CN107194011A - A kind of position prediction system and method based on social networks - Google Patents
A kind of position prediction system and method based on social networks Download PDFInfo
- Publication number
- CN107194011A CN107194011A CN201710488445.2A CN201710488445A CN107194011A CN 107194011 A CN107194011 A CN 107194011A CN 201710488445 A CN201710488445 A CN 201710488445A CN 107194011 A CN107194011 A CN 107194011A
- Authority
- CN
- China
- Prior art keywords
- social networks
- prediction
- mrow
- data
- msub
- 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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Abstract
The invention belongs to social network position electric powder prediction, a kind of position prediction system and method based on social networks is disclosed, including：Social networks is crawled to register data；Data of being registered to the social networks crawled are pre-processed, and wash invalid data, and the openness of data of registering is handled using core smooth interpolation technology；The output probability and the output probability of unconventional position prediction predicted with reference to rotine positioning, whether prediction the next position is rotine positioning：Top m list of locations is obtained by rotine positioning prediction；By the unstructured information gathered in extraction and analysis data, applied to top m list of locations, top k list of locations is obtained.By knowledge mapping and position commending system to unconventional position prediction.The present invention solves the problems, such as cold start-up in position prediction, unconventional position prediction problem；Improve precision of prediction.
Description
Technical field
The invention belongs to social network position electric powder prediction, more particularly to a kind of position prediction based on social networks
System and method.
Background technology
With internet fast development and can location equipment a large amount of popularizations, the network application based on geo-location service
Increasingly popularize, such as targeted ads (targeted advertisement), tracking movement of population, prevent a disease from spreading, network
Safety, performance optimization etc., address location is widely used as a kind of high information resources of quality.The online society of simultaneous
The development of network is handed over, location-based service and online social networks gradually tend to fusion, that is, generate LBSN.Location-based LBSN is
Position and social combination, it supports user to be recorded whenever and wherever possible in social platform and share the geography information of oneself, and it is
Using communication network as medium, using intelligent terminal as the Novel platform of main carriers.In LBSN, a large number of users is by registering to friend
Friendly sharing positional information or geographical labels.Position social networks allows location-based social as a kind of new social patterns, makes
Obtain social activity and social activity under line on line organically to be combined, greatly change the life style of people.Social networks is expedited the emergence of
Many location Based services, in order to provide more preferable service, the most possible next position of prediction user is extremely important
's.Such as by predicting the next position of user, businessman can be with significantly more efficient dispensing targeted ads.Existing Forecasting Methodology is based on
The position prediction of GPS track historical data, the position prediction for having data of being registered based on social networks.Social networks register data and
GPS track historical data has obvious difference.Social networks is registered, and historical data is sparse, and position prediction scope is larger.Phase
Than being registered data in social networks, 5-10 meters are spaced between continuous recording gps data.But gps data only include longitude,
Latitude and timestamp information, not including semantic information, it is impossible to carry out position prediction according to social networks.It is existing to be based on social activity
The position prediction of network mainly has the prediction of movement locus and the prediction in next place.The prediction of movement locus is relative complex, opens
Pin is larger, shows good to periodicity trajectory predictions, but poor to periodically unconspicuous trajectory predictions precision.It is existing to be based on
Social networks the next position prediction hypothesis the next position user was once accessed, and the next position is only selected from case history position
Select, easily cause " cold start-up ", cause rotine positioning prediction well, unconventional position prediction precision is relatively low.
In summary, the problem of prior art is present be：It is existing that semantic letter is not included based on the prediction of GPS historical data locations
Breath, it is impossible to carry out position prediction according to social networks.It is existing that motion is had based on the prediction of social networks track Affinity Location
Trajectory predictions are relative complex, and expense is larger, poor to periodically unconspicuous trajectory predictions precision, easily cause " cold start-up ".
The content of the invention
The problem of existing for prior art, the invention provides a kind of position prediction system based on social networks and side
Method.
The present invention is achieved in that a kind of position predicting method based on social networks, described based on social networks
Position predicting method comprises the following steps：
Step one, social networks is crawled to register data；
Step 2, data of being registered to the social networks crawled are pre-processed, filter out register number of times be less than averagely register
The data of number of times, wash invalid data, and the openness of data of registering is handled using core smooth interpolation technology；In f
(x) in, if entering row interpolation using the average of neighborhood sample, make f (x) unsmooth, so using a kernel function to estimate
Smoothly；Specifically used core weighted average, formula is：
Wherein K () uses gaussian kernel function, it is seen then that from x0Nearer influence power is bigger, right
The Quan Yue great that should be exported, meets the realistic simulation for data of registering；
Step 3, the output probability P predicted with reference to rotine positioningr(loc) and unconventional position prediction output probability Pu
(loc), whether prediction the next position is rotine positioning；
Step 4, by rotine positioning prediction module, obtains top-m list of locations；Pass through extraction and analysis data acquisition
The unstructured information gathered in module, applied to top-m list of locations, improves position prediction precision, obtains top-k position
List, k<=m.
Further, whether described prediction the next position is that rotine positioning formula is：
P (loc)=λ Pr (loc)+(1- λ) Pu (loc).
Wherein Pr (loc) is rotine positioning prediction probability, and Pu (loc) is unconventional position prediction probability, and λ joins for regulation
Number, λ ∈ { 0,1 }.
Further, the rotine positioning prediction uses MHMM algorithms, and HMM binding times feature and space characteristics enter to position
Row prediction.The influence in time and space is not considered, identical observation sequence is given, and HMM always obtains identical and predicted the outcome；Examine
Consider the behavior of registering of social user is influenceed by time and space, and the next position is predicted from mixed HMM algorithm.
Wherein Ct+1For the position classification at t+1 moment, StFor the observation sequence of t
Column-shaped state,For time and space vector.
Further, the unconventional position prediction, which is combined, builds knowledge mapping, excavates social networks, is closed using fusion is social
The Markov model binding site commending system of system is predicted to unconventional position.Using registering, data set builds knowledge graph
Spectrum, made inferences on knowledge mapping, excavate similar users, based on history register data and merge similar users train a horse
Er Kefu models are predicted to the next position.Finally Markov model and position commending system are combined together, improved
Position prediction precision.
Further, by the use of data set of registering as data source, social knowledge mapping is built, is pushed away on knowledge mapping
Reason.Inference method has three classes：Embedding-based technologies, Pathranking algorithms, and Probabilistic
Graphicalmodels probabilistic models.The reasoning of social networks uses Embedding-based technologies.Embedding-based
Technology is the method using implicit factor model as basic thought, and it is the method for expressing based on low-dimensional vector, by knowledge mapping
Entity and relation expressed in the vector space of low-dimensional, then make inferences.Entity and relation are entered into row vector first
Represent；Secondly, scoring functions are defined to weigh the possibility of relation establishment.Furthermore, parameter Estimation, according to scoring functions reasoning phase
Like user.Similar users reasoning integrates the Interest Similarity that scoring functions structured message and unstructured analysis module are extracted,
It is expressed as follows：
Sim (u, v)=α s (u, v)+(1- α) w (u, v)；
Wherein, α is regulation parameter, and value is [0,1], reaction structure information and unstructured information similarity institute accounting
Weight, s (u, v) representative structure information similarity, w (u, v) represents non-structural information similarity.Unstructured information similarity is
Non-structural information analysis module extracts interest keyword, then calculates the similarity between user according to cosine similarity.Calculate
It is as follows：
Wherein ui,ujRepresent that user i and user's j interest keyword vectors are represented.Cosine value span is [0,1], 0 table
Show entirely different, 1 represents identical.
On the basis of history registers data, the influence of similar users is merged, training Markov model is carried out to position
Prediction, i.e. Lm=maxP (Am| H, sim (u, v)), wherein AmRepresent Markov algorithm, H represents history and registered data, sim (u,
V) user's similarity is represented.
Finally, the Markov model and position commending system of comprehensive fusion similar users, improve position prediction precision.Its
Formula is as follows：
Ltop-n=β Lm+(1-β)Sr；
Wherein LmRepresent the Markov model of fusion similar users, SrPosition commending system is represented, β is regulation weight, is taken
0.6.The Markov model and position commending system of comprehensive fusion similar users, obtain top-n list of locations.
Another object of the present invention is to provide a kind of position predicting method based on social networks based on social activity
The position prediction system of network includes：
Data acquisition module, using crawler system, crawls social networks and registers data；
Data preprocessing module, data of being registered to the social networks crawled are pre-processed, and wash invalid data, profit
The openness of data of registering is handled with core smooth interpolation technology；
Judge module, the output probability Pr (loc) and the output probability of unconventional position prediction predicted with reference to rotine positioning
Pu (loc), whether prediction the next position is rotine positioning；
Rotine positioning prediction module, for classifying to predicted position, the classification of first predicted position, predicted position；
Unstructured data analysis module, by rotine positioning prediction module, obtains top-m list of locations；By carrying
The unstructured information gathered in analyze data acquisition module is taken, applied to top-m list of locations, position prediction precision is improved,
Obtain top-k list of locations；
Unconventional position prediction module, with reference to knowledge mapping is built, excavates similar users, using the horse of fusion similar users
Er Kefu models couplings position commending system is predicted to unconventional position.
Another object of the present invention is to provide a kind of social activity of the position predicting method based on social networks described in application
The network terminal.
Advantages of the present invention and good effect are：Registered data based on social networks, with reference to fuzzy clustering, knowledge mapping and
Position recommends to be predicted next place, is not only suitable for normal mode position prediction, is applied to unconventional mode position again pre-
Survey.The influence of binding time and positional factor of the present invention to position prediction, fully excavates the semanteme that social networks is registered in information
Information, solves rotine positioning forecasting problem, while " cold to open in position prediction by being solved to unconventional position prediction
It is dynamic " problem.The present invention uses knowledge mapping, excavates implicit similar users；Predict that the next position is routine with probability theory knowledge
Position or unconventional position.For unconventional position, binding site commending system improves unconventional position prediction precision.
Brief description of the drawings
Fig. 1 is the position predicting method flow chart provided in an embodiment of the present invention based on social networks.
Fig. 2 is the position predicting method implementation process figure provided in an embodiment of the present invention based on social networks.
Fig. 3 is the position prediction system structure diagram provided in an embodiment of the present invention based on social networks.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to
Limit the present invention.
The application principle of the present invention is explained in detail below in conjunction with the accompanying drawings.
As shown in figure 1, the position predicting method provided in an embodiment of the present invention based on social networks comprises the following steps：
S101：Social networks is crawled to register data；
S102；Data of being registered to the social networks crawled are pre-processed, and wash invalid data, are smoothly inserted using core
Value technology is handled the openness of data of registering；
S103：The output probability P predicted with reference to rotine positioningr(loc) and unconventional position prediction output probability Pu
(loc), whether prediction the next position is rotine positioning：
S104：By rotine positioning prediction module, top-m list of locations is obtained；Pass through extraction and analysis data acquisition module
The unstructured information gathered in block, applied to top-m list of locations, improves position prediction precision, obtains top-k location column
Table (k<=m).
Rotine positioning prediction uses MHMM algorithms, and HMM binding times feature and space characteristics are predicted to position.Do not examine
Worry time and the influence in space, give identical observation sequence, and HMM always obtains identical and predicted the outcome；In view of social activity use
The behavior of registering at family is influenceed by time and space, and the next position is predicted from mixed HMM algorithm.
Wherein Ct+1For the position classification at t+1 moment, StFor the observation sequence of t
Column-shaped state,For time and space vector.
The unconventional position prediction, which is combined, builds knowledge mapping, social networks is excavated, using the horse of fusion social networks
Er Kefu models couplings position commending system is predicted to unconventional position.Using registering, data set builds knowledge mapping,
Made inferences on knowledge mapping, excavate similar users, based on history register data and merge similar users train a Ma Erke
Husband's model is predicted to the next position.Finally Markov model and position commending system are combined together, position is improved
Precision of prediction.
By the use of data set of registering as data source, social knowledge mapping is built, is made inferences on knowledge mapping.Reasoning
Method has three classes：Embedding-based technologies, Path ranking algorithms, and Probabilistic
Graphical models probabilistic models.The reasoning of social networks uses Embedding-based technologies.Embedding-
Based technologies are the methods using implicit factor model as basic thought, and it is the method for expressing based on low-dimensional vector, by knowledge graph
Entity and relation in spectrum are expressed in the vector space of low-dimensional, are then made inferences.Entity and relation are carried out first
Vector representation；Secondly, scoring functions are defined to weigh the possibility of relation establishment.Furthermore, parameter Estimation is pushed away according to scoring functions
Manage similar users.It is similar with the interest that unstructured analysis module is extracted that similar users reasoning integrates scoring functions structured message
Degree, is expressed as follows：
Sim (u, v)=α s (u, v)+(1- α) w (u, v)；
Wherein, α is regulation parameter, and value is [0,1], reaction structure information and unstructured information similarity institute accounting
Weight, s (u, v) representative structure information similarity, w (u, v) represents non-structural information similarity.Unstructured information similarity is
Non-structural information analysis module extracts interest keyword, then calculates the similarity between user according to cosine similarity.Calculate
It is as follows：
Wherein ui,ujRepresent that user i and user's j interest keyword vectors are represented.Cosine value span is [0,1], 0 table
Show entirely different, 1 represents identical.
On the basis of history registers data, the influence of similar users is merged, training Markov model is carried out to position
Prediction, i.e. Lm=maxP (Am| H, sim (u, v)), wherein AmRepresent Markov algorithm, H represents history and registered data, sim (u,
V) user's similarity is represented.
Finally, the Markov model and position commending system of comprehensive fusion similar users, improve position prediction precision.Its
Formula is as follows：
Ltop-n=β Lm+(1-β)Sr；
Wherein LmRepresent the Markov model of fusion similar users, SrPosition commending system is represented, β is regulation weight, is taken
0.6.The Markov model and position commending system of comprehensive fusion similar users, obtain top-n list of locations.
Fig. 2 is the position predicting method implementation process figure provided in an embodiment of the present invention based on social networks.
As shown in figure 3, the position prediction system provided in an embodiment of the present invention based on social networks includes：
Data acquisition module, using crawler system, crawls social networks and registers data.
Data preprocessing module, data of being registered to the social networks crawled are pre-processed, and wash invalid data, so
The openness of data of registering is handled using core smooth interpolation technology afterwards.
Judge module, the output probability P predicted with reference to rotine positioningr(loc) and unconventional position prediction output probability Pu
(loc), whether prediction the next position is rotine positioning：
P (loc)=λ Pr(loc)+(1-λ)Pu(loc), λ ∈ { 0,1 }.
Rotine positioning prediction module, rotine positioning is frequent mode, periodicity pattern；Such as the working of 8 thirty, 12 noon exists
Lunch near company, 6 pm partly comes home from work, is in after going home and sees that TV stays in.Rotine positioning precision of prediction is by then
Between factor, geographic factor and historical data influence.Using GHMM algorithms, HMM binding times feature and space characteristics are to position
It is predicted.In order to overcome the difficulty that estimation range is big, predicted position is classified first, the classification of first predicted position, so
Laggard one-step prediction position.
Unstructured data analysis module, by rotine positioning prediction module, obtains top-m list of locations.Non-structural
Change prediction module by the unstructured information gathered in extraction and analysis data acquisition module, applied to top-m list of locations, carry
High position precision of prediction, obtains top-k list of locations (k<=m).
Unconventional position prediction module, position prediction is directed not only to rotine positioning, due to the new kink characteristics of people, with many moulds
Formula, can show the exploration to unconventional position on Move Mode.As Saturday goes to the cinema, Zhou Tian goes shopping.It is unconventional
Position prediction module combines unstructured analysis module, while building knowledge mapping, excavates similar users, is used using fusion is similar
The Markov model binding site commending system at family is predicted to unconventional position.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
Any modifications, equivalent substitutions and improvements made within refreshing and principle etc., should be included in the scope of the protection.
Claims (7)
1. a kind of position predicting method based on social networks, it is characterised in that the position prediction side based on social networks
Method comprises the following steps：
Step one, social networks is crawled to register data；
Step 2, data of being registered to the social networks crawled are pre-processed, and are filtered out number of times of registering and are less than number of times of averagely registering
Data, wash invalid data, the openness of data of registering handled using core smooth interpolation technology；In f (x)
In, if entering row interpolation using the average of neighborhood sample, make f (x) unsmooth, so flat to estimate using a kernel function
It is sliding；Specifically used core weighted average, formula is：
<mrow>
<mi>f</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mn>0</mn>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mn>0</mn>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
</mrow>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>K</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mn>0</mn>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>;</mo>
</mrow>
Wherein K () uses gaussian kernel function, it is seen then that the influence power nearer from x0 is bigger, the Quan Yue great of correspondence output, meets label
To the realistic simulation of data；
Step 3, the output probability P predicted with reference to rotine positioningr(loc) and unconventional position prediction output probability Pu(loc),
Predict whether the next position is rotine positioning；
Step 4, by rotine positioning prediction module, obtains top-m list of locations；Pass through extraction and analysis data acquisition module
The unstructured information of middle collection, applied to top-m list of locations, improves position prediction precision, obtains top-k location column
Table, k<=m.
2. the position predicting method as claimed in claim 1 based on social networks, it is characterised in that described prediction the next position
Whether it is that rotine positioning formula is：
P (loc)=λ Pr (loc)+(1- λ) Pu (loc)；
Wherein Pr (loc) is rotine positioning prediction probability, and Pu (loc) is unconventional position prediction probability, and λ is regulation parameter, λ ∈
{ 0,1 }.
3. the position predicting method as claimed in claim 1 based on social networks, it is characterised in that the rotine positioning prediction
Using MHMM algorithms, HMM binding times feature and space characteristics are predicted to position；From mixed HMM algorithm to next bit
Put and be predicted；
Wherein Ct+1For the position classification at t+1 moment, StFor the observation sequence shape of t
State,For time and space vector.
4. the position predicting method as claimed in claim 1 based on social networks, it is characterised in that the unconventional position is pre-
Survey combination and build knowledge mapping, excavate social networks, the Markov model binding site using fusion social networks recommends system
System is predicted to unconventional position；First with registering, data set builds knowledge mapping, is made inferences on knowledge mapping, its
It is secondary based on history register data and merge similar users train a Markov model the next position is predicted；Finally will
Markov model and position commending system are combined together, and improve position prediction precision.
5. the position predicting method as claimed in claim 4 based on social networks, it is characterised in that the unconventional position is pre-
Survey combination and build knowledge mapping, excavate social networks, the Markov model binding site using fusion social networks recommends system
System is predicted to unconventional position；Specifically include：
With data set of registering as data source, social knowledge mapping is built, is made inferences on knowledge mapping；Inference method bag
Include three classes：Embedding-based technologies, Pathranking algorithms and Probabilistic graphical
Models probabilistic models；The reasoning of social networks uses Embedding-based technologies；Embedding-based technologies are with hidden
Formula factor model be basic thought method, for based on low-dimensional vector method for expressing, by the entity and relation in knowledge mapping
Expressed, then made inferences in the vector space of low-dimensional；
Entity and relation are subjected to vector representation first；Secondly, scoring functions are defined to weigh the possibility of relation establishment；Again
Person, parameter Estimation, according to scoring functions reasoning similar users；Similar users reasoning integrates scoring functions structured message and non-knot
The Interest Similarity that structure analysis module is extracted, is expressed as follows：
Sim (u, v)=α s (u, v)+(1- α) w (u, v)；
Wherein, α is regulation parameter, and value is [0,1], reaction structure information and unstructured information similarity proportion, s
(u, v) representative structure information similarity, w (u, v) represents non-structural information similarity；Unstructured information similarity is non-knot
Structure information analysis module extracts interest keyword, then calculates the similarity between user according to cosine similarity；It is calculated as follows：
<mrow>
<mi>w</mi>
<mrow>
<mo>(</mo>
<mi>u</mi>
<mo>,</mo>
<mi>v</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>&CenterDot;</mo>
<msub>
<mi>u</mi>
<mi>j</mi>
</msub>
</mrow>
<mrow>
<mo>|</mo>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>u</mi>
<mi>j</mi>
</msub>
<mo>|</mo>
</mrow>
</mfrac>
<mo>;</mo>
</mrow>
Wherein ui,ujRepresent that user i and user's j interest keyword vectors are represented；Cosine value span is [0,1], and 0 has represented
Complete different, 1 represents identical；
On the basis of history registers data, the influence of similar users is merged, training Markov model is predicted to position,
That is Lm=maxP (Am| H, sim (u, v)), wherein AmRepresent Markov algorithm, H represents history and registered data, sim (u, v) generation
Table user's similarity；
Finally, the Markov model and position commending system of comprehensive fusion similar users, improve position prediction precision；Its formula
It is as follows：
Ltop-n=β Lm+(1-β)Sr；
Wherein LmRepresent the Markov model of fusion similar users, SrPosition commending system is represented, β is regulation weight, takes 0.6；
The Markov model and position commending system of comprehensive fusion similar users, obtain top-n list of locations.
6. a kind of position prediction system based on social networks of the position predicting method based on social networks as claimed in claim 1
System, it is characterised in that the position prediction system based on social networks includes：
Data acquisition module, using crawler system, crawls social networks and registers data；
Data preprocessing module, data of being registered to the social networks crawled are pre-processed, and are washed invalid data, are utilized core
Smooth interpolation technology is handled the openness of data of registering；
Judge module, the output probability Pr (loc) and the output probability Pu of unconventional position prediction predicted with reference to rotine positioning
(loc), whether prediction the next position is rotine positioning；
Rotine positioning prediction module, for classifying to predicted position, the classification of first predicted position, predicted position；
Unstructured data analysis module, by rotine positioning prediction module, obtains top-m list of locations；Divided by extracting
The unstructured information gathered in analysis data acquisition module, applied to top-m list of locations, improves position prediction precision, obtains
Top-k list of locations；
Unconventional position prediction module, with reference to knowledge mapping is built, excavates similar users, using the Ma Erke of fusion similar users
Husband's models coupling position commending system is predicted to unconventional position.
7. a kind of social networks of the position predicting method based on social networks described in application Claims 1 to 5 any one is whole
End.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710488445.2A CN107194011A (en) | 2017-06-23 | 2017-06-23 | A kind of position prediction system and method based on social networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710488445.2A CN107194011A (en) | 2017-06-23 | 2017-06-23 | A kind of position prediction system and method based on social networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107194011A true CN107194011A (en) | 2017-09-22 |
Family
ID=59879517
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710488445.2A Pending CN107194011A (en) | 2017-06-23 | 2017-06-23 | A kind of position prediction system and method based on social networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107194011A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108345662A (en) * | 2018-02-01 | 2018-07-31 | 福建师范大学 | A kind of microblog data weighted statistical method of registering considering user distribution area differentiation |
| CN109902883A (en) * | 2019-03-25 | 2019-06-18 | 重庆邮电大学 | A kind of position predicting method of registering based on personalized level Density Estimator |
| CN109978983A (en) * | 2017-12-27 | 2019-07-05 | 艾迪普（北京）文化科技股份有限公司 | A kind of three-dimensional material real-time manufacturing system based on big data |
| CN110020168A (en) * | 2017-12-27 | 2019-07-16 | 艾迪普（北京）文化科技股份有限公司 | A kind of three-dimensional material recommended method based on big data |
| CN110222273A (en) * | 2019-05-14 | 2019-09-10 | 上海交通大学 | Shopping center's promotion method and system in social networks based on geographic community |
| CN110334293A (en) * | 2019-07-12 | 2019-10-15 | 吉林大学 | A kind of facing position social networks has Time Perception position recommended method based on fuzzy clustering |
| CN111898572A (en) * | 2020-08-05 | 2020-11-06 | 杭州云栖智慧视通科技有限公司 | Case intelligent serial-parallel method based on shape recognition |
| CN112347222A (en) * | 2020-10-22 | 2021-02-09 | 中科曙光南京研究院有限公司 | Method and system for converting non-standard address into standard address based on knowledge base reasoning |
| CN112488384A (en) * | 2020-11-27 | 2021-03-12 | 香港理工大学深圳研究院 | Method, terminal and storage medium for predicting target area based on social media sign-in |
| CN112749209A (en) * | 2020-12-31 | 2021-05-04 | 南开大学 | Method for constructing movement behavior map facing to space-time data |
| CN113010803A (en) * | 2021-05-26 | 2021-06-22 | 南京航空航天大学 | Prediction method for user access position in geographic sensitive dynamic social environment |
| CN113111581A (en) * | 2021-04-09 | 2021-07-13 | 重庆邮电大学 | LSTM trajectory prediction method combining space-time factors and based on graph neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104462190A (en) * | 2014-10-24 | 2015-03-25 | 中国电子科技集团公司第二十八研究所 | On-line position prediction method based on mass of space trajectory excavation |
| CN104680250A (en) * | 2015-02-11 | 2015-06-03 | 北京邮电大学 | Position predicting system |
| CN104750829A (en) * | 2015-04-01 | 2015-07-01 | 华中科技大学 | User position classifying method and system based on signing in features |
| CN106202236A (en) * | 2016-06-28 | 2016-12-07 | 联想(北京)有限公司 | A kind of customer location Forecasting Methodology and device |
| CN106528614A (en) * | 2016-09-29 | 2017-03-22 | 南京邮电大学 | Method for predicting geographical location of user in mobile social network |
-
2017
- 2017-06-23 CN CN201710488445.2A patent/CN107194011A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104462190A (en) * | 2014-10-24 | 2015-03-25 | 中国电子科技集团公司第二十八研究所 | On-line position prediction method based on mass of space trajectory excavation |
| CN104680250A (en) * | 2015-02-11 | 2015-06-03 | 北京邮电大学 | Position predicting system |
| CN104750829A (en) * | 2015-04-01 | 2015-07-01 | 华中科技大学 | User position classifying method and system based on signing in features |
| CN106202236A (en) * | 2016-06-28 | 2016-12-07 | 联想(北京)有限公司 | A kind of customer location Forecasting Methodology and device |
| CN106528614A (en) * | 2016-09-29 | 2017-03-22 | 南京邮电大学 | Method for predicting geographical location of user in mobile social network |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978983A (en) * | 2017-12-27 | 2019-07-05 | 艾迪普（北京）文化科技股份有限公司 | A kind of three-dimensional material real-time manufacturing system based on big data |
| CN110020168A (en) * | 2017-12-27 | 2019-07-16 | 艾迪普（北京）文化科技股份有限公司 | A kind of three-dimensional material recommended method based on big data |
| CN108345662A (en) * | 2018-02-01 | 2018-07-31 | 福建师范大学 | A kind of microblog data weighted statistical method of registering considering user distribution area differentiation |
| CN109902883A (en) * | 2019-03-25 | 2019-06-18 | 重庆邮电大学 | A kind of position predicting method of registering based on personalized level Density Estimator |
| CN110222273B (en) * | 2019-05-14 | 2021-08-17 | 上海交通大学 | Business point promotion method and system in social network based on geographic community |
| CN110222273A (en) * | 2019-05-14 | 2019-09-10 | 上海交通大学 | Shopping center's promotion method and system in social networks based on geographic community |
| CN110334293A (en) * | 2019-07-12 | 2019-10-15 | 吉林大学 | A kind of facing position social networks has Time Perception position recommended method based on fuzzy clustering |
| CN110334293B (en) * | 2019-07-12 | 2021-08-27 | 吉林大学 | Position social network-oriented position recommendation method with time perception based on fuzzy clustering |
| CN111898572A (en) * | 2020-08-05 | 2020-11-06 | 杭州云栖智慧视通科技有限公司 | Case intelligent serial-parallel method based on shape recognition |
| CN112347222A (en) * | 2020-10-22 | 2021-02-09 | 中科曙光南京研究院有限公司 | Method and system for converting non-standard address into standard address based on knowledge base reasoning |
| CN112488384A (en) * | 2020-11-27 | 2021-03-12 | 香港理工大学深圳研究院 | Method, terminal and storage medium for predicting target area based on social media sign-in |
| CN112749209A (en) * | 2020-12-31 | 2021-05-04 | 南开大学 | Method for constructing movement behavior map facing to space-time data |
| CN112749209B (en) * | 2020-12-31 | 2023-08-29 | 南开大学 | Method for constructing mobile behavior patterns oriented to space-time data |
| CN113111581A (en) * | 2021-04-09 | 2021-07-13 | 重庆邮电大学 | LSTM trajectory prediction method combining space-time factors and based on graph neural network |
| CN113111581B (en) * | 2021-04-09 | 2022-03-11 | 重庆邮电大学 | LSTM trajectory prediction method combining space-time factors and based on graph neural network |
| CN113010803A (en) * | 2021-05-26 | 2021-06-22 | 南京航空航天大学 | Prediction method for user access position in geographic sensitive dynamic social environment |
| CN113010803B (en) * | 2021-05-26 | 2021-08-24 | 南京航空航天大学 | Prediction method for user access position in geographic sensitive dynamic social environment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107194011A (en) | A kind of position prediction system and method based on social networks | |
| Lau et al. | A survey of data fusion in smart city applications | |
| Dong et al. | Occupant behavior modeling methods for resilient building design, operation and policy at urban scale: A review | |
| Kim et al. | Graph convolutional network approach applied to predict hourly bike-sharing demands considering spatial, temporal, and global effects | |
| Solomon et al. | Predict demographic information using word2vec on spatial trajectories | |
| Chang et al. | Understanding user’s travel behavior and city region functions from station-free shared bike usage data | |
| CN105183870A (en) | Urban functional domain detection method and system by means of microblog position information | |
| Hao et al. | A trajectory-based recruitment strategy of social sensors for participatory sensing | |
| Xia et al. | UPTP vehicle trajectory prediction based on user preference under complexity environment | |
| Al-Molegi et al. | Move, attend and predict: An attention-based neural model for people’s movement prediction | |
| Fang et al. | Identifying activities and trips with GPS data | |
| Tian et al. | Twitter user location inference based on representation learning and label propagation | |
| Pang et al. | Efficient point-of-interest recommendation with hierarchical attention mechanism | |
| Qiao et al. | Improvement of thermal comfort for underground space: Data enhancement using variational autoencoder | |
| Fu et al. | A novel spatiotemporal behavior-enabled random walk strategy on online social platforms | |
| Liu et al. | Unified route representation learning for multi-modal transportation recommendation with spatiotemporal pre-training | |
| CN106339769A (en) | User travel forecasting method for mobile social network | |
| Chen et al. | Trip-chain-based travel-mode-shares-driven framework using cellular signaling data and web-based mapping service data | |
| Campos-Cordobés et al. | Big data in road transport and mobility research | |
| Basyoni et al. | Speed prediction from mobile sensors using cellular phone‐based traffic data | |
| Shan et al. | Follow a guide to solve urban problems: the creation and application of urban knowledge graph | |
| Zhan et al. | A semantic sequential correlation based lstm model for next poi recommendation | |
| Bhatnagar et al. | Role of machine learning in sustainable engineering: a review | |
| Liao et al. | Location prediction through activity purpose: integrating temporal and sequential models | |
| Song et al. | HERMAS: a human mobility embedding framework with large-scale cellular signaling data |
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: 20170922 |
|
| RJ01 | Rejection of invention patent application after publication |