CN108829744A - A kind of travel mode recommended method based on situation element and user preference - Google Patents
A kind of travel mode recommended method based on situation element and user preference Download PDFInfo
- Publication number
- CN108829744A CN108829744A CN201810509174.9A CN201810509174A CN108829744A CN 108829744 A CN108829744 A CN 108829744A CN 201810509174 A CN201810509174 A CN 201810509174A CN 108829744 A CN108829744 A CN 108829744A
- Authority
- CN
- China
- Prior art keywords
- user
- travel mode
- travel
- travelling
- probability
- 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.)
- Granted
Links
Classifications
-
- 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
- G06Q50/10—Services
- G06Q50/14—Travel agencies
Landscapes
- Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Navigation (AREA)
Abstract
The travel mode recommended method based on situation element and user preference that the invention discloses a kind of matches longitude and latitude and POI point first according to user's travel histories data, constructs user's travelling rule database based on situation element；Then the situation element in user's travelling is extracted, building travel mode recommends the relationship between user preference；Eventually form user's travel mode suggested design.The present invention can not only solve single user and be recommended based on the travel mode under certain situation, but also can solve the approximate multi-user's travel mode of travelling preference and recommend；When situation changes, only it need to can solve the problems, such as that the travel mode of different situations under more elements is recommended by the joint probability of change situation element and the single element of calculating；The two big key factors that the situation and its preference of user's trip are recommended as user's travel mode, are comprehensively considered, and reach preferable facilitation to the promotion for recommending accuracy rate.
Description
Technical field
The invention belongs to field of artificial intelligence, it is related to a kind of travel mode based on situation element and user preference and pushes away
Method is recommended, selection, period corresponding Path selection, the selection of travel mode and travel mode including the period of travelling change to four
Aspect.
Background technique
Recommended based on situation element and user preference for user's travel mode, the choosing period of time that needs to coordinate to travel,
Period corresponding Path selection, travel mode select and whether change to the relationship between four factors.It is registered data with user's history
The mechanics of user can be concluded for reference, and obtain the movement rule of user (including single user or user group),
So that the recommendation research of user's travel mode is possibly realized.Comprehensively considering safety, convenience, timeliness and warp
Under the premise of Ji property, so that user is obtained best travel solutions as this patent research key content and a kind of challenge.
In view of the above-mentioned problems, domestic and foreign scholars expand many relevant researchs.Jing He et al. proposes one kind and is used for
Two step methods (document 1) that next POI recommends.Steffen Rendle et al.[Propose a kind of general optimum standard
BPR-OPT is used for personalized ordering (document 2).Carl Yang proposes a kind of general and orderly property semi-supervised learning frame
Structure (document 3) alleviates Deta sparseness by filtering adjacent user and POI, and inclined by adjusting the user based on context figure
It is good to handle different contexts.Eliezer de Souza da Silva et al. proposes a kind of band content and social trust letter
The Poisson matrix factorisation method (document 4) of breath.Huayu Li et al. people, which proposes, a kind of can effectively learn fine granularity and can solve
User interest is released, and the unified approach (document 5) of adaptive modeling can be carried out to missing data.Hideaki Kim et al. structure
A kind of real-time POI recommender system (document 6) is built, which updates recommendation referring to the time cycle without access.Ren Xing
Happy et al. geography, text, society, classification and the popularity information using point of interest, and these information are subjected to fusion and are proposed
A kind of probability matrix decomposition point of interest proposed algorithm (document 7) of context-aware.At the same time, they also proposed a kind of
Generative probabilistic model analog subscriber is closed to register the decision process (document 8) of behavior.Gao Rong et al. is based on matrix decomposition and merges emerging
The factors such as comment, user social contact association and the geography information of interest point, propose a kind of new point of interest recommended models (document 9).
Liao state fine jade et al. is directed to the problems such as without display user preference, interest nonuniformity and Deta sparseness, proposes one kind and is directed to
Property dual fine granularity POI Generalization bounds (document 10), while they with Higher-order Singular value decomposition algorithm to user-theme-
Time, three rank tensors was decomposed, and is obtained user to the interest scores of theme, is preferably resolved data sparsity problem.Yu Yong
It is red et al. to propose a kind of Poisson matrix decomposition point of interest proposed algorithm (document 11) based on Ranking.
Defect existing for current techniques sums up following points：
It is 1, less that there are technologies that the situation element that user travels is extracted and concluded；
It is 2, less that there are technologies, and situation element is combined to mode of being travelled and (combined) with the travel mode of user preference
Recommendation；
3, the existing recommended method for user's travel mode is usually to use to the single dimension of user (for example, the time
Dimension or spatial position dimension etc.) recommended models are established, compared with the joint situation element under major general's various dimensions in view of travel mode
In recommended models.
Bibliography
[document 1] Jing He, Xin Li, Lejian Liao.Category-aware Next Point-of-
Interest Recommendation via Listwise Bayesian Personalized Ranking.IJCAI
2017:1837-1843.
[document 2] Steffen Rendle, Christoph Freudenthaler, Zeno Gantne et al.BPR:
Bayesian Personalized Ranking from Implicit Feedback.UAI 2009:452-461.
[document 3] Carl Yang, Lanxiao Bai, Chao Zhang et al.Bridging Collaborative
Filtering and Semi-Supervised Learning:A Neural Approach for POI
Recommendation.KDD2017:1245-1254.
[document 4] Eliezer de Souza da Silva, Helge Langseth, Heri
Ramampiaro.Content-Based Social Recommendation with Poisson Matrix
Factorization.ECML/PKDD(1)2017:530-546.
[document 5] Huayu Li, Yong Ge, Defu Lian et al.Learning User's Intrinsic and
Extrinsic Interests for Point-of-Interest Recommendation:A Unified
Approach.IJCAI 2017:2117-2123.
[document 6] Hideaki Kim, Tomoharu Iwata, Yasuhiro Fujiwara et al.Read the
Silence:Well-Timed Recommendation via Admixture Marked Point Processes.AAAI
2017:132-139.
[document 7] Ren Xingyi, Song Meina, Song Jun moral based on user register behavior point of interest recommend [J] computer
Report, 2017,40 (1):28-51.
The point of interest of context-aware of [document 8] Ren Xingyi, Song Meina, Song Jun moral based on position social networks is recommended
[J] Chinese journal of computers, 2017,40 (4):824-841.
A kind of fusion scene of [document 9] Gao Rong, the such as Li Jing, Du Bo and the position social networks point of interest of comment information push away
Recommend model [J] Journal of Computer Research and Development, 2016,53 (4):752-763.
[document 10] Liao Guoqiong, Jiang Shan, all identical of will are recommended based on the dual fine granularity point of interest of position community network
[J] Journal of Computer Research and Development, 2017,54 (11):2600-2610.
[document 11] Yu Yonghong, Gaoyang, Poisson matrix decomposition point of interest proposed algorithm of the such as Wanghao based on Ranking
[J] Journal of Computer Research and Development, 2016,53 (8):1651-1663.
Summary of the invention
In view of the drawbacks of the prior art, the invention proposes a kind of travel modes based on situation element and user preference to push away
Recommend method.
The technical scheme adopted by the invention is that：A kind of travel mode recommendation side based on situation element and user preference
Method, which is characterized in that include the following steps：
Step 1：According to user's travel histories data, longitude and latitude and POI point are matched, constructs the user trip based on situation element
Line discipline database；
Step 2：The situation element in user's travelling is extracted, building travel mode recommends the relationship between user preference；
Step 3：Form user's travel mode suggested design.
Beneficial effects of the present invention are：
1, the situation elements recognition in user's travelling and rule construct, and can not only solve single user based under certain situation
Travel mode is recommended, and can solve the approximate multi-user's travel mode of travelling preference and recommend；
2, it when situation changes, need to can only be solved by the joint probability of change situation element and the single element of calculating
The travel mode of different situations recommends problem under certainly more elements；
3, the two big key factors that the situation and its preference of user's trip are recommended as user's travel mode, are integrated and are examined
Consider, preferable facilitation is reached to the promotion for recommending accuracy rate.
Detailed description of the invention
Fig. 1 is transfer point, course of action and the travel mode triadic relation's schematic diagram of the embodiment of the present invention；
Fig. 2 is the rule database schematic diagram of building user's travelling of the embodiment of the present invention；
Fig. 3 is that the travel mode of the embodiment of the present invention recommends the relation schematic diagram between user preference.
Specific embodiment
Understand for the ease of those of ordinary skill in the art and implement the present invention, with reference to the accompanying drawings and embodiments to this hair
It is bright to be described in further detail, it should be understood that implementation example described herein is merely to illustrate and explain the present invention, not
For limiting the present invention.
It provides first defined below：
Transfer point (Transfer hubs)：The combination of transfer time and position between different travel modes are referred to as to change to
Point < ts,ls>.Wherein, tsIndicate transfer time, lsIndicate lowering position.
Course of action (Movement routes)：Movement track of the user from trip initial position to target position, by more
A transfer point composition.
Travel mode (Transportation Mode)：With set M={ M1,M2,...,MiIndicate, wherein i indicates trip
Capable mode type；It is transfer point, course of action and the triangular relationship of travel mode wherein such as Fig. 1.
Travelling expectation：User's expectation is related with its preference.Travelling expectation includes time, cost and convenience；Generally, when
Between, cost and convenience three be difficult to reach expectation maximization simultaneously.Although the selection private car trip for example, certain user travels,
But private car is sometimes difficult to look for parking stall, thus causes convenience not high.
A kind of travel mode recommended method based on situation element and user preference provided by the invention, including following step
Suddenly：
Step 1：According to user's travel histories data, longitude and latitude and POI point are matched, constructs the user trip based on situation element
Line discipline database；
See Fig. 2, user's travel histories data of the present embodiment include that register time, user of user registers position and user
Comment on three classes data；Longitude and latitude is registered the location information that position includes from user, and POI point is public transit facility point of interest
(such as public transport, subway and other modes of transport website, parking lot etc.)；User comment refers to user to the feedback of customization suggested design push
Evaluation excavates to obtain user rating by attitude to user comment and emotion word；User time, user position of registering of registering is logical
It crosses Frequent Pattern Mining and constitutes user's trip trajectory model；Trajectory model and user rating collectively form the regular number of user's travelling
According to library, which customizes for user's travel solutions and personalized recommendation.
Step 2：The situation element in user's travelling is extracted, building travel mode recommends the relationship between user preference；
Specific implementation includes the following steps：
Step 2.1：Situation element in user's travelling includes user U, time T, section L and travel mode W；Respectively to
The historical data of family travelling presses a variable and three constants according to tetra- kinds of user U, time T, section L and travel mode W elements
Combination obtains following four classes probabilistic relation：
①PΔU,T,L,W- same time, same to a road section, different user select the probability of same travel mode；
②PU,ΔT,L,W- different time, same to a road section, same user select the probability of same travel mode；
③PU,T,ΔL,W- same time, different sections of highway, same user select the probability of same travel mode；
④PU,T,L,ΔW- same time, same to a road section, same user select the probability of different travel modes；
Step 2.2：Establish 1. -4. in probability and corresponding grade Rate relationship：
When the value of probability is between 0≤P≤0.2, corresponding grade Rate value is 1；
When the value of probability is between 0.2 P≤0.4 <, corresponding grade Rate value is 2；
When the value of probability is between 0.4 P≤0.6 <, corresponding grade Rate value is 3；
When the value of probability is between 0.6 P≤0.8 <, corresponding grade Rate value is 4；
When the value of probability is between 0.8 P≤1 <, corresponding grade Rate value is 5；
Wherein, probability P represents PΔU,T,L,W、PU,ΔT,L,W、PU,T,ΔL,WOr PU,T,L,ΔW；
Step 2.3：Construct travel mode recommendation and the relationship between user preference；
Define comparison rule：Travel mode to be compared is divided into two class of i and j, and comparison is divided into+,-and？Three
Class；Travel mode of the user u within t period and itinerary l is selected as two class of i and j, and the comparison of i and j is+, show
I is compared with two kinds of travel modes of j, and user u is more likely to i, is denoted as i > j；The comparison of i and j be-, show that i is compared with j,
User u is more likely to j, is denoted as i < j；The comparison of i and j is？, show that two kinds of travel modes of i and j can not compare, be denoted as i
< > j；
See Fig. 3, by taking the first row of the left side as an example, i1And j1Comparison be+, show i1And j1Two kinds of travel mode phases
Than user U is more likely to i1, it is denoted as i1> j1, similarly i1> j5。i1And j2Comparison be-, show i1And j2It compares, user
U is more likely to i2, it is denoted as i1< j2, similarly i1< j4。i1And j1Comparison be？, show i1And j3Two kinds of travel modes can not
Compare, is denoted as i1< > j3。
If the complete or collected works of travel mode are I, then comparison rule meets following property：
1. closed：
2. skew-symmetry：
3. transitivity：
Wherein, property 1. -3. in s, s' indicate user rating, value range be 1~5；
Step 3：Form user's travel mode suggested design；
User's travel histories data are trained using comparison rule, it is corresponding with grading of the multi-user to each travel mode
Travel mode under different road conditions selects ranking, then by the continuous adjustment transformation and optimization of ranking, ultimately forms user trip
Line mode suggested design.
Wherein, user's travel histories data are trained using comparison rule；Firstly, to the historical data of user's travelling
It is divided into five classes according to situation element (user, time, position, event, event inducement), to embody the situation of user's travelling.Wherein,
Event refers to starting point, terminating point and the used travel mode of user's travelling, and event inducement refers to user using certain
The reason of kind travel mode, inducement can be most users and select the probability of certain combination travel mode (for example, 80% use
Family is in the itinerary that certain period is constituted from certain starting point to certain terminating point using " subway-shares bicycle-walking "
Combination), then the inducement that said combination travel mode is used on the period corresponding at this time and the route is exactly 80%, it is right
The user rating answered is 5.And same period and itinerary, it is right when the probability of corresponding " subway " travel mode is 60%
The user rating answered is 4, and both front and back travel mode is compared rear (5>4), then being based on situation element, by comparing rule
After being trained, it can show that the travel mode for recommending user is the combination of " subway-shares bicycle-walking ".
Wherein, ranking is selected with the travel mode under the corresponding different road conditions of grading of the multi-user to each travel mode, is root
According to user's travelling demand, determines the situation element of this travelling demand of user, the joint probability of each situation element is calculated, to user
The travel mode for carrying out different situations under more elements is recommended；
Wherein the joint probability of each situation element is：
(1) in, certain user uiBelong to user set U, for describing the content in 2.1 sections 1., P (Δ U, j) shows more greatly
User is more in same time, the number for preferring to a road section travel mode j.Similarly, (2)-(4) can be respectively obtained.
(2) in, tiBelong to time set T, for describing the content in 2.1 sections 2., P (Δ T, j) shows that more greatly user exists
The time accounting that same a road section of different time prefers to travel mode j is bigger, i.e., the major travel of user in the section one day
Mode is j.
(3) in, liBelong to section set L, for describing the content in 2.1 sections 3., P (Δ L, j) shows that more greatly user exists
The same time, the accounting that different sections of highway prefers to travel mode j is bigger, can be used for side and reflects the city in the master of fixed period
Wanting travel mode is j.
(4) in, wjBelong to travel mode set W, respectively indicates content 4. in 2.1 sections.P (Δ W, j) shows to use more greatly
In the same time, the accounting for preferring to travel mode j with a road section is bigger at family, can be used for side reflection set time and section
The pouplarity and convenience of travel mode j.
It, can be by these four types of factors since user has demand to the time T of travelling, route Pa, cost Mo and convenience C
Tensor resolution is done, such as (5)
χ≈[T,P,M,C]≈T×P×M×C (5)
Wherein, T, P, M and C respectively indicate time, route, cost and convenience four indices, and χ indicates the meter of four factors
Calculate result.Therefore, next only the result need to repeatedly be compared, keeps its difference minimum.Thus obtain χ and T, Pa,
The calculated relationship of tetra- factors of Mo, C, such as (6)-(9).
(6) in, the calculated result of four factors in χ representation formula (5),It indicates any two in many experiments result
The difference of item,Indicate the difference at any two moment in many experiments result, P (Δ T, j) indicates user in different time
Same a road section prefer to the time accounting of travel mode j；Calculated result shows that the minimum value of the difference existsWhen obtain.Due toWithIt is directly proportional, thereforeIt is minimized, and if only if t=tn-1,
When, the time is most short.
(7) in,Indicate the difference of any two location point longitude and latitude in many experiments result, P (Δ L, j) table
Show user in the same time, different sections of highway prefers to the accounting of travel mode；Due toWithIt is directly proportional, therefore with
(6) similarly,It is minimized, and if only if pa=pan-1,When, path is most short.
(8) in,The difference of any travel cost twice in expression many experiments result, P (Δ W | Δ T, Δ L)
For joint probability,
Wherein：P (Δ W, Δ T, Δ L)=P (Δ W, j) P (Δ T, j) P (Δ L, j), P (Δ T, Δ L)=P (Δ T,
j)·P(ΔL,j)；Due toWithIt is directly proportional, therefore similarly with (6),It is minimized, and if only if
Mo=mon-1,When, cost is minimum.
(in above-mentioned formula, parameters are required to remark additionally one by one the meaning of its representative！)
(9) in,The difference for indicating any convenience of travelling twice in many experiments result, due toWith
It is directly proportional, therefore similarly with (6),It is minimized, and if only if c=cn-1,Convenience is best.Wherein, c=cv×
cp,cvIndicate the convenience of travel mode selection, cpIndicate the convenience of user's travelling.By taking private car as an example, private car travelling with
(can be traditionally arranged to be 100m) near destination to find legal parking space or the convenience in parking lot is maximum valueWith
Parking position is far from destination travel distance dtIncrease, convenience is gradually reduced [].User's travelling convenience cpMeasurement and user
The travel distance of travelling starting point and destinationRoad conditions jam situation Ltc, available travel mode type KwotEtc. because
It is plain related.For example, working asSmaller (generally 1km) and when road conditions are compared with congestion, user selects the likelihood ratio automobile of shared bicycle more
It is high.By analyzing above, it is known thatTherefore
dtIt is parking position far from destination travel distance,It is the travel distance of user's travelling starting point and destination；
By the way that user's travel mode can be obtained in the calculating process of convenience in conjunction with (9) and (10).In order to user's travel mode
Ranking optimizes adjustment, needs to T, Pa, Mo, tetra- factors of C assign weight (generally 0~1 value).By four because
Result after plain weighted sum carries out ranking, and the recommendation of user's travel mode is used for this.As shown in (11),
Wherein, S indicates to calculate score, and score value height determines the recommended ranking of travel mode.
χ=[T, Pa, Mo, C] is salt matrices,For adjustable weight matrix, the value range of each weight
For 0-1.
The present invention can effectively solve the problems, such as：
1, congestion period and its corresponding congested link can be obtained by the statistics to urban transportation historical data and non-gathered around
The stifled period selects the non-congested link of congestion period or non-congestion period to suggest for best trip to suggest that user goes on a journey；
2, whether the congestion period is also the non-congestion period, all corresponding to have non-congested link.Therefore, by that can be user
Recommend the corresponding relatively non-congested link of any time period for going on a journey；
3, recommend optimal travel mode in any time period for user, be considered in conjunction with user preference, the time, position,
When, where, because of what etc. whether five situation elements such as event and inducement change to for user, provide complementary decision.
It should be understood that the part that this specification does not elaborate belongs to the prior art.
It should be understood that the above-mentioned description for preferred embodiment is more detailed, can not therefore be considered to this
The limitation of invention patent protection range, those skilled in the art under the inspiration of the present invention, are not departing from power of the present invention
Benefit requires to make replacement or deformation under protected ambit, fall within the scope of protection of the present invention, this hair
It is bright range is claimed to be determined by the appended claims.
Claims (5)
1. a kind of travel mode recommended method based on situation element and user preference, which is characterized in that include the following steps：
Step 1：According to user's travel histories data, longitude and latitude and POI point are matched, constructs user's travelling rule based on situation element
Then database；
Step 2：The situation element in user's travelling is extracted, building travel mode recommends the relationship between user preference；
Step 3：Form user's travel mode suggested design.
2. the travel mode recommended method according to claim 1 based on situation element and user preference, which is characterized in that
The specific implementation process of step 1 is：
User's travel histories data include that register time, user of user registers position and user comment three classes data；The longitude and latitude
Degree is registered the location information that position includes from user, and the POI point is public transit facility point of interest；User comment refers to
User excavates to obtain user and comment to the Feedback Evaluation of customization suggested design push by attitude to user comment and emotion word
Grade；User time, user position of registering of registering passes through Frequent Pattern Mining and constitutes user and goes on a journey trajectory model；Trajectory model
The rule database of user's travelling is collectively formed with user rating, which customizes for user's travel solutions and individual character
Change and recommends.
3. the travel mode recommended method according to claim 1 based on situation element and user preference, which is characterized in that
The specific implementation of step 2 includes the following steps：
Step 2.1：Situation element in user's travelling includes user U, time T, section L and travel mode W；Respectively to
The historical data of family travelling presses a variable and three constants according to tetra- kinds of user U, time T, section L and travel mode W elements
Combination obtains following four classes probabilistic relation：
①PΔU,T,L,W- same time, same to a road section, different user select the probability of same travel mode；
②PU,ΔT,L,W- different time, same to a road section, same user select the probability of same travel mode；
③PU,T,ΔL,W- same time, different sections of highway, same user select the probability of same travel mode；
④PU,T,L,ΔW- same time, same to a road section, same user select the probability of different travel modes；
Step 2.2：Establish 1. -4. in probability and corresponding grade Rate relationship：
When the value of probability is between 0≤P≤0.2, corresponding grade Rate value is 1；
When the value of probability is between 0.2 P≤0.4 <, corresponding grade Rate value is 2；
When the value of probability is between 0.4 P≤0.6 <, corresponding grade Rate value is 3；
When the value of probability is between 0.6 P≤0.8 <, corresponding grade Rate value is 4；
When the value of probability is between 0.8 P≤1 <, corresponding grade Rate value is 5；
Wherein, probability P represents PΔU,T,L,W、PU,ΔT,L,W、PU,T,ΔL,WOr PU,T,L,ΔW；
Step 2.3：Construct travel mode recommendation and the relationship between user preference；
Define comparison rule：Travel mode to be compared is divided into two class of i and j, and comparison is divided into+,-and？Three classes；With
Travel mode of the family u within t period and itinerary l is selected as two class of i and j, and the comparison of i and j is+, show i and j
Two kinds of travel modes are compared, and user u is more likely to i, are denoted as i > j；The comparison of i and j be-, show that i is compared with j, user u
It is more likely to j, is denoted as i < j；The comparison of i and j is？, show that two kinds of travel modes of i and j can not compare, be denoted as i < > j；
If the complete or collected works of travel mode are I, comparison rule meets following property：
1. closed：
2. skew-symmetry：
3. transitivity：
Wherein, property 1. -3. in s, s' indicate user rating, value range be 1~5.
4. the travel mode recommended method according to claim 3 based on situation element and user preference, which is characterized in that
The specific implementation process of step 3 is：
User's travel histories data are trained using comparison rule, it is corresponding different with grading of the multi-user to each travel mode
Travel mode under road conditions selects ranking, then by the continuous adjustment transformation and optimization of ranking, ultimately forms user travelling side
Formula suggested design.
5. the travel mode recommended method according to claim 4 based on situation element and user preference, it is characterised in that：
Travel mode under the corresponding different road conditions with grading of the multi-user to each travel mode selects ranking, is travelled according to user
Demand, determines the situation element of this travelling demand of user, calculates the joint probability of each situation element, carries out more elements to user
The travel mode of lower difference situation is recommended；
Wherein the joint probability of each situation element is：
Wherein, j=1,2 ..., m, certain user uiBelonging to user's set U, P (Δ U, j) shows more greatly user in same time, same
The number that a road section prefers to travel mode j is more；tiBelong to time set T, P (Δ T, j) shows more greatly user when different
Between same a road section prefer to travel mode j time accounting it is bigger, i.e., the major travel mode of user is in the section one day
j；liBelong to section set L, P (Δ L, j) shows more greatly user in the same time, and different sections of highway prefers to accounting for for travel mode j
Than bigger, reflect that major travel mode of the city in the fixed period is j for side；wjBelong to travel mode set W, P (Δ
W, j) show user more greatly in the same time, the accounting for preferring to travel mode j with a road section is bigger, fixes for side reflection
The pouplarity and convenience of the travel mode j in time and section；
Since user has demand to the time T of travelling, route Pa, cost Mo and convenience C, therefore these four types of factors are done into tensor point
Solution：
χ≈[T,Pa,Mo,C]≈T×Pa×Mo×C (5)
Wherein, T, Pa, Mo and C respectively indicate time, route, cost and convenience four indices, and χ indicates the calculating of four factors
As a result；
Calculated result is repeatedly compared, keeps its difference minimum；Thus the calculating for obtaining χ and tetra- factors of T, Pa, Mo, C is closed
System：
(6) in, the calculated result of four factors in χ representation formula (5),Indicate any two differences in many experiments result
Value,Indicate the difference at any two moment in many experiments result, P (Δ T, j) indicates user in the same of different time
Section prefers to the time accounting of travel mode j；Calculated result shows the minimum value of the difference in χ=χn-1,When take
?；Due toWithIt is directly proportional, thereforeIt is minimized, and if only if t=tn-1,When, the time is most short；
(7) in,Indicate the difference of any two location point longitude and latitude in many experiments result, P (Δ L, j) indicates to use
In the same time, different sections of highway prefers to the accounting of travel mode at family；Due toWithIt is directly proportional,It takes most
Small value, and if only if pa=pan-1,When, path is most short；
(8) in,The difference of any travel cost twice in expression many experiments result, and P (Δ W | Δ T, Δ L) it is connection
Probability is closed,
Wherein：P (Δ W, Δ T, Δ L)=P (Δ W, j) P (Δ T, j) P (Δ L, j), P (Δ T, Δ L)=P (Δ T, j) P
(ΔL,j)；Due toWithIt is directly proportional,It is minimized, and if only if mo=mon-1,When,
Cost is minimum；
(9) in,The difference for indicating any convenience of travelling twice in many experiments result, due toWithCheng Zheng
Than,It is minimized, and if only ifConvenience is best；Wherein, c=cv×cp,cvIndicate travel mode choosing
The convenience selected, cpIndicate the convenience of user's travelling；User's travelling convenience cpMeasurement and user travel starting point and purpose
The travel distance on groundRoad conditions jam situation Ltc, available travel mode type KwotEtc. factors it is related；By above
Analysis, it is known thatTherefore
dtIt is parking position far from destination travel distance,It is the travel distance of user's travelling starting point and destination；Pass through knot
It closes (9) and (10) and obtains user's travel mode in the calculating process of convenience；
In order to which the ranking to user's travel mode optimizes adjustment, needs to assign weight to tetra- factors of T, Pa, Mo, C, pass through
Result after four factor weighted sums carries out ranking, and the recommendation of user's travel mode is used for this：
Wherein, S indicates to calculate score, and score value height determines the recommended ranking of travel mode；
χ=[T, Pa, Mo, C] is salt matrices,For adjustable weight matrix, the value range of each weight is 0-
1。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810509174.9A CN108829744B (en) | 2018-05-24 | 2018-05-24 | Travel mode recommendation method based on situation elements and user preferences |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810509174.9A CN108829744B (en) | 2018-05-24 | 2018-05-24 | Travel mode recommendation method based on situation elements and user preferences |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108829744A true CN108829744A (en) | 2018-11-16 |
| CN108829744B CN108829744B (en) | 2021-07-06 |
Family
ID=64145292
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810509174.9A Active CN108829744B (en) | 2018-05-24 | 2018-05-24 | Travel mode recommendation method based on situation elements and user preferences |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108829744B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109829588A (en) * | 2019-02-13 | 2019-05-31 | 苏州大学 | Tensor trajectory path planing method based on context |
| CN109977322A (en) * | 2019-03-05 | 2019-07-05 | 百度在线网络技术（北京）有限公司 | Trip mode recommended method, device, computer equipment and readable storage medium storing program for executing |
| CN110046174A (en) * | 2019-03-07 | 2019-07-23 | 特斯联（北京）科技有限公司 | A kind of population migration analysis method and system based on big data |
| CN110110013A (en) * | 2019-05-10 | 2019-08-09 | 成都信息工程大学 | A kind of entity competitive relation data digging method based on time-space attribute |
| CN110222278A (en) * | 2019-06-18 | 2019-09-10 | 东北大学 | The acquisition methods of user location under composite factor |
| CN110647693A (en) * | 2019-09-23 | 2020-01-03 | 京东城市(北京)数字科技有限公司 | Path recommendation method and device |
| CN111442779A (en) * | 2019-01-16 | 2020-07-24 | 百度在线网络技术（北京）有限公司 | Travel mode recommendation method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101963960A (en) * | 2009-07-22 | 2011-02-02 | 刘旸 | Travel route and scheduling generation method and server |
| CN103955464A (en) * | 2014-03-25 | 2014-07-30 | 南京邮电大学 | Recommendation method based on situation fusion sensing |
| CN105159933A (en) * | 2015-08-06 | 2015-12-16 | 北京百度网讯科技有限公司 | Tourism information recommendation method and apparatus |
| CN106096000A (en) * | 2016-06-22 | 2016-11-09 | 长江大学 | A kind of user based on mobile Internet travel optimization recommend method and system |
| US20170262874A1 (en) * | 2016-03-10 | 2017-09-14 | Mastercard International Incorporated | Methods and system for identifying consumer preferences |
-
2018
- 2018-05-24 CN CN201810509174.9A patent/CN108829744B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101963960A (en) * | 2009-07-22 | 2011-02-02 | 刘旸 | Travel route and scheduling generation method and server |
| CN103955464A (en) * | 2014-03-25 | 2014-07-30 | 南京邮电大学 | Recommendation method based on situation fusion sensing |
| CN103955464B (en) * | 2014-03-25 | 2017-10-03 | 南京邮电大学 | It is a kind of that the recommendation method perceived is merged based on situation |
| CN105159933A (en) * | 2015-08-06 | 2015-12-16 | 北京百度网讯科技有限公司 | Tourism information recommendation method and apparatus |
| US20170262874A1 (en) * | 2016-03-10 | 2017-09-14 | Mastercard International Incorporated | Methods and system for identifying consumer preferences |
| CN106096000A (en) * | 2016-06-22 | 2016-11-09 | 长江大学 | A kind of user based on mobile Internet travel optimization recommend method and system |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111442779A (en) * | 2019-01-16 | 2020-07-24 | 百度在线网络技术（北京）有限公司 | Travel mode recommendation method and device |
| US11313694B2 (en) | 2019-01-16 | 2022-04-26 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for recommending travel way |
| CN109829588A (en) * | 2019-02-13 | 2019-05-31 | 苏州大学 | Tensor trajectory path planing method based on context |
| CN109977322A (en) * | 2019-03-05 | 2019-07-05 | 百度在线网络技术（北京）有限公司 | Trip mode recommended method, device, computer equipment and readable storage medium storing program for executing |
| CN110046174A (en) * | 2019-03-07 | 2019-07-23 | 特斯联（北京）科技有限公司 | A kind of population migration analysis method and system based on big data |
| CN110046174B (en) * | 2019-03-07 | 2020-01-31 | 特斯联（北京）科技有限公司 | population migration analysis method and system based on big data |
| CN110110013A (en) * | 2019-05-10 | 2019-08-09 | 成都信息工程大学 | A kind of entity competitive relation data digging method based on time-space attribute |
| CN110110013B (en) * | 2019-05-10 | 2020-03-24 | 成都信息工程大学 | Entity competition relation data mining method based on space-time attributes |
| CN110222278A (en) * | 2019-06-18 | 2019-09-10 | 东北大学 | The acquisition methods of user location under composite factor |
| CN110222278B (en) * | 2019-06-18 | 2022-05-17 | 东北大学 | Method for acquiring user position under comprehensive factors |
| CN110647693A (en) * | 2019-09-23 | 2020-01-03 | 京东城市(北京)数字科技有限公司 | Path recommendation method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108829744B (en) | 2021-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108829744A (en) | A kind of travel mode recommended method based on situation element and user preference | |
| CN110555112B (en) | Interest point recommendation method based on user positive and negative preference learning | |
| Barth et al. | Shared-use vehicle systems: Framework for classifying carsharing, station cars, and combined approaches | |
| CN110399565A (en) | Based on when null cycle attention mechanism recurrent neural network point of interest recommended method | |
| Bhat et al. | Disaggregate attraction-end choice modeling: formulation and empirical analysis | |
| Zhang et al. | Modeling hierarchical category transition for next POI recommendation with uncertain check-ins | |
| Niu et al. | An online-traffic-prediction based route finding mechanism for smart city | |
| CN108629457B (en) | Travel prediction mode and method and device for building prediction model | |
| Lindner et al. | Estimating motorized travel mode choice using classifiers: An application for high-dimensional multicollinear data | |
| Chen et al. | CEM: A convolutional embedding model for predicting next locations | |
| Ashik et al. | Investigating the impacts of transit-oriented development on transport-related CO2 emissions | |
| CN112800210B (en) | Crowd portrayal algorithm based on mass public transport data | |
| CN116738066B (en) | Rural travel service recommendation method and device, electronic equipment and storage medium | |
| Ha et al. | Determination of the influence factors on household vehicle ownership patterns in Phnom Penh using statistical and machine learning methods | |
| He et al. | Next point-of-interest recommendation via a category-aware Listwise Bayesian Personalized Ranking | |
| Veillette et al. | Does one bicycle facility type fit all? Evaluating the stated usage of different types of bicycle facilities among cyclists in Quebec City, Canada | |
| Xia et al. | A Random Effect Bayesian Neural Network (RE-BNN) for travel mode choice analysis across multiple regions | |
| Huang et al. | PRACE: A taxi recommender for finding passengers with deep learning approaches | |
| CN106446191A (en) | Logistic regression based multi-feature network popular tag prediction method | |
| CN112784177B (en) | Spatial distance adaptive next interest point recommendation method | |
| Liao et al. | An integrated model based on deep multimodal and rank learning for point-of-interest recommendation | |
| Zhou et al. | Refined taxi demand prediction with ST-Vec | |
| He et al. | Gravity model of passenger and mobility fleet origin–destination patterns with partially observed service data | |
| Zheng et al. | A generalized comfort function of subway systems based on a nested logit model | |
| Jiang et al. | Analyzing parking sentiment and its relationship to parking supply and the built environment using online reviews |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |