information.Cloud Server
The a plurality of route information from origin information to destination information is obtained, and obtains every route information history and adopts number, will be gone through
History adopts the most route information of number and feeds back to navigation user as recommended route information.
Wherein, to make recommendation results more meet the preference of user, it is preferable that the present invention is from navigation user's history action data
Middle excavation user personality, including：
All routes and preference are extracted from historical record, establishes the preference route probability function under Y preference, and Y is
Integer；
The route in historical record is divided into a plurality of Route Set according to the navigation action of user；For every Route Set, root
The user preference probability function of the Route Set is established according to the preference distribution of single user；
According to the weight of the user preference probability function of every Route Set to the user preference probability function of all Route Sets
Be weighted summation, obtain all users Y preference user preference probability function；
Route recommendation list is generated according to preference route probability function and user preference probability function；It is pushed away based on the route
It recommends list and carries out route recommendation.
Wherein, the dividing mode of navigation action can be according to recommendation and the difference of route guidance system application environment
It is selected.According to a preferred embodiment of the invention, the navigation action of user includes：Browsing route, compares road at analogue navigation
Line preserves route and executes navigation.Route in historical record is divided into a plurality of route by the present invention according to the navigation action of user
Collection, and the influence to recommendation results is acted by analysis and research user's different navigation to every Route Set.As long as navigation action
Type disclosure satisfy that analysis demand.
The quantity of Route Set can be identical as the type of navigation action, i.e. every Route Set and a kind of navigation action one are a pair of
It answers, the route in historical record is divided into browsing Route Set, analogue navigation Route Set, alternative line collection, preservation Route Set and holds
Row navigation routine collection.As long as the dividing mode of Route Set disclosure satisfy that analysis demand, the present invention is to the specific of navigation action
Dividing mode is not construed as limiting.
Influence for exam arrangement feature to recommendation results, according to a preferred embodiment of the invention, according to each preference
Under the preference route probability function established under preference of route distribution include：
For all routes in historical record, the route distribution under preference i is generatedEstablish the preference road under Y preference
Line probability function
It is Y × V matrixes in formula, i is integer, and 1≤i≤Y；V is the route quantity in historical record, and V is integer.
For every Route Set every is generated according to the frequency that every route of each user occurs in each preference
User preference probability function in Route Set.It enables browsing Route Set, analogue navigation Route Set, alternative line collection, preserve Route Set
It is respectively with navigation routine concentration user preference probability function is executed：θ1, θ2, θ3, θ4, θ5。
In order to which summation considers influence of each navigation action to recommendation results, according to the user preference probability of every Route Set
The weight of function is weighted summation to the user preference probability function of N Route Set, obtain all users Y preference use
Family preference probability function θ.I.e.：
θ=w1θ1+w2θ2+w3θ3+w4θ4+w5θ5
Wherein, w1,…,w5Respectively θ1,…,θ5Weight.Preferably, every Route Set is obtained using logistic regression
User preference probability function weight.
Route of user probability function is obtained based on preference route probability function and user preference probability function.Route of user is general
Each user is contained in rate function has every route the probability of preference, this probability value can be considered as meter of the user to route
Point, probability value is bigger, shows that user is higher to the preference of route.The route probability function RS of user is：
In formula, RS is M × V matrixes；θ is M × Y matrixes,For Y × V matrixes；Y is preference quantity, and M is number of users, and V is
Route quantity.
After obtaining the route probability function of user, according to user to the interested probability of every route, by all roads
Line arranges according to descending and generates route recommendation list, then carries out topN recommendations according to route recommendation list, i.e., from recommendation list
Middle selection scores highest N route recommendation to user.
In guidance path search process, the present invention is not complete connected graph to be established before search, but will connect
The process of establishing of logical figure incorporates in route searching.Only construction related may be connected to side with optimal path in search process, improve
The efficiency of global path planning.Specifically, connected graph is stored using adjacency list.Node to be extended is stored using EXPAND tables,
Search efficiency and store path are improved using SEARCH tables.It is as follows in the entry of EXPAND tables storage：
EXPAND (i)={ index, prev, status, gn, hn, visitedCount }
In formula：Index is number of the node in connected graph point set V, and prev indicates a upper node, and status is node shape
Whether state was traversed from a upper node to this node, 0 indicate be not traversed or 2 points between access be not connection
Side, 1 indicates that the access between 2 points is connection side.The value of gn is length of the upper node gn values plus access between 2 nodes, hn
This node is indicated to the estimated value of destination, visitedCount is the access times of node, for avoiding repeated accesses identical
Node.
Each nodal information following SEARCH (i)={ index, prev, gn } in SEARCH tables
In formula：Prev is call number of the upper node in SEARCH tables, and prev indicates that the node is starting point for 0, and gn is
Actual path length of the starting point to this node.
Including initialization and route searching two parts.In initialization, destination G is put into EXPAND tables, by starting point S
It is stored in SEARCH tables, it will be in deposit connected graph in side adjacent in congestion regions；During route searching, need to take out
The node of assessed value minimum is extended.Access is constructed according to the state of node, then completes the structure of connected graph on this basis
Make the search with path.It is as follows：
(1) EXPAND tables are initialized as EXPAND (1)={ G, S, 0, DSG, 0,0 } by step 1
In formula：DSG is Euclidean distances of the starting point S to destination G.By SEARCH tables be initialized as SEARCH (1)=S, 0,
0}
(2) step 2 takes out minimum node.The node for taking out evaluation function value minimum is extended.Evaluation function defines such as
Lower f (x)=gn+hn+visitedCount × MAXDISTANCE
In formula：MAXDISTANCE indicates the possible maximum distance from starting point to destination.Ensuring in this way will not repeated accesses
Identical node.Select in the node not being accessed the gn+hn values minimum, i.e. the shortest node of estimated path extends.
(3) step 3, decision node access times.Judge whether the access times of present node are equal to 1, if being equal to 1 table
Show that all feasible paths are traversed, search the path to arrive at the destination less than 1, algorithm terminates.
(4) step 4, according to nodal information accessed path.It is divided into 2 different branches according to the state of node to execute.Work as section
When dotted state is 0, step 5 is executed, it is no to then follow the steps 6.
(5) step 5 traverses a upper node to the path of present node.Structurally access of the node to present node.Such as
The fruit line be present in connected graph or with congestion regions Lothrus apterus, update node state be 1, if the access is not in connected graph
In, it is added in connected graph as connection side.Conflict if access has with congestion regions, congestion regions are divided into 2 by access
Vertex farthest apart from access in 2 parts is taken out in a part respectively.If this 2 vertex are not in SEARCH and EXPAND tables
In, it is added in EXPAND tables.
EXPAND (size+1)={ P, node.prev, 0, node.prev.gn+DPP, DPG, 0 }
In formula：P is the vertex for passing through congestion regions, and node indicates the node of current extensions, and what DPP was indicated is a upper node
To the Euclidean distance of vertex P, DPG is Euclidean distances of the vertex P to destination G.The access times of present node are finally added 1.
(6) step 6, the path of traversal present node to destination.Access of the construction present node to destination G.If
The line does not conflict in connected graph or with congestion regions, and expression has found path, outgoing route, and algorithm terminates.Such as
The fruit line has with congestion regions to conflict, which is added in SEARCH tables, while being deleted from EXPAND tables.With step 5
The method of middle addition congestion regions boundary point is the same, adds in the node to EXPAND tables Jing Guo congestion regions.
EXPAND (size+1)={ P, node, 0, node.gn+DCP, DPG, 0 }
In formula：DCP is Euclidean distance of the present node to congestion regions boundary point P.
To overcome the recommendation quality of navigation system that can not cope with the growth of mass data, using based on route cluster and part
The recommendation rules of preference fitting, include the following steps：
1) in alternative route set determining offline, criterion is maximized using similarity and determines K initial cluster center；
2) all routes are clustered according to similarity and maximum k-means clustering algorithms；
3) on the basis of cluster, local adjacent user is found；
4) by local adjacent user and local adjacent user using similarity and the ratio between be fitted as weights, according to the overall situation
Score of the adjacent user with local adjacent user to route, score of the estimation navigation user to route, completes to recommend.
Step 1) is specially：
1.1) all route i, the similarity sim (i, j) between j are calculated using cosine similarity, wherein i, j=1,
2,…,n,i≠j：
Wherein, Ui,jIndicate the Route Set that route i and route j scores jointly, UiIndicate that route i has the Route Set of score, Uj
Indicate that route j has the Route Set of score, ru,iIndicate score of the navigation user u to route i, ru,jIndicate navigation user u to route j
Score utilize pearson similarity calculations so that the similarity of K clustering and the value of J reach maximum after cluster.
Using the two lines of similarity minimum between all routes as the first two initial cluster center μ1And μ2, k=2；
1.2) the other route i to each cluster centre μ being selected for not being selected as initial cluster center are calculated1,
μ2,…,μkSimilarity sim (i, μ1),sim(i,μ2),…,sim(i,μk)；
1.3) route i to the cluster centre μ being selected is selected1,μ2,…μkSimilarity in similarity maximum turn to road
Similarity sim (i, μ)=max { sim (i, μs of the line i to cluster centre collection1),sim(i,μ2),…,sim(i,μk)}；
1.4) the route i* for choosing the similarity minimum value for arriving cluster centre collection is+1 cluster centre μ of kth increased newlyk+1,
Sim (i*, μ)=min sim (1, μ), sim (2, μ) ... sim (i, μ) ... sim (n, μ) }；
If 1.5) k+1 ＜ K, 2) assignment k=k+1 is gone to step, and otherwise maximizing criterion using similarity determines K
The process of initial cluster center terminates.
Step 2) is specially：
2.1) K initial cluster center μ is determinedc, the cluster centre μ of each cluster ccFor a route, c=1,2 ...,
K；
2.2) for remaining other routes i, i=1,2 ..., n, i ≠ uc, c=1,2 ..., K, then according to route i and step
It is rapid 2.2) described in cluster centre μcSimilarity sim (i, μc), by the maximum criterion of similarity, route i is distributed to it most
In class c* representated by similar cluster centre, sim (i, uc*)=max { sim (i, μ1),sim(i,μ2),…,sim(i,
μK)}；
2.3) calculate the similarity that each clusters andWherein, IcIndicate all routes in cluster c；
Calculate it is all K cluster similarity andJ values when different routes are as cluster centre in cluster are calculated, by J values
Maximum principle, chooses the maximum route of J values as new cluster centre, i.e., using the route as cluster centre when, J values maximum；Such as
Fruit J values become larger, then return to step 2.1), otherwise cluster terminates.
Step 3) is specially：
3.1) the global similarity between navigation user, navigation user u and phases of the user v in all routes of navigating are calculated
Like degree sim (u, v), formula is as follows：
Wherein, IuvIndicate the Route Set that navigation user u and navigation user v score jointly, IuIndicate that navigation user u has score
Route Set, IvIndicate that navigation user v has the Route Set of score, ru,iIndicate score of the navigation user u to route i, rv,iIt indicates
Score of the navigation user v to route i；
By the model split navigation user preference of cluster, the user that navigates in each cluster is calculated by cosine similarity
Between similarity simc(u, v), formula are as follows：
Wherein, c indicates the cluster where route i, Ic uvIndicate that navigation user u and navigation user v are counted jointly in clustering c
The Route Set divided, Ic uIndicate that navigation user u has the Route Set of score, I in clustering cc vIndicate that navigation user v has in clustering c
The Route Set of score, ru,iIndicate score of the navigation user u to route i, rv,iIndicate score of the navigation user v to route i；
3.2) selection is made with the maximum top n navigation user of target navigation user similarity on the cluster c where route i
For the local adjacent user of target navigation user.
In the step 4), the estimation score value formula based on global adjacent user is as follows：
Wherein, p 'u,iIndicate that navigation user u is to route i's when the estimation score value of the global adjacent user based on navigation user
Estimate score value,Indicate the mean scores of navigation user u, nbuIndicate that the neighbouring set of the overall situation of navigation user u, sim (u, v) indicate
The global similarity of navigation user u and the user v that navigates, eu,vIndicate the similarity related weighing of navigation user u and the user v that navigates,
rv,iIndicate practical score of the navigation user u to route i,Indicate the mean scores of navigation user v；Related weighing formula is as follows：
Wherein, QuvFor the common number scoring of navigation user u and the user v that navigates, T is that preset adjacent user counts jointly
Score threshold.To target navigation user u, top n sim (u, v) * e are selectedu,vMaximum navigation user is as adjacent user.
The estimation score value formula of local adjacent user based on navigation user is as follows：
Wherein, p "u,iThe user u that navigates for indicating that the score of local adjacent user obtains to the estimation score value of route i,Table
Show mean scores of the navigation user u on cluster c, indicates the neighbouring set in parts of the navigation user u on cluster c, simc(u,v)
Indicate navigation user u and local similarities of the user v on cluster c that navigate, rv,iIndicate practical meters of the navigation user v to route i
Point,Indicate mean scores of the navigation user v on cluster c；
Estimation score value formula after local adjacent user is combined with global adjacent user with corresponding weights is as follows：
pU, i=p 'U, i*e′+p″U, i*e″；
Local adjacent user and global adjacent user estimate the weights distribution of score value by the neighbouring similarity in part and with it is complete
The ratio setting e ' and e " of the neighbouring similarity sum of office, formula is as follows：
Wherein, "=1 e '+e.
After obtaining the relation data between navigation user and route, operating procedure is carried out for route data：
A. route data is expressed as navigation user-route score matrix；
B. the similarity between the user that navigates on all routes is calculated, navigation user overall situation adjacent user is found；
C. the similarity between route is calculated, maximizing criterion using similarity determines similarity and maximum k-means
K initial cluster center of clustering algorithm, and the clustering algorithm is clustered for route；
D. local similarity and the local adjacent user to be navigated based on route cluster calculation between user；
E. global adjacent user and local adjacent user are fitted with the distribution of corresponding weights, according to navigation user's
The score of global adjacent user and local adjacent user to the score estimation navigation user of route to route, recommends to generate.
According to further embodiment of this invention, the method for generating recommendation further comprises：
1. extracting characteristic sequence：High frequency destination is pre-processed, including participle, word frequency and duplicate removal, obtains z high frequency
The foundation characteristic keyword of keyword as a purpose.Further, as unit of the user of destination, to training data, test
Data are pre-processed, including the processing of participle, stop words.According still further in each user's history navigation routine of sequence of event, t
This procedural abstraction is discrete Markov Chain, the preference of user is described with transfer matrix by the sequence that a high-frequency key words occur
Feature, and user preference is estimated based on this.
In the present embodiment, the process of participle is：Using Chinese automatic word-cut, in conjunction with self-defined route of user library to purpose
Ground information is segmented；Stop words processing method be：It is tabled look-up using quick indexing and garbage is filtered, to reduce
The noise of destination information.
2. establishing Markov chain model
Markov chain model can be expressed as a four-tuple：<X,K,P(C),MC>.Wherein, X is that Discrete Stochastic becomes
Amount, codomain are { x1,x2,...,xn, each xiA corresponding keyword, referred to as model a state, { x1,x2,...,xn}
For user characteristics keyword sequence；K indicates user's classification quantity that model includes；C={ c1,c2,...,cKIndicate dividing for user
Class, distribution function P (C) indicate the probability function of different classifications user；MC={ mc1,mc2,...,mcKIt is class Markov
Chain set, each element mckBe interpretive classification be ckUser navigate preference profiles Markov Chain, calculate transfer matrix
AkIn each elementWith initial state probabilities λkIn each single item
Wherein,It indicates in user characteristics keyword sequence, state is to (xi,xj) occur number；For Bayes
Background knowledge in estimation：Assuming that in every a kind of user preference feature critical word sequence, institute is stateful to (xi,xj) appearance
Number is all identical, then：
Wherein, constant beta value is the size n in problem space domain.
The learning process of Markov chain model needs to complete two tasks：First, being clustered to user characteristics sequence；
Second is that generating class Markov Chain for each classification.First regard each user as an independent classification, generates class Markov
Chain.Then these Markov Chains are clustered and is merged, after cluster result reaches standard, by calculating cluster result institute
Determining network posterior probability, and determine that the maximum cluster result of posterior probability is optimal.
The bayes decision rule based on minimal error rate, judgement user's classification c are used firstk.Point of user is determined
Class ckAfterwards, so that it may with its class Markov Chain mckThe preference profiles of user are described, are estimated using following formula：
V (t)=wt-1H(t-1)×A1 k+wt-2H(t-2)×A2 k+…+wt-hH(t-h)×Ah k
Wherein, Ai kFor class Markov Chain mckThe i-th row of transfer matrix, wt-iIt is respective weights.H (t) indicates that t moment should
The characteristic key words of user, namely vector H (t) indicate the state in time point t, and even user is in state x at this timei, then
The i-th dimension of the vector is 1, other dimensions are all 0.Vector v (t) indicates that in moment t, each state occurs general in system
Rate.That maximum one-dimensional corresponding state of probability value in V (t), namely in the moment most probable state of user.
3. constructing more preference profiles vectors：
After Markov chain model is established, in judgement user's classifying step of estimation procedure, by P=P ((x1,x2,…,
xn) | C=ck)P(Ck) sort in descending order, the corresponding class Markov Chain of s P value estimates user before taking, and obtains user
The possible interested keyword sequence of subsequent time, keyword sort according to corresponding P values size order, form the more preferences of user
Feature vector, s are the positive integer more than 1.
4. route recommendation
According to preference profiles keyword and corresponding weights, matching route recommends user, namely is tied according to estimation
The possible interested keyword of fruit, i.e. user's subsequent time will according to the relevant routes in the keyword search navigational route database
It recommends user.
In conclusion the present invention proposes a kind of immediate processing method of network data, by dividing user preference, and root
Overall situation and partial situation is mutually fitted adjacent to recommendation results according to user preference local similarity, to improve recommendation precision, improves user
Use the convenience of navigation feature.User terminal can obtain recommended route information according to predefined recommendation rules, meet user
Different demands.
Obviously, it should be appreciated by those skilled in the art, each module of the above invention or each steps can be with general
Computing system realize that they can be concentrated in single computing system, or be distributed in multiple computing systems and formed
Network on, optionally, they can be realized with the program code that computing system can perform, it is thus possible to they are stored
It is executed within the storage system by computing system.In this way, the present invention is not limited to any specific hardware and softwares to combine.
It should be understood that the above-mentioned specific implementation mode of the present invention is used only for exemplary illustration or explains the present invention's
Principle, but not to limit the present invention.Therefore, that is done without departing from the spirit and scope of the present invention is any
Modification, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.In addition, appended claims purport of the present invention
Covering the whole variations fallen into attached claim scope and boundary or this range and the equivalent form on boundary and is repairing
Change example.
Claims (4)
1. a kind of immediate processing method of network data, which is characterized in that including：
In response to the inquiry instruction for the user that navigates, the route inquiry for including origin information and destination information is sent to Cloud Server
Request；
Cloud Server generates a plurality of route information, selectes optimal recommended route information and feeds back to navigation user.
2. according to the method described in claim 1, it is characterized in that, the selected optimal recommended route information feeds back to navigation
User further comprises：
Using the adopted times of route as recommendation rules；
The Cloud Server obtains a plurality of route information from origin information to destination information, and obtains every route information and go through
History adopts number；
History is adopted into the most route information of number and feeds back to navigation user as recommended route information.
3. according to the method described in claim 1, it is characterized in that, the selected optimal recommended route information feeds back to navigation
User further comprises：
User personality is obtained from the historical action data of navigation user；
According to the user personality that navigation history acts, selectes optimal recommended route information and feed back to navigation user.
4. according to the method described in claim 3, it is characterized in that, further comprising：
The preference route probability function under multiple preferences is obtained according to the action of the navigation history of user, according to preference route probability letter
Number generates route recommendation list, and optimal recommended route information is selected from the list and feeds back to navigation user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810064016.7A CN108280181A (en) | 2018-01-23 | 2018-01-23 | The immediate processing method of network data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810064016.7A CN108280181A (en) | 2018-01-23 | 2018-01-23 | The immediate processing method of network data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108280181A true CN108280181A (en) | 2018-07-13 |
Family
ID=62804549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810064016.7A Pending CN108280181A (en) | 2018-01-23 | 2018-01-23 | The immediate processing method of network data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108280181A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113758494A (en) * | 2021-08-31 | 2021-12-07 | 北京百度网讯科技有限公司 | Navigation path planning method, device, equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100183183A1 (en) * | 2007-07-20 | 2010-07-22 | Electronics And Telecommunications Research Institute | Headset for user specific audio service and method for user specific audio service using the same |
| CN102099805A (en) * | 2008-07-15 | 2011-06-15 | 皇家飞利浦电子股份有限公司 | Method and apparatus for selecting a multimedia item |
| CN104165634A (en) * | 2014-07-28 | 2014-11-26 | 广州视源电子科技股份有限公司 | Path planning method based on use habit of user |
| CN105373597A (en) * | 2015-10-27 | 2016-03-02 | 华侨大学 | Collaborative filtering recommendation method for user based on k-medoids project clustering and local interest fusion |
| CN105973255A (en) * | 2016-06-17 | 2016-09-28 | 北京奇虎科技有限公司 | Route recommendation method and terminal |
| CN107025311A (en) * | 2017-05-18 | 2017-08-08 | 北京大学 | A kind of Bayes's personalized recommendation method and device based on k nearest neighbor |
| CN107256241A (en) * | 2017-05-26 | 2017-10-17 | 北京工业大学 | The film recommendation method for improving multi-objective genetic algorithm is replaced based on grid and difference |
-
2018
- 2018-01-23 CN CN201810064016.7A patent/CN108280181A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100183183A1 (en) * | 2007-07-20 | 2010-07-22 | Electronics And Telecommunications Research Institute | Headset for user specific audio service and method for user specific audio service using the same |
| CN102099805A (en) * | 2008-07-15 | 2011-06-15 | 皇家飞利浦电子股份有限公司 | Method and apparatus for selecting a multimedia item |
| CN104165634A (en) * | 2014-07-28 | 2014-11-26 | 广州视源电子科技股份有限公司 | Path planning method based on use habit of user |
| CN105373597A (en) * | 2015-10-27 | 2016-03-02 | 华侨大学 | Collaborative filtering recommendation method for user based on k-medoids project clustering and local interest fusion |
| CN105973255A (en) * | 2016-06-17 | 2016-09-28 | 北京奇虎科技有限公司 | Route recommendation method and terminal |
| CN107025311A (en) * | 2017-05-18 | 2017-08-08 | 北京大学 | A kind of Bayes's personalized recommendation method and device based on k nearest neighbor |
| CN107256241A (en) * | 2017-05-26 | 2017-10-17 | 北京工业大学 | The film recommendation method for improving multi-objective genetic algorithm is replaced based on grid and difference |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113758494A (en) * | 2021-08-31 | 2021-12-07 | 北京百度网讯科技有限公司 | Navigation path planning method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110555112B (en) | Interest point recommendation method based on user positive and negative preference learning | |
| CN109241291B (en) | Knowledge graph optimal path query system and method based on deep reinforcement learning | |
| CN110110094B (en) | Cross-network character association method based on social network knowledge graph | |
| CN105740401B (en) | A kind of interested site recommended method and device based on individual behavior and group interest | |
| CN107247961B (en) | Track prediction method applying fuzzy track sequence | |
| US20110282798A1 (en) | Making Friend and Location Recommendations Based on Location Similarities | |
| CN108509461A (en) | A kind of sequence learning method and server based on intensified learning | |
| CN110147421B (en) | Target entity linking method, device, equipment and storage medium | |
| CN105808590A (en) | Search engine realization method as well as search method and apparatus | |
| CN111931505A (en) | Cross-language entity alignment method based on subgraph embedding | |
| CN108280548A (en) | Intelligent processing method based on network transmission | |
| CN104778284A (en) | Spatial image inquiring method and system | |
| CN110222838B (en) | Document sorting method and device, electronic equipment and storage medium | |
| CN112818262B (en) | Map POI searching method, system, equipment and medium based on user data | |
| CA3059929A1 (en) | Text searching method, apparatus, and non-transitory computer-readable storage medium | |
| Cai et al. | An improved random walk based clustering algorithm for community detection in complex networks | |
| CN115270007B (en) | POI recommendation method and system based on mixed graph neural network | |
| Thiengburanathum et al. | A decision tree based recommendation system for tourists | |
| Huang et al. | Ponas: Progressive one-shot neural architecture search for very efficient deployment | |
| CN108182273A (en) | Network data processing method based on cloud storage | |
| CN111538916A (en) | Interest point recommendation method based on neural network and geographic influence | |
| CN110580252A (en) | Space object indexing and query method under multi-objective optimization | |
| CN114254615A (en) | Volume assembling method and device, electronic equipment and storage medium | |
| CN107909498B (en) | Recommendation method based on area below maximized receiver operation characteristic curve | |
| CN105760965A (en) | Pre-estimated model parameter training method, service quality pre-estimation method and corresponding devices |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180713 |
|
| RJ01 | Rejection of invention patent application after publication |