CN103870604B - Method and apparatus is recommended in tourism - Google Patents
Method and apparatus is recommended in tourism Download PDFInfo
- Publication number
- CN103870604B CN103870604B CN201410136090.7A CN201410136090A CN103870604B CN 103870604 B CN103870604 B CN 103870604B CN 201410136090 A CN201410136090 A CN 201410136090A CN 103870604 B CN103870604 B CN 103870604B
- Authority
- CN
- China
- Prior art keywords
- interest
- point
- model
- user
- type
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 77
- 238000005457 optimization Methods 0.000 claims abstract description 12
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 238000013507 mapping Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 description 12
- 238000001914 filtration Methods 0.000 description 12
- 239000002245 particle Substances 0.000 description 12
- 238000012360 testing method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000000694 effects Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—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
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the present invention provides a kind of tourism and recommends method and apparatus, the method to include：According to the point of interest Type model and point of interest cost model of setting, the utility function model of point of interest is set up；Object function is generated according to the scoring of the historical interest point of the user and utility function model；With the user preference parameters in object function as optimization aim, optimum utility function model is determined；The value of utility of point of interest to be selected is calculated according to optimum utility function model；At least one maximum point of interest to be selected of value of utility is recommended into the user.The present invention considers the tourism favor of user individual and tourism expense, the tour interest point that accuracy can be recommended higher for user.
Description
Technical Field
The invention relates to the field of tourism application, in particular to a tourism recommendation method and device.
Background
The tourism industry is rapidly developed, and becomes one of the largest industries in the world. According to the prediction of world tourism and travel council, the contribution rate of the tourism industry to global GDP is improved from 9.1% to 9.6% in 2011 by 2021. The trend toward numerous travel sites (e.g., Expedia, travel network) is to provide online travel services to tourists. However, the rapidly growing online travel information has brought great difficulty for tourists to select scenic spots that meet their personalized needs. On the other hand, to obtain more business and profits, the tourist enterprise must know these personalized needs and preferences of the tourist and improve better and attractive services. Therefore, the intelligent travel service is urgently needed to be developed and improved for both tourists and travel enterprises.
The existing tourism recommendation system only utilizes basic information of users and scores of scenic spots of various large websites to calculate similarity between the users, and recommends the scenic spots for the users according to the similarity between the users.
Disclosure of Invention
The invention provides a travel recommendation method and device. The personalized travel preference and the personalized travel expense of the user are considered, and the travel interest points with higher accuracy can be recommended to the user.
The invention provides a travel recommendation method, which comprises the following steps:
establishing a utility function model of the interest points according to the set interest point type model and the set interest point cost model;
generating a target function according to the historical interest point scores of the users and the utility function model;
determining an optimal utility function model by taking the user preference parameters in the objective function as optimization objectives;
calculating the utility value of the interest point to be selected according to the optimal utility function model;
and recommending the at least one to-be-selected interest point with the maximum utility value to the user.
The invention also provides a travel recommendation device, which comprises:
the establishing module is used for establishing a utility function model of the interest points according to the set interest point type model and the set interest point cost model;
the generating module is used for generating a target function according to the historical interest point scores of the users and the utility function model;
the determining module is used for determining an optimal utility function model by taking the user preference parameters in the objective function as optimization targets;
the calculation module is used for calculating the utility value of the interest point to be selected according to the optimal utility function model;
and the recommending module is used for recommending the at least one to-be-selected interest point with the maximum utility value to the user.
The invention relates to a travel recommendation method and a device, which establish a utility function model of an interest point according to a set interest point type model and an interest point cost model; generating a target function according to the historical interest point scores of the users and the utility function model; determining an optimal utility function model by taking the user preference parameters in the objective function as optimization targets; calculating the utility value of the interest point to be selected according to the optimal utility function model; and recommending at least one point of interest with the maximum utility value to the user, and when recommending the point of interest to the user, considering the personalized travel preference and travel cost of the user, and recommending the point of interest with higher accuracy to the user.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of one embodiment of a travel recommendation method of the present invention;
FIG. 2 is a graph of experimental results of the scenic spot accuracy of the recommendation of the present invention and other tourist recommendation methods;
FIG. 3 is a graph of experimental results of the recommended scenic spot error rates for the travel recommendation method of the present invention and other travel recommendation methods;
FIG. 4 is a diagram of the results of experiments on the relevance between the recommended sequence of the scenic spots and the optimal sequence of the scenic spots in the method for recommending tourist and other methods for recommending tourist of the present invention;
FIG. 5 is a schematic structural diagram of an embodiment of the travel recommendation device of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a flowchart of an embodiment of a travel recommendation method of the present invention, as shown in fig. 1, an execution subject of the embodiment is a travel recommendation device, which may be implemented by software, hardware, or a combination of software and hardware, and the method includes:
step 101, establishing a utility function model of the interest point according to a set interest point type model and an interest point cost model.
In this embodiment, the interest points may represent scenic spot resources in a tour, such as a great wall, a pyramid, a museum, and the like. The food resources in the travel, such as local speciality restaurants, hot pot restaurants, western-style restaurants, japanese cuisine restaurants, etc., may also be represented, and the interest points may also represent other resources in the travel, which is not limited in this embodiment.
Specifically, the point-of-interest resources have different types and different fees, when the point-of-interest indicates a scenic spot resource in a tour, the fee indicates a ticket fee of the scenic spot, when the point-of-interest indicates a food resource in the tour, the fee indicates per-person consumption of food, and the meaning of the fee indication is different according to the difference of the resources in the tour indicated by the point-of-interest. Taking the point of interest as a tourist attraction resource as an example, the attraction resource can be divided into geological and landform resources, water landscape resources, biological tourist resources, humanistic and ancient resources, religious culture resources, classical garden resources and the like according to different types, and each attraction resource has different entrance ticket fees.
In this embodiment, the set interest point type model is a function of interest point types and user preference parameters of the user, the preference parameters of the user indicate the degree of importance of the user to each interest point type, the user interest point cost model is a function of cost of interest points of the user, in the utility function model for establishing the interest points, when the scenery spot resource types are closer to the preference of the user, the value of the utility function model is larger, when the cost of the scenery spot resource is smaller, the value of the utility function model is larger, and if the value of the utility function model is larger, the scenery spot resource is more worthy of being recommended to the user.
And 102, generating a target function according to the historical interest point score of the user and the utility function model.
In this embodiment, the interest points that each user has gone through may be different, and for different users, the historical interest points that the user has gone through are collected, the collected historical interest points of the user include the types of scenic spots of each historical interest point that the user has gone through, ticket costs, and a score of the user for each historical interest point, and an objective function is generated according to the historical interest point score of the user and a utility function model, and when a difference between the historical interest point score of the user and a value of the utility function model is smaller, a smaller value of the objective function indicates that the utility function model conforms to the preference of the user for selecting the interest point.
And 103, determining an optimal utility function model by taking the user preference parameters in the objective function as optimization targets.
In this embodiment, a minimum value point of the objective function is obtained, when the objective function takes a minimum value, it indicates that the utility function model conforms to the preference of the user for selecting the interest point, and the user preference parameter at this time is the optimal preference parameter of the user, and the optimal preference parameter of the user is brought into the utility function model, so as to determine the optimal utility function model of the user.
And 104, calculating the utility value of the interest point to be selected according to the optimal utility function model.
In this embodiment, the to-be-selected interest points are interest points that have not been passed by the user, the types of the scenic spots and the cost of the to-be-selected interest points are brought into the optimal utility function model, and the utility value of each to-be-selected interest point is calculated.
105, recommending at least one interest point to be selected with the maximum utility value to the user
Specifically, one to-be-selected interest point with the largest utility value can be recommended to the user, the calculated utility values of the to-be-selected interest points can be sorted in a descending order, the to-be-selected interest points corresponding to the first K large utility values are selected and recommended to the user, the user selects the first K recommended to-be-selected interest points, and the to-be-selected interest points satisfied by the user are obtained in the interaction process with the user.
In the embodiment, a utility function model of the interest point is established according to a set interest point type model and an interest point cost model; generating a target function according to the historical interest point score of the user and the utility function model; determining an optimal utility function model by taking the user preference parameters in the objective function as optimization targets; calculating the utility value of the interest point to be selected according to the optimal utility function model; and recommending at least one point of interest with the maximum utility value to the user, and when recommending the point of interest to the user, considering the personalized travel preference and travel cost of the user, and recommending the point of interest with higher accuracy to the user.
Further, establishing a utility function model of the interest point according to the set interest point type model and the set interest point cost model, specifically:
establishing a utility function model of the interest point by adopting an equation (1),
U(X,C)＝α*U1(X)+β*U2(C) (1)
wherein α + β =1,0 < α < 1,0 < β < 1, U1(X) represents a point of interest type model, U2(C) The model representing the cost of the point of interest, α represent the weight of the model representing the type of the point of interest and the model representing the cost of the point of interest respectively, and for different users, the types and the cost of the point of interest have different attention degrees, so that for different users, the value of α can be changed according to the requirements of the users.
In this embodiment, the established utility function model of the interest point is linearly separable with respect to the interest point type model and the interest point cost model, different weights can be set for the types and cost attention degrees of the interest points of different users, and the requirements of different users can be more flexibly met.
In economics, the kobuki-douglas function can measure how well a consumer likes a commodity, in the form:
u(y1,y2)＝y1 a*y2 b(2)
wherein, y1And y2The quantities of the two types of goods are represented, a and b describe the preference of the consumer for the two types of goods 0 ≦ a, b ≦ 1, and a + b ≦ 1, respectively.
Therefore, in this embodiment, in order to measure the preference of the user for the type of the point of interest, the set point of interest category model is:
wherein,x represents an interest point type vector, and X is (X)1,x2，……xi,……，xn)，xiIs the ith dimension component of the interest point type vector X and indicates whether the interest point belongs to the ith type or not, when the interest point belongs to the ith type, XiThe value is 1, when the interest point does not belong to the ith type, xiA value of 0, xi∈ {0,1}, when the interest point belongs to one or more different interest point types, one or more components in the interest point type vector represented by X are 1, and when the interest point does not belong to any interest point type, the values of the components in the interest point type vector represented by X are all 0.αiA preference parameter representing the user's preference for the ith type of point of interest, 0 ≦ αiIs less than or equal to 1, n represents the type number of the interest points, and the preference parameters of the n interest point types are taken as elements to form an n-dimensional vector representing the preference parameters of the interest type of the user, wherein the vector is represented by A = (α)1,α2，……αi,……，αn). The value of the vector representing the interest type preference parameter will be different for different users.
The scenic spot resources are explained by taking the interest points as the scenic spot resources, for example, the scenic spot resources are divided into six types of resources, namely geological and geomorphic resources, water landscape resources, biological tourism resources, humanistic and ancient resources, religious culture resources and classical garden resources according to different types, and the six types of scenic spot resources are sequentially divided into (X) and (X) by using a vector X1,x2，……xi,……，xn) One component in (1) indicates that if the attraction resource is a great wall, the attraction resource type of the great wall is a human historic resource, X = (0,0,0,1,0,0), and if the user has a preference for only one attraction resource type, human historic resource, a = (0,0,0,1,0, 0).
In this embodiment, U1(X) has a value range of [0,1 ]]. When a certain interest point does not belong to any interest point type, i.e. X ═ 0,0,0, the interest point type model takes the minimum value, i.e. U1(X) =0, and when the type of the interest point is completely matched with the user preference distribution, the interest point type model takes the maximum value, namely U1(X) = 1. As in the above example, the type of great wall is a perfect match to the user preference distribution, where the point of interest type model takes the maximum value, U1(X)=1。
Further, the cost model of the interest point can be expressed as shown in formula (4)
Wherein C represents the cost of the point of interest, CmaxRepresenting the most expensive cost in the historical interest points which are passed by the user, when the cost of the interest points is higher, the value of the interest point cost model is smaller, which indicates that the probability of the user selecting the interest points is smaller, when the cost exceeds a certain value, the value of the interest point cost model is zero, U is equal to U2(C) Has a value range of [0,1 ]]When the cost of the interest point is zero, the value of the cost model of the interest point is maximum, and when the cost spent by the user is the historical cost, the interest point is shown to reach the maximum cost acceptable by the user, and the value of the cost model of the interest point is minimum.
In this embodiment, generating a target function according to the historical interest point score of the user and the utility function model specifically includes:
generating an objective function according to equation (5);
wherein,m represents the historical interest point number of the user, rjRepresents the user's score for the jth historical point of interest, and K represents the mapping constant. Since the user scores r for historical points of interestjIs any integer of 0-5, and the value of the utility function model is [0,1 ]]Therefore, the mapping constant K takes a value of 5. In the objective function represented by the formula (5), the smaller the value of the objective function is, the more the utility function model meets the requirement of the user for selecting the interest point.
Preferably, a gradient descent method is used when determining the optimal value of the user preference parameter in the objective function.
The idea of the gradient descent method is to use the negative gradient direction to determine a new search direction for each iteration, so that each iteration can gradually reduce the optimized objective function, when the difference between the optimized objective function value obtained in the last iteration and the optimized objective function value obtained after the current iteration is smaller than a certain threshold value, the iteration is ended, and at this time, the corresponding A in the objective function is the optimal value of the user preference parameter.
And (6) determining an optimal utility function model.
Wherein α + β =1,0 < α < 1,0 < β < 1, U1(X) represents a point of interest type model, U2(C) Representing an interest point cost model, α representing the weight of the interest point type model and the interest point cost model respectively, X representing an interest point type vector, XiRepresenting the ith dimension component of the point of interest type vector X, αiAnd the preference parameter of the user on the ith interest point type is shown, and n is the type number of the interest points.
And (3) bringing the optimal user preference parameter into the formula (6) to obtain an optimal utility function model of the user, bringing the relevant parameter of the interest point to be selected by the user into the optimal utility function to obtain the utility value of the interest point to be selected, and recommending at least one interest point with the maximum utility value to the user.
The effective results of this example can be further illustrated by the following simulations:
1, simulation content: in the tourism recommendation method (GD) and the method (PSO) adopting the particle swarm algorithm to calculate the user preference parameter, the training set is used for determining the optimal utility function model of the user, and in the traditional collaborative filtering tourism recommendation method (CF), the training set is used for determining other users with the highest similarity to the user so as to recommend the scenic spots which are historically used by other users with the highest similarity to the user. The test set is used for testing the performance of scenic spots recommended by the tourism recommendation method (GD) and the method (PSO) for calculating the user preference parameter by adopting a particle swarm algorithm and the traditional collaborative filtering tourism recommendation method. And evaluating the performances of the methods from the recommendation accuracy of the scenic spots recommended by each method, the recommendation error rate of the scenic spots recommended by each method, and the relevance between the sequence of the scenic spots recommended by each method and the optimal sequence of the scenic spots.
2 simulation experiment results
A, experimental results of scenic spots recommended by the methods in the aspect of recommendation accuracy
By using the tourism recommendation method (GD) and the user preference parameter method (PSO) calculated by adopting the particle swarm algorithm, the traditional collaborative filtering tourism recommendation method (CF) respectively recommends the first K most attractive interest points for a user, P represents the set of scenic spots with the user score not less than 4 in the recommended scenic spots, T represents the set of scenic spots in the test set of the user, and the recommendation accuracy Precision @ K of the scenic spots is represented by formula (6):
FIG. 2 is a graph of experimental results of the tourist recommendation method and other tourist recommendation methods of the present invention in the accuracy of recommended scenic spots, where the X-axis in FIG. 2 represents the value of the recommended number of points of interest K, and the ordinate represents the recommendation accuracy, and it can be seen from FIG. 2 that the tourist recommendation method (GD) of the present invention has the highest recommendation accuracy, the recommendation accuracy of the particle swarm algorithm for calculating the user preference Parameter (PSO) is slightly lower than that of the tourist recommendation method (GD) of the present invention, and the recommendation accuracy of the conventional collaborative filtering tourist recommendation method (CF) is the lowest, so the tourist recommendation accuracy of the tourist recommendation method of the present invention is superior to that of the particle swarm algorithm for calculating the user preference Parameter (PSO) and that of the conventional collaborative filtering tourist recommendation method (CF).
Experiment result of scenic spots recommended by methods B in recommendation error rate aspect
The tourism recommendation method (GD) and the particle swarm algorithm-based user preference parameter calculation method (PSO) are used, the traditional collaborative filtering tourism recommendation method (CF) respectively recommends the first K most attractive interest points for the user, when the score of the user is less than 4, the user is considered to dislike the scenic spot, and if the scenic spots exist in the recommendation result, the recommendation result is inaccurate. The recommended error rate is therefore defined as follows: q is a scenic spot set with the user score smaller than 4 in recommended scenic spots, T is a scenic spot set in test concentration, and the recommended error rate ErrorRate @ K of the scenic spot is expressed as shown in formula (7):
FIG. 3 is a graph of the experimental results of the recommendation error rate of the travel recommendation method and other travel recommendation methods of the present invention, wherein the X-axis in FIG. 3 represents the value of the recommended number of interest points K, the ordinate represents the recommendation error rate, and the lower the recommendation error rate represents that the recommended scenic spots of the method meet the user requirements better, and it can be seen from FIG. 3 that the travel recommendation method (GD) of the present invention has the lowest recommendation error rate, the particle swarm algorithm-based method for calculating user preference Parameters (PSO) has a slightly higher recommendation error rate than the method (GD) of the present invention, and the traditional collaborative filtering travel recommendation method (CF) has the highest recommendation error rate, so the travel recommendation method of the present invention is superior to the particle swarm algorithm-based method for calculating user preference Parameters (PSO) and the traditional collaborative filtering travel recommendation method (CF) in the recommendation error rate.
C, the method recommends experimental results on the correlation between the sequences of the scenic spots and the optimal sequences of the scenic spots
The tourism recommendation method (GD) and the particle swarm algorithm-based method (PSO) for calculating the user preference parameters are used, the traditional collaborative filtering tourism recommendation method (CF) respectively recommends the first K most attractive interest points for the user, then the scenic spot sequences recommended by each method are compared with the optimal sequence of the scenic spot selected by the user, and the NDCG @ K of the relevance between the scenic spot sequences recommended by each method and the optimal sequence of the scenic spot is expressed as the following formula (8):
wherein,
i is subscript of scenery in scenery sequence, riFor the user's score for the ith attraction, iDCG @ K is the value of DCG @ K under the optimal sequence of attractions.
The standard for NDCG @ K is based on the following two assumptions:
(1) the higher the correlation, the better the sights that are ranked in front.
(2) Scenic spots with high relevance can better meet the requirements of the user than scenic spots with low relevance.
FIG. 4 is a graph of the experimental results of the correlation between the recommended sequence of the scenery spot and the optimal sequence of the scenery spot for the tourism recommendation method of the present invention and other tourism recommendation methods, wherein the black bar in FIG. 4 represents the correlation between the sequence of the scenery spot with the number of 5 and the optimal sequence of the scenery spot with the number of 5 recommended by each method, and the white bar represents the correlation between the sequence of the scenery spot with the number of 10 and the optimal sequence of the scenery spot with the number of 10 recommended by each method, and the greater the correlation between the recommended sequence of the scenery spot and the optimal sequence of the scenery spot indicates that the scenery spot recommended by the method meets the requirements of the user better, as can be seen from FIG. 4, when the tourism recommendation method (GD) of the present invention is K =5 and K =10, the NDCG @ K values reach 0.6502 and 0.6649 respectively, which are higher than the NDCG @ K values when the particle swarm algorithm is used to calculate the user preference parameter method (PSO) and the traditional collaborative filtering tourism recommendation method (CF) at K =, therefore, the method (GD) is superior to a method (PSO) for calculating user preference parameters by adopting a particle swarm algorithm and a traditional collaborative filtering tourism recommendation method (CF) in the aspect of correlation between the recommended sequence of the scenic spot and the optimal sequence of the scenic spot.
Therefore, the method for recommending the tourism (GD) is superior to a method for calculating user preference Parameters (PSO) and a traditional collaborative filtering tourism recommendation method (CF) by adopting a particle swarm algorithm in the three aspects of recommending accuracy of scenic spots, recommending error rate of scenic spots, and correlation between recommended sequences of the scenic spots and optimal sequences of the scenic spots.
Fig. 5 is a schematic structural diagram of an embodiment of the travel recommendation device of the present invention, as shown in fig. 5, the device may include: an establishing module 501, a generating module 502, a determining module 503, a calculating module 504 and a recommending module 505.
The establishing module 501 is configured to establish a utility function model of the point of interest according to a set point of interest type model and a set point of interest cost model.
A generating module 502, configured to generate an objective function according to the historical interest point score of the user and the utility function model.
A determining module 503, configured to determine an optimal utility function model by using the user preference parameter in the objective function as an optimization target.
And the calculating module 504 is configured to calculate a utility value of the to-be-selected interest point according to the optimal utility function model.
And the recommending module 505 is configured to recommend the at least one point of interest to be selected with the largest utility value to the user.
The apparatus of this embodiment may execute the technical solution of the method embodiment shown in fig. 1, and the implementation principle and the technical effect are similar, which are not described herein again.
Further, the establishing module 501 is configured to establish a utility function model of the interest point according to the set interest point type model and the set interest point cost model, specifically:
using U (X, C) ═ α × U1(X)+β*U2(C) Establishing a utility function model of the interest points,
wherein α + β =1,0 < α < 1,0 < β < 1, U1(X) represents a point of interest type model, U2(C) Representing the point of interest cost model, α representing the weight of the point of interest type model and the point of interest cost model, respectively.
Specifically, the interest point category model is as follows:
wherein,x denotes a point of interest type vector, XiRepresenting the ith dimension component of the point of interest type vector X, αiA preference parameter representing the user's preference for the ith type of interest point, n represents the interest pointThe number of types of (2).
The point of interest cost model is:
wherein C represents the cost of the point of interest, CmaxIndicating the most expensive cost of historical points of interest that the user has gone.
A generating module 502, configured to generate a target function according to the historical interest point score and the utility function model of the user, specifically:
according toGenerating the objective function;
wherein,m represents the historical interest point number of the user, rjRepresents the jth historical point of interest score, K represents the mapping constant, and K is 5.
The determining module 503 is configured to determine an optimal utility function model by using the user preference parameter in the objective function as an optimization objective, and specifically includes:
determining the optimal value of the user preference parameter in the objective function by adopting a gradient descent method;
according to the optimal value of the user preference parameter, adoptingAnd determining an optimal utility function model.
Wherein α + β =1,0 < α < 1,0 < β < 1, U1(X) represents a point of interest type model, U2(C) Representing an interest point cost model, α representing the weight of the interest point type model and the interest point cost model respectively, X representing an interest point type vector, XiRepresenting interestsThe i-th component of the point type vector X, αiAnd the preference parameter of the user on the ith interest point type is shown, and n is the type number of the interest points.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A travel recommendation method is characterized by comprising the following steps:
establishing a utility function model of the interest points according to the set interest point type model and the set interest point cost model;
generating a target function according to the historical interest point scores of the users and the utility function model;
determining an optimal utility function model by taking the user preference parameters in the objective function as optimization objectives;
calculating the utility value of the interest point to be selected according to the optimal utility function model;
and recommending the at least one to-be-selected interest point with the maximum utility value to the user.
2. The method according to claim 1, wherein the utility function model of the point of interest is established according to the set point of interest type model and the set point of interest cost model, and specifically comprises:
using U (X, C) ═ α × U1(X)+β*U2(C) Establishing a utility function model of the interest points;
wherein α + β is 1,0 < α < 1,0 < β < 1, U1(X) represents the point of interest type model, U2(C) Representing the point of interest cost model, α representing the weight of the point of interest type model and the point of interest cost model, respectively.
3. The method of claim 2, wherein the point of interest type model is:
wherein,
x denotes a point of interest type vector, XiRepresenting the i-th component of the point of interest type vector X, αiThe preference parameter of the user on the ith interest point type is represented, and n represents the type number of the interest points;
the point of interest cost model is as follows:
wherein C represents the cost of the point of interest, CmaxRepresenting the most expensive cost among the historical points of interest that the user has gone.
4. The method according to claim 2 or 3, wherein the generating of the objective function according to the historical interest point scores of the user and the utility function model comprises:
according toGenerating the objective function;
wherein,m represents the historical interest point number of the user, rjRepresents the jth historical point of interest score and K represents the mapping constant.
5. The method according to claim 4, wherein the determining an optimal utility function model with the user preference parameter in the objective function as an optimization objective specifically comprises:
determining the optimal value of the user preference parameter in the objective function by adopting a gradient descent method;
according to the optimal value of the user preference parameter, adoptingDetermining an optimal utility function model;
wherein α + β is 1,0 < α < 1,0 < β < 1, U1(X) represents the point of interest type model, U2(C) Representing the interest point cost model, α representing the weight of the interest point type model and the interest point cost model respectively, X representing the interest point type vector, X representing the weight of the interest point type vectoriRepresenting the i-th component of the point of interest type vector X, αiAnd the preference parameter of the user to the ith interest point type is represented, and n represents the type number of the interest points.
6. A travel recommendation device, comprising:
the establishing module is used for establishing a utility function model of the interest points according to the set interest point type model and the set interest point cost model;
the generating module is used for generating a target function according to the historical interest point scores of the users and the utility function model;
the determining module is used for determining an optimal utility function model by taking the user preference parameters in the objective function as optimization targets;
the calculation module is used for calculating the utility value of the interest point to be selected according to the optimal utility function model;
and the recommending module is used for recommending the at least one to-be-selected interest point with the maximum utility value to the user.
7. The apparatus of claim 6,
the establishing module is used for establishing a utility function model of the interest point according to the set interest point type model and the set interest point cost model, and specifically comprises the following steps:
using U (X, C) ═ α × U1(X)+β*U2(C) Establishing a utility function model of the interest points;
wherein α + β is 1,0 < α < 1,0 < β < 1, U1(X) represents the point of interest type model, U2(C) Representing the point of interest cost model, α representing the weight of the point of interest type model and the point of interest cost model, respectively.
8. The apparatus of claim 7, wherein the point of interest category model is:
wherein,
x denotes a point of interest type vector, XiRepresenting the i-th component of the point of interest type vector X, αiThe preference parameter of the user to the ith interest point type is represented, and n represents the type number of the interest points;
the point of interest cost model is as follows:
wherein C represents the cost of the point of interest, CmaxRepresenting the most expensive cost of the user's historically visited points of interest.
9. The apparatus according to claim 7 or 8,
the generating module is configured to generate a target function according to the historical interest point score of the user and the utility function model, and specifically includes:
according toGenerating the objective function;
wherein,m represents the historical interest point number of the user, rjRepresents the jth historical point of interest score and K represents the mapping constant.
10. The apparatus of claim 9,
the determining module is configured to determine an optimal utility function model by using the user preference parameter in the objective function as an optimization objective, and specifically includes:
determining the optimal value of the user preference parameter in the objective function by adopting a gradient descent method;
according to the optimal value of the user preference parameter, adoptingDetermining an optimal utility function model;
wherein α + β is 1,0 < α < 1,0 < β < 1, U1(X) represents the point of interest type model, U2(C) Representing the interest point cost model, α representing the weight of the interest point type model and the interest point cost model respectively, X representing the interest point type vector, X representing the weight of the interest point type vectoriRepresenting the i-th component of the point of interest type vector X, αiAnd the preference parameter of the user to the ith interest point type is represented, and n represents the type number of the interest points.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410136090.7A CN103870604B (en) | 2014-04-04 | 2014-04-04 | Method and apparatus is recommended in tourism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410136090.7A CN103870604B (en) | 2014-04-04 | 2014-04-04 | Method and apparatus is recommended in tourism |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103870604A CN103870604A (en) | 2014-06-18 |
| CN103870604B true CN103870604B (en) | 2017-03-29 |
Family
ID=50909134
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410136090.7A Active CN103870604B (en) | 2014-04-04 | 2014-04-04 | Method and apparatus is recommended in tourism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103870604B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699812A (en) * | 2015-03-23 | 2015-06-10 | 苏州大学张家港工业技术研究院 | Tourism information pushing method and system |
| CN106202252A (en) * | 2016-06-29 | 2016-12-07 | 厦门趣处网络科技有限公司 | Method, system are recommended in a kind of trip analyzed based on user emotion |
| CN109409916B (en) * | 2017-08-18 | 2021-10-22 | 重庆赫皇科技咨询有限公司 | Tourism marketing system based on big data platform |
| CN107545471B (en) * | 2017-09-19 | 2021-06-11 | 北京工业大学 | Big data intelligent recommendation method based on Gaussian mixture |
| CN109376294A (en) * | 2018-08-22 | 2019-02-22 | 中国平安人寿保险股份有限公司 | User interest point Weight Determination and device, storage medium and electronic equipment |
| CN112149010B (en) * | 2020-11-01 | 2024-05-24 | 云境商务智能研究院南京有限公司 | Group travel route recommendation method based on attention mechanism |
| CN116595259A (en) * | 2021-03-25 | 2023-08-15 | 支付宝(杭州)信息技术有限公司 | Position recommendation processing method and device |
| CN112989214A (en) * | 2021-05-20 | 2021-06-18 | 湖北游游智游网络科技有限公司 | Tourism information display method and related equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102483827A (en) * | 2009-04-14 | 2012-05-30 | Jda软件集团有限公司 | Travel price optimization (tpo) |
| CN103020254A (en) * | 2012-12-20 | 2013-04-03 | 北京趣拿信息技术有限公司 | Information recommending method and device |
| CN103020308A (en) * | 2013-01-07 | 2013-04-03 | 北京趣拿软件科技有限公司 | Method and device for recommending travel strategy project |
| CN103678429A (en) * | 2012-09-26 | 2014-03-26 | 阿里巴巴集团控股有限公司 | Recommendation method and device of tour routes |
-
2014
- 2014-04-04 CN CN201410136090.7A patent/CN103870604B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102483827A (en) * | 2009-04-14 | 2012-05-30 | Jda软件集团有限公司 | Travel price optimization (tpo) |
| CN103678429A (en) * | 2012-09-26 | 2014-03-26 | 阿里巴巴集团控股有限公司 | Recommendation method and device of tour routes |
| CN103020254A (en) * | 2012-12-20 | 2013-04-03 | 北京趣拿信息技术有限公司 | Information recommending method and device |
| CN103020308A (en) * | 2013-01-07 | 2013-04-03 | 北京趣拿软件科技有限公司 | Method and device for recommending travel strategy project |
Non-Patent Citations (3)
| Title |
|---|
| A travel sequence recommendation approach based on markov model;dalu guo et al;《IADIS international conference on WWW/Internet 2012》;20121021;全文 * |
| inter-temporal variation in the travel time and travel cost parameters of transport models;borjesson.m;《transportation》;20140331;第41卷(第2期);全文 * |
| 基于用户兴趣建模的推荐方法及应用研究;刘淇;《中国博士学位论文全文数据库》;20131015(第10期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103870604A (en) | 2014-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103870604B (en) | Method and apparatus is recommended in tourism | |
| Chen et al. | Distributed individuals for multiple peaks: A novel differential evolution for multimodal optimization problems | |
| CN108460101B (en) | Point of interest recommended method of the facing position social networks based on geographical location regularization | |
| CN107451894B (en) | Data processing method, device and computer readable storage medium | |
| JP6356789B2 (en) | Main knowledge point recommendation method and system | |
| CN103473291A (en) | Personalized service recommendation system and method based on latent semantic probability models | |
| CN106021456B (en) | Interest point recommendation method fusing text and geographic information in local collaborative arrangement | |
| CN105631037B (en) | A kind of image search method | |
| Guàrdia-Sebaoun et al. | Latent trajectory modeling: A light and efficient way to introduce time in recommender systems | |
| CN105430099A (en) | Collaborative Web service performance prediction method based on position clustering | |
| CN103377296A (en) | Data mining method for multi-index evaluation information | |
| CN104239496A (en) | Collaborative filtering method based on integration of fuzzy weight similarity measurement and clustering | |
| CN108897750B (en) | Personalized place recommendation method and device integrating multiple contextual information | |
| CN108074016A (en) | Customer relationship intensity prediction method, device and equipment based on position social networks | |
| Thiengburanathum et al. | A decision tree based recommendation system for tourists | |
| CN106095974A (en) | Commending system score in predicting based on network structure similarity and proposed algorithm | |
| Mesran et al. | Comparative Analysis of MOORA and MOOSRA Methods in Determining Prospective Students Recipient of the Indonesian Smart Card (KIP) | |
| CN107305559A (en) | Method and apparatus are recommended in one kind application | |
| CN107944750A (en) | A kind of poverty depth analysis method and system | |
| Ioannou et al. | Adalip: An adaptive learning rate method per layer for stochastic optimization | |
| CN110263250A (en) | A kind of generation method and device of recommended models | |
| CN109460474A (en) | User preference trend method for digging | |
| CN103218457B (en) | People search based on social networks and discovery method and system | |
| CN108985898B (en) | Site scoring method and device and computer readable storage medium | |
| CN110609963A (en) | Recommendation list generation method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |