Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Research on the Recommendation Algorithm of Rural Tourism Routes Based on the Fusion Model of Multiple Data Sources

Research on the Recommendation Algorithm of Rural Tourism Routes Based on the Fusion Model of... Hindawi Discrete Dynamics in Nature and Society Volume 2022, Article ID 2262148, 10 pages https://doi.org/10.1155/2022/2262148 Research Article Research on the Recommendation Algorithm of Rural Tourism Routes Based on the Fusion Model of Multiple Data Sources 1 1 2 Hong Li, Man Qiao, and Shuai Peng Henan University of Animal Husbandry & Economy, Zhengzhou 450046, Henan, China Guangdong University of Finance and Economics, Guangzhou 510320, Guangdong, China Correspondence should be addressed to Shuai Peng; 20191021@gdufe.edu.cn Received 23 November 2021; Accepted 21 March 2022; Published 14 April 2022 Academic Editor: Wei Zhang Copyright©2022HongLietal.+isisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Rural tourism has become an important force in implementing the rural revitalisation strategy and accelerating rural economic development. +e hectic pace of life has made more and more city dwellers yearn for rural life, and travelling in the countryside has become their weekend choice. However, the current level of rural tourism informationization is low, the publicity is in- sufficient, the tourists’ awareness is low, and the source of customers is seriously insufficient. To this end, this paper proposes a relatively novel multidata source fusion tourism recommendation algorithm, which adopts the idea of tensor orthogonal de- composition and fuses multisource data models to predict the target domain’s for rating. +e integrated consideration of multiple data sources under the do-it-yourself approach assists the target domain to discover the target user neighbourhood users more quickly and to discover the user’s interest degree more accurately. It is worth pointing out that the recommendation algorithm proposedinthispaperunderthefusionofmultipledatasourcesisnotnecessarilyapplicabletodatasourceswithweakcorrelation, such as travel data sources and music data sources, which are relatively weakly correlated, and the algorithm is slightly weak in making predictions of user preferences. operation [3]. However, in the traditional matrix decom- 1. Introduction position process, the data structure information is often lost With the national economy entering a new normal, tourism in the matrix decomposition process due to data sparsity, hasusheredinagoldenperiodofrapiddevelopment[1],and making the results distorted. Tensor is a way of storing rural tourism has become an important part of China’s multidimensional data, and the concept of tensor decom- tourism industry. +e busy pace of life has made more and position is based on the idea of matrix completion, which more city dwellers yearn for rural life, and travelling in the aims to fill in the missing (or unobservable) parts of the countryside has become their weekend choice. However, at target matrix [4]. In simple terms, this means that matrix A present, the level of information technology for rural is used to approximately evaluate matrix B (there is some tourism is low, the publicity is not strong enough, tourists inherent correlation between A and B). +e invisible parts of the matrix B are filled in by the matrix A. are less aware of it, and the source of visitors is seriously insufficient. In this context, rural tourism urgently needs +is study, through the platform of rural tourism information technology to increase publicity and improve products on the malefactor research, found that the content service levels [2]. oftheseruraltourismproductsismainlyconcentratedinthe Inpracticalrecommendationsystems,themostcommon better economy and tourism industry of more developed method of fusing multiple data sources is matrix decom- areas, while the economy in relatively backward areas is not position. Decomposed user feature matrix U and item covered [5] or there is also incomplete information, not feature matrix V are obtained by training the loss function, enough to meet the rising demand of users. For example, the and finally the scoring matrix is reduced by matrix inverse interface of searching for any area on the Nongjiale platform 2 Discrete Dynamics in Nature and Society +rough the study of the above rural tourism wisdom simply shows a little of the same scenery and cuisine in- formation, there is no user search function on the rural products, we found that these rural tourism wisdom products are mainly concentrated in the more developed tourism service platform, and the product description is simple, which cannot hook the people’s desire to travel [6]. tourism industry and the better economic regions, while the +e development of rural tourism products has a very far- relatively more economically backward regions rural tour- reaching significance in promoting the economic develop- ism wisdom products either do not have complete infor- ment of rural areas, combining field research, literature mation or are not covered, and the scope of operation is analysis, and data from existing rural tourism platforms to narrow, the services provided are limited and not well construct a rural tourism product model and a user model. known, and their service content cannot meet the needs of +e aim is to improve the quality and efficiency of users’ users [11]. For example, the interface of searching for any access to useful information, so that rural tourism products area on the WeChat platform of Nongjiale is just to show a littlesceneryanddishinformation,thereisnousersearchon are more accurately submitted to users. Combining users’ personalised characteristics to recommend the rural tourism the WeChat miniprogram of rural tourism service platform, and the product introduction is simple, which cannot hook products they really need to have has become a valuable and challenging research topic. the people’s desire to travel. +rough literature research, the academic world has also seen moreresearchonintelligentruraltourism;forexample, 2. Related Work [12] published in the China Tourism News about the new While China’s economy is growing rapidly and people’s mission given to tourism in the new era, [13] conducted an living standards aresteadily rising, the tourismindustry, as a in-depth analysis of the problems in the development of sunrise industry, is receiving increasing attention from the rural tourism information technology and believes that the government and enterprises [7], and tourism has become an factors that inhibit the development of rural tourism in- formation technology are publicity, information manage- importantdrivertostimulateconsumptionandinducerapid upgrading and transformation of the industry. Just as ment level, information infrastructure construction, etc., [14] argues that there is a certain difference between the China’s tourism industry is moving towards mass tourism and global tourism, the development of information tech- demand and supply of tourists in rural tourism and that nology and mobile Internet applications has given rise to a measures should be provided to reduce the difference be- new concept of smart tourism for the tourism experience. tween the two, [15] designed an ecological service system for +e combination of tourism and information technology rural tourism from three aspects: service design process, constitutes smart tourism, which is a necessary path for the interactive experience, and branding, and [16] argues that, current development of tourism [8]. with the help of mobile Internet, a seamless connection Tourism websites, tourism APPs, and tourism WeChat between tourism enterprises and tourists can be achieved applets are current manifestations of smart tourism, and and a bridge for efficient communication between them can be built. with the rapid development of smart phones and mobile networks, many smart tourism products have emerged, such Although smart tourism is a unique concept in China, similar projects with smart tourism have emerged inter- as Ctrip, founded in 1999, whose mobile APP was launched in 2010, and its acquisition of the UK-based airfare search nationally earlier than in China, with USA, Singapore, platform Skyscanner Limited was in November 2016, which Korea, England, and other countries being among the more means that Ctrip has started to enter the road of inter- representative ones. +e smart wrist system online in the nationalisation [3, 9]. Where to Go was established in 2005, United States in 2005 opened the beginning of smart and in 2010 it launched its APP, the same as Ctrip; it is a tourism; the feedback system equipped with radio frequency comprehensive travel APP that provides a collection of technology positioning device one by one Mountainwatch wasthefirsttobeusedbytheColoradoSteamboatSkiResort business travel management, hotels, airline tickets, holiday booking, and travel information. In October 2015, the two in the United States, which can provide tourists with real- time consumption and ski routes. Touchwood, a service travelgiantAPPs announcedamerger. In2006, theMaHive travel website was launched and became popular with users, platformin Seoul,South Korea, is aimed at self-guided,rural travellers, who can use the platform to perceive tourist and its mobile APP was launched in 2011. Just as smart tourism was being widely promoted and applied, China’s information [17, 18]. rural tourism industry also stepped into a path of rapid development [10]. Previous rural tourism products can no 3. Construction of User Ontology Model for longer meet the tourism needs of the people, and there is an Rural Tourism Platform urgent need to introduce intelligent rural tourism into rural tourism. +at is why MPPs such as Meiju Countryside, Find Personalised recommendation for users is the ultimate goal the Yard, Go Farm, He Xiang You, Meet the Countryside, of an intelligent recommendation system, and the con- and Down to the Country Guest have emerged, and cor- struction of a good user ontology model is a prerequisite for responding WeChat platforms such as Countryside Tourism implementing intelligent recommendations. +e user on- Service Platform, Countryside Tourism Merchant, Nongjia tology model in this study requires the acquisition of data Platform, and Shanghai Nongjia Platform have appeared on relatedtotheuser’sinterestsinthefieldofruraltourism,and WeChat mini-programs. then combining the user’s personal information to Discrete Dynamics in Nature and Society 3 automatically build their interest models. +e user model is eventually build a model that can be recognised by our computer. +e construction process of the user personalised an abstract representation of the personal information and interest information of the system users. +e basic personal interest model [19] is shown in Figure 1. information in the user model is relatively stable, while the user’s interest information fluctuates greatly with time and 3.1. Countryside Tourism Platform User Information. +e product attributes, so how to accurately express the user’s format and quality of the acquired data directly affect the interest and facilitate the calculation becomes the key to the quality of the user model. Currently, there are two common implementation of the user model. techniques used to obtain user interest information, Considering the probabilistic topic representation, the specifically: core idea is to understand each text message as a mixture of multiple topical features,where each topic is thedistribution (1) Displayfeedback technology: the main wayto record probability of the corresponding feature [22]. user preferences is through the user’s evaluation of In the rural tourism platform study, the rural tourism the corresponding product. +is requires the user to product ontology is used to represent the user’s interests, activelyparticipateandactivelyevaluate theproduct, which is relatively simple in structure and can be imple- which takes up more of the user’s time and does not mented by the keyword vector space representation method. allow for good access to the user’s personal interest +e user model consists of the user’s interest set, the user’s information when the user’s participation is not interest attribute set, the user’s attention and weight for each high. interestintheinterestattributevalues,and3parts,whichare (2) Implicit feedback technology: it is mainly through the user’s interest set, i.e., the product’s key attribute set; the the system to view and analyse the user’s behavioural user’s interest attribute set, i.e., the product attribute value data to obtain personalised information about the set. user’s interests. Implicit feedback technology does not require active participation of the user, mainly throughtheplatform’sback-endsystemtoobtainthe 3.4. Dynamic User Interest Model. +e user model contains corresponding information of the user, such as the two types of information: static, such as basic information number of visits to a product, the length of stay of about the user, such as gender, occupation, and knowledge each visit (because of the number of visits to background; and dynamic, which changes over time. In this products of interest to the user, the length of stay study, we mainly consider the part of dynamic updates. may be longer), the number of searches, etc. Users’ interests change with the environment and psycho- logical factors, so the common methods used to update dynamicusermodelsaretheforgettingfunctionmethodand 3.2. Personal Information of Rural Tourism Platform Users. the time window method. +e forgetting function method is Usually, the user’s basic personal situation will make the similar to the law of memory forgetting in that, without user’s interests relatively stable [20]; for example, users with external stimuli, a user’s interest in something will decay young children have a greater interest in parent-child rural over time [23]. In this study, the user interest model is tourism tours, and low-income people will generally choose considered to change with the combination of time for- rural tourism products with lower or free costs. getting and frequency of access. User interest forgetting In this paper, the authors conclude from the corre- function is used to dynamically update the user interest sponding literature, research on existing user information model based on the forgetting factor, which is calculated on rural tourism platforms, and field surveys that the per- using sonal information affecting users’ choice of rural tourism products mainly includes income status, age range, address t − T min f(t) � exp − k 􏼠 􏼡, T ≤ t ≤ T 􏼁, (1) min max of residence, user’s gender, nature of work, knowledge T − T max min composition, and education level received. where T is minimum forgetting interval and represents min the forgetting process, i.e., the buffer period, and the value is 3.3. User Model Construction. Rural tourism websites cur- the difference between the interest Ts and the interest ref- rently on the market have functions such as searching and erence time; T ismaximumforgettinginterval, indicating max recommending [21], but the following problems still exist: the decay cycle, i.e., the time required for the interest to decrease to its original value; t is user access interval; Tis the (1) +e accuracy rate of search and recommendation is timeofthelastvisitonthecurrentdate(indays); kisinterest not high. decay rate, where the value of k is proportional to the decay (2) +e correlation between recommended attractions is rate, here defined as 1 (the specific value can be adjusted not great. +rough analysis, it is found that the root according to the actual needs of the user). cause of the problems is the lack of modelling of user Equation (1) considers the case of a forgotten decay information. cycle, i.e., (T ≤ t ≤ T ), when t > T or t < T is not min max max min In this study, the user model is constructed by using the involved, as defined in this study. relevant information generated by the platform users in the When t < T , f() �1, it means that the interest is not min process of browsing and purchasing products to decreasing. 4 Discrete Dynamics in Nature and Society Keywords based Probabilistic Implicit Explicit inward agent model feedback feedback space model User User interest User interest information information User interest acquisition information representation User interest model learning User Modeling User interest representation of user interest modeling interest information model User interest model update Figure 1: Selected photographic records of participants. When t > T , �0, it means that the user has lost the essentially evaluates the preference information of most max interest and is removed from the user model. users and then makes a recommendation for a particular +eexponentialforgettingfunctionallowsusers’interest user. Once the set of users’ rating vectors is constructed, the weights to decay according to the length of the time interval. target user’s rating vectors from other rows are used to +is forgetting function takes into account the laws of predict the target user’s rating of the product (filling in the human memory and treats interest as a special kind of missing values of the target user’s rating in one row). A memory. tensor is amultilinear vector space, where first-ordertensors +e dynamic user interest weighting formula that can be considered as vectors and second-order tensors can combines thefrequency of visits tothe dynamicuser interest berepresentedbymatrices.Tensorslargerthansecondorder model can consider not only the one factor of time, but also are uniformly called higher order tensors. +e rank of a the frequency of visits to the interest. In this study, the tensor is defined as if a tensor can be expressed as an outer dynamic user interest model [24] is constructed by com- product of N vectors, and then the rank of the tensor is said bining the time forgetting factor and the frequency of access to be N. +e rank of a tensor means a tensor of rank one, to the interest, as shown in (1) which can be expressed by the outer product of vectors [25]. Figure 2 shows a third-order tensor, whose outer product is 1, ⎧ ⎪ of the form x .x .x . ⎪ 1 2 3 +e cp decomposition of a tensor is based on the basic concept of a rank one tensor, which is a decomposition of a v t, N � (2) 􏼁 exp􏼒− 􏼓∗ fi(t), i i ⎪ multiorder tensor into the form of an outer product of Ni ⎪ multiple rank one tensors. For example, 0, r r r X � 􏽘 x , x , . . . , x , (3) 1 2 N where N indicates the frequency of the user’s visit to the ith i r�1 interest point; v (t, N ) indicates the weight of the user’s i i where K denotes the number of rank one quantities. interest in the ith interest point at time t. +e initial value of In the description of this paper, we need to consider the user’s interest in all interests is 0, the weight of the user’s predicting thescoringdata inthetarget data sourcefrom the interest in “parent-child” recreation at the current moment feature vectors of the secondary data source, and here we is V , and according to the time cycle of rural tourism, the enhance the correlation between multiple data sources with minimum forgetting interval is 30 days and the maximum the help of tensor decomposition techniques between the forgetting interval is one year, i.e., 365 days. target and secondary data sources. 4. Tensor Decomposition-Based Fusion 4.2. Basic Ideas of the Algorithm. It is assumed that the user Model for Multiple Data Sources has had relevant Internet operations on multiple data 4.1. Tensor Models. +e reason for applying the idea of sources and has generated a rating matrix in the corre- tensor decomposition to the collaborative filtering recom- sponding data source. +e set of data sources is mendation algorithm is that collaborative filtering R � 􏼈r , r . . . r . . .􏼉, and the user’s rating matrix is i 1 2 i Discrete Dynamics in Nature and Society 5 Figure 2: +ird-order tensor. ′ 1 λ λ λ represented by the set q, where X � 􏽮x , . . . , x 􏽯 rep- 2 M 2 T 2 B 2 1 1p t min(M, T, B) � ‖R − [[M, T, B]]‖ + ‖M‖ + ‖T‖ + ‖B‖ , F F F 2 2 2 2 resents the user’s rating of item Y � y , . . . , y . 􏽮 􏽯 t D D (4) Here, X, Y denote the corresponding user ratings of items in different data sources, which may be in the same where [[M, T, B]] � 􏽐 M T B , a regularization ° ° r�1 D D D r r r data source or in different data sources where the user term is added to prevent overfitting of the training results, λ ratings of items overlap, e.g., the user ratings of items in the represents the regularization parameter, and the model is data source and in the data source at the same time in the trained using stochastic gradient descent. corresponding data source, and there is also the possibility To ensure that the local optimum result can be achieved that the user ratings of items in each data source are un- in finite time, it needs to be shown that partial derivatives related. +e data sources are also independent of each other existfortheCdatasourcesintheirrespectivedirections.+e with no overlap [26]. +e data sources are considered to be proof procedure is as follows: the target data sources, and other relevant data sources are considered to be the secondary data sources. zf (5) � 􏼐X − [[M, T, B]] 􏼑(BΘT) + λ M. (1) (1) M Assuming that there is partial overlap and interpopu- zM lation of users between data sources, the corresponding Similarly it can be shown that the partial derivatives of ratings made by users in different domains are abstracted and c in the respective directions are into n-order tensors according to the idea of tensor, and the tensorsinthisexampleallbelongtorankonetensor,andthe zf � 􏼐X − [[M, T, B]] 􏼑(MΘB) + λ T, (2) (2) T vector model diagram is shown in Figure 3. zT (6) zf � 􏼐X − [[M, T, B]] 􏼑(MΘT) + λ B. (3) (3) B 4.3. Model Building. +e multisource tourism information zB fusion model proposed in this paper is based on tensor decomposition, and the user’s rating matrix in the target data source is regarded as a reorganization of the approx- 4.4. Travel Recommendation Algorithm Design. Since the imation matrix of the secondary data source through the user’s scoring matrix in a multidata source environment idea of matrix complementation, and then the elements of may not be a regular tensor, it is not possible to use the the approximation matrix are used as the unobservable part tensor decomposition model directly, so here it is nec- of the target matrix for rating estimation [27]. essary to introduce an invertible transformation to ensure +e main objective of the multidata source fusion travel that the ones in different domains can be transformed into recommendation algorithm is to construct a global rating the same dimensional information matrix; i.e., the matrix matrix model, where the matrix fuses the global rating product of the [M, T, B] score components is expressed in matrices of users in different data sources. +e matrix fusion the form of process is shown in Figure 4. Fromtheabovemodel,weabstractthethreedatasources Y � M 􏽘 B + E , K k (7) into a third-order tensor model, assuming a global data domain of R, the third-order tensor is described as R � [[M, T, B]] � 􏽐 M ∘ T ∘ B , where M, T, B where A,B denote the tensor matrix vectors in the auxiliary r�1 D D D r r r correspond to the user ratings in the three data sources of data sources. 􏽐 � diag(C ) denotes the diagonal matrix of k k movies, travel, and data, respectively. According to the R × R of the target data sources. E denotes the residual tensor decomposition model, the fusion of multiple data terms of the model training. +e global scoring matrix for sources can be normalised into a minimisation solution multiple data sources can be obtained by minimizing the problem. +e solution formula is as follows: objective function: 6 Discrete Dynamics in Nature and Society Book Movie Join Filed User Traval User Domain Figure 3: Data fusion model. 4 3 4 2 3 1 3 4 1 2 2 2 3 2 1 2 3 2 2 1 2 1 4 3 3 4 1 1 3 4 4 3 3 2 1 4 3 Figure 4: Matrix fusion process. � � � �2 � � 1� � λ λ λ T 2 2 2 � � M T B � � (8) U ≈ min(M, T, B) � Y − M 􏽘 B + ‖M‖ + ‖T‖ + ‖B‖ . � K � F F F � � 2 2 2 2 � � Here it is assumed that X � 􏼈X , X , . . . , X 􏼉 is a Guangzhou, Guilin, Hangzhou, and Haikou) were crawled D 1 2 K scoring matrix for K different domains, where X has N × from a domestic travel website using the dynamic agent M dimensions, N denotes the number of users under the technique of the PythonScrapy crawler framework, which current dimension, and M denotes the number of items in contains 8720 users and 23,305 rating records of the at- the K th data source. tractions by users; each user can make a range of ratings for +e vectors under different data sources can be itera- the attractions, along with 1,500 travel tips, corresponding tively trained and fused under the same data source. With types of attractions, the number of visitors to the attractions the help of a unified data source model, the similarity of the under different seasons, etc. +e information is mined to target data source domain is calculated and the nearest analyse the relevant characteristics, including gender, age, neighbours are selected based on this to obtain global occupation, travel time, as well as the route, type, and rating recommendation results [9]. of the attraction, and stored in the database for subsequent When the global user rating matrix is obtained the target analysis. user can be selected to calculate the nearest neighbours for +e types of attributes and the corresponding number of predictedratingsandobtainthefinalsetofrecommendation ratings for the attractions are shown in Table 1. results. +e rating prediction formula is as follows: +euserratingsforthescenicspotsareshowninTable2. 􏽐 sim(u, u) ∗ R − R 􏼁 u∈U(u) u u 􏼌 􏼌 P � R + , (9) 5.2. Experimental Results and Discussion. +e experimental u,i 􏼌 􏼌 􏼌 􏼌 􏽐 sim u, u 􏼁 􏼌 􏼌 u∈U(u) dataset was divided into a training set and a test set in the ratio of 8:2, and the results were validated by evaluating the where U(u) represents the global data rating matrix, data from the training set and using the data from the test sim(u, u)representsthesimilarityofusersontheglobaldata set. +e test dataset contains over 17 attraction types (his- domain, and R represents the average user rating. torical, scenic, natural, human, etc.) [11]. 5. Analysis of Experimental Results 5.2.1. Effect of Weighting Parameters. +e rating similarity 5.1. Experimental Data Sets. In order to obtain sufficient and attribute similarity of attractions are combined through experimental data to verify the feasibility of the algorithm, (9) to construct a global similarity of attractions, and the the experimental dataset was processed as follows: firstly, proportion in which these parameters are adapted is a fo- 1039 attractions in six Chinese cities (Beijing, Shanghai, cused point of investigation for this section of the 1 5 2 1 3 4 4 3 1 2 1 Discrete Dynamics in Nature and Society 7 Table 1: Types of attractions. 0.60 POI_ID POI_NAME Rating Type 1 +e Great Wall 456 History, scenery 0.55 2 Palace Museum 432 History, humanity 3 XiHu 112 Scenery, park 4 Nanshan Temple 321 Buddhism 0.50 5 Tiananmen Square 222 Politics, park 6 Fenjiezhou Island 532 Landscape, park 0.45 Table 2: User-attraction rating scale. 0.40 User_ID POI_ID Rating Times stamp 189 212 3 865454 231 331 2 823213 0.0 0.2 0.4 0.6 0.8 1.0 123 213 4 872443 Alpha 412 23 3 865231 k = 20 k = 80 11 321 5 865454 k = 40 k = 100 6 221 3 843233 k = 60 Figure 5: Effect of weighting parameters. experiment. From (9), μ + μ � 1, to ensure a single con- a p trollableexperimentalvariable;hereweproposeahypothesis of α � μ ,andthen μ � 1 − α,sothatasinglevariablecanbe a p controlled to observe the effect of the weighting parameters 0.75 on the experimental results. In this experiment, the uniform parameter α was taken in the range [0, 1], and the perfor- 0.74 mance of the algorithm was observed by adjusting the different neighbourhood users selected. As can be seen from Figure 5, the horizontal coordinate 0.73 representstherangeof theparameter values,andthevertical coordinate represents the MAE values, which change dif- ferently between different numbers of neighbours k as the 0.72 parameter goes from 0 to 1. +e MAE of the algorithm is optimal when the weight parameter a �0.6 and the number 0.71 of neighbours k �60. +e MAE value decreases at the be- ginning as the weight parameter a increases and starts to 05 10 20 30 400 60 70 increase when it exceeds α �0.6. +is is mainly because the Number of neighbor algorithm gradually ignores the evaluation of attraction attributes when the weight parameter exceeds 0.6. In order alpha = 0 to find the optimal set of neighbours for the target item, we alpha = 0.6 alpha = 1 set the range of neighbours for the target item to 60 and use three different weight control parameters a �0.061 to ob- Figure 6: Effect of number of residences. serve the influence of different number of neighbours on the recommendation result, which was observed by using three number of unrated scenic items. +e experimental results different weight control parameters; a �0.061. are shown in Figure 7. As shown in Figure 6, the overall performance of the algorithm’s ME value is low after fusing the optimal weight From Figure 7, it can be seen that the MAE values of the algorithms in this paper are relatively low when crossing the parameters, and the MAE of the algorithm gradually de- other two algorithms. Traditional algorithms IBCF and creases as the number of project neighbours increases and ITEM-CFhavedifficultyinobtainingthenearestneighbours gradually stabilizes when the number of neighbours exceeds of the target items due to the lack of basis for calculating the 50. +e experimental results show that when the number of similarity matrix due to the absence of a large amount of neighbours is chosen around 40, the algorithm can achieve rating data. In this paper, the algorithm uses a combination the optimal recommendation result. Comparing different of attraction scores and project attributes to calculate the algorithms, this section compares the algorithms in this global similarity in the absence of project scores, combined paper (RACF with the traditional user-based collaborative with the inherent project attributes to assist in the calcu- filtering algorithm (IBCF) and the improved item-based lation of global similarity, which alleviates the problem of collaborative filtering algorithm IITEM-CF). 800 users were selected in the experimental dataset, which contained a large data sparsity to a certain extent. MAE MAE 8 Discrete Dynamics in Nature and Society 0.95 4000 0.90 3500 0.85 0.80 0.75 0.70 0.65 0.60 0 20 40 60 80 movie travel book IRACF Figure 8: Distribution of test data statistics. IBCF ITEM-CF Figure 7: Comparison results of the same algorithm. (5) Without distinguishing the data sources, the global data is considered, and the movie data source, book data source, and tourism data source are regarded as 5.3. Real Life Examples. Based on the tourism data of the overall target data domain, and the similarity is domestic Ctrip, we used the dynamic agent technology calculated for the target user in the target data do- of Python scratch crawler framework to capture 1039 main and the nearest neighbours are selected, and scenic spots in 6 cities in China (Beijing, Shanghai, themissingratingitemsofthetargetuserarefilledto Guangzhou, Guilin, Hangzhou, and Haikou) from a make the user’s rating of the attractions in the domestic tourism website. It contains 8720 users and tourism data source. 23305 scoring records of scenic spots. Each user can store the scenic spots within the range of [1, 5]. At the +e data set division in the experiments is uniformly same time, there are 1500 tourism strategies, corre- constructed in the form of proportional division; i.e., the sponding scenic spot types, number of scenic spots ratio of the training set to the test set is 8:2, and the al- visited in different seasons, etc. gorithmmodelistrainedbasedonthedatainthetrainingset In total, the experiments in this section involve three to predict the users’ scores for the unrated items in the test data sources: the movie data sources, book data sources, and set. Firstly, as shown in the figure, we conducted a statistical travel data sources, using movies and books as secondary analysis of the scores in the secondary data sources to verify data sources to make predictions on the target data source, the correlation between the data sources, and the experi- the travel domain. +e experiments in this paper are divided mental results are shown in Figure 8. into the following tasks. Figure 8 shows the number of users who have rated books in the different data sources. Since the number of (1) +e movie data source and the book data source, users selected is fixed (5000), we can observe that there is a respectively, are used as auxiliary data sources to certain coverage of users’ behaviour in the different data calculate the similarity between users, to predict sources. +e main reason for this phenomenon is that if a users who did not make a rating, to calculate the user has marked books on the topic of history and hu- nearest neighbours, to calculate the correlation co- manities several times in the book data source, he will also efficient using the modified cosine similarity, and pay more attention to movies related to history and hu- thus to predict the items that populate the user who manities in the movie field, and by sending similar users in did not make a rating. thesecondarydatasource,ithelpsthetargetdatasourcefield (2) Based on the similarity between users calculated to discover similar users in the target field faster. +is is also from the auxiliary data sources, predictions were a prerequisite for fusion modelling of multiple data sources. made to the target according to the source of the Figure 9 shows the performance of the algorithm in the items that were not rated. case of different data sources. (3) Fusion of the target data source and the auxiliary In Figure 9, the horizontal coordinate indicates the data source was carried out. +e algorithm in this number of neighbours of the selected target user, and the paper is used to score the target data source. vertical coordinate indicates the value of the system average (4) Without differentiating data sources, the global data absolute error MAE. From the above figure, it can be seen areconsidered, andthe movie datasource, book data that the algorithm’s prediction in the single data source source, and travel data source are considered as the environment is difficult to find suitable neighbours in ef- overall target data domain. fective time due to the sparsity of the data, resulting in low MAE Discrete Dynamics in Nature and Society 9 the traditional culture of unique ethnic minorities in 0.94 Yunnan (no. 20BMZ164). 0.93 References 0.92 [1] M. Wang, “Research on convergence of rural tourism and related industries based on the perspective of holistic tourism: practice analysis from China,” Ekoloji, vol. 28, no. 107, 0.91 pp. 4295–4302, 2019. [2] P. Yochum, L. Chang, T. Gu, and M. Zhu, “Linked open data 0.90 in location-based recommendation system on tourism do- main: a survey,” IEEE Access, vol. 8, pp. 16409–16439, 2020. 0.89 [3] Y. Yi, C. Zhao, and J. Fu, “Research on the integration and development of modern agriculture and rural tourism based onAHP—takeYangjia towninMianyangcityasanexample,” 52 10 15 205 Open Journal of Social Sciences, vol. 7, no. 7, pp. 76–87, 2019. Neighbor [4] D. Wu, C. Zhang, L. Ji, R. Ran, H. Wu, and Y. Xu, “Forest fire travel book recognition based on feature extraction from multi-view travel movie images,” Traitement du Signal, vol. 38, no. 3, pp. 775–783, travel book movie [5] H. Zhang, Y. Duan, and Z. Han, “Research on spatial patterns Figure9:Performanceofthealgorithminthispaperwiththesame and sustainable development of rural tourism destinations in data source. the yellow river basin of China,” Land, vol. 10, no. 8, p. 849, [6] J.ZhangandL.Dong,“Imagemonitoringandmanagementof recommendation accuracy. As new data sources are added, the hot tourism destination based on data mining technology in algorithm’s MAE value starts to decrease compared to the big data environment,” Microprocessors and Microsystems, single data source environment, and the algorithm’s MAE vol. 80, Article ID 103515, 2021. value is lowest when fusing global data sources and reaches the [7] G. Alexandridis, A. Chrysanthi, G. E. Tsekouras, and optimal result when the number of neighbours is 25. G. Caridakis, “Personalized and content adaptive cultural heritage path recommendation: an application to the Gournia 6. Conclusions and Çatalhoy ¨ uk ¨ archaeological sites,” User Modeling and User-Adapted Interaction, vol. 29, no. 1, pp. 201–238, 2019. +is paper proposes a relatively novel travel recommendation [8] Q. Li, S. Li, S. Zhang, J. Hu, and J. Hu, “A review of text algorithm with fusion of multiple data sources. +e algorithm corpus-based tourism big data mining,” Applied Sciences, adopts the idea of tensor orthogonal decomposition and fuses vol. 9, no. 16, p. 3300, 2019. multipledatamodelstopredicttheratingofthetargetdomain. [9] F. Randelli and F. Martellozzo, “Is rural tourism-induced built-up growth a threat for the sustainability of rural areas? +eintegratedconsiderationofmultipledatasourcesunderthe +e case study of Tuscany,” Land Use Policy, vol. 86, use of the target domain is to assist the target domain to pp. 387–398, 2019. discoverthetargetuserneighbourhoodusersmorequicklyand [10] H. Li, D. Zeng, L. Chen, Q. Chen, M. Wang, and C. Zhang, more accurately to discover the user’s interest degree. It is “Immune multipath reliable transmission with fault tolerance worth pointing out that the recommendation algorithm pro- in wireless sensor networks,” in Proceedings of the Interna- posed in this paper under the fusion of multiple data sources is tional Conference on Bio-Inspired Computing: ?eories and not necessarily applicable to data sources with weak correla- Applications, pp. 513–517, Springer, Singapore, 2016 October. tion. +e algorithm has the lowest MAE value when fusing [11] I. Ibrahim, K. Zakariya, and N. A. Wahab, “Satellite image global data sources and reaches the optimal result when the analysis along the kuala selangor to sabak bernam rural number of neighbours is 25. tourism routes,” in Proceedings of the IOP Conference Series: Earth and Environmental Science, vol. 117, no. 1, Article ID 012013, United Kingdom, 2018. Data Availability [12] G. Bambi, S. Iacobelli, G. Rossi, P. Pellegrini, and M. Barbari, “Rural tourism to promote territories along the ancient roads +e dataset used in this paper is available from the corre- of communication: case study of the rediscovery of the st. sponding author upon request. Francis’s ways between fv,” European Countryside, vol. 11, no. 3, pp. 462–474, 2019. Conflicts of Interest [13] R. Beridze, I. Kordzaia, M. Diasamidze, and N. Beridze, “Sustainable rural tourism development recovery from covid- +e authors declare that they have no conflicts of interest 19 (in adjara region),” Globalization and Business, vol. 10, regarding this work. pp. 287–294, 2020. [14] S. Gorm ¨ u¨¸,s E. Atmi¸,s H. B. Gun ¨ ¸en, s N. K. Ozkazanç, and Acknowledgments M. Artar, “+e importance of mapping natural and cultural routes in rural tourism: bartın case,” International Journal of +is work was supported by the National Social Science Agricultural and Natural Sciences, vol. 10, no. 1, pp. 32–38, Fund Item, Research on Ecological Aesthetics in line with 2017. MAE 10 Discrete Dynamics in Nature and Society [15] I. Triˇsic,´ S. Stetic,´ D. Privitera, and A. Nedelcu, “Wine routes in Vojvodina Province, Northern Serbia: a tool for sustainable tourism development,” Sustainability, vol. 12, no. 1, p. 82, [16] T. Wijijayanti, Y. Agustina, A. Winarno, L. Istanti, and B. Dharma, “Rural tourism: a local economic development,” Australasian Accounting, Business and Finance Journal, vol. 14, no. 1, pp. 5–13, 2020. [17] C. J. A. Mitchell and M. Shannon, “Establishing the routes to ruralin-migrantproprietorshipinaCanadiantourismregion: a mobilities perspective,” Population, Space and Place, vol. 24, no. 3, p. e2095, 2018. [18] L.Wang,C.Zhang,Q.Chenetal.,“Acommunicationstrategy of proactive nodes based on loop theorem in wireless sensor networks,” in Proceedings of the 2018 Ninth International Conference on Intelligent Control and Information Processing (ICICIP), pp. 160–167, IEEE, Wanzhou, China, 2018, November. [19] M. A. Uwaisy, Z. K. A. Baizal, and M. Y. Reditya, “Recom- mendation of scheduling tourism routes using tabu search method (case study bandung),” Procedia Computer Science, vol. 157, pp. 150–159, 2019. [20] S. Li, S. Takahashi, K. Yamada, M. Takagi, and J. Sasaki, “Analysis of SNS photo data taken by foreign tourists to Japan and a proposed adaptive tourism recommendation system,” in Proceedings of the 2017 International Conference on Progress in Informatics and Computing (PIC), pp. 323–327, IEEE, Nanjing, China, 2017, December. [21] W. Zheng, Z. Liao, and Z. Lin, “Navigating through the complex transport system: a heuristic approach for city tourism recommendation,” Tourism Management, vol. 81, Article ID 104162, 2020. [22] L.Zhang,“Intelligenttourismrouterecommendationmethod based on big data,” International Journal of Autonomous and Adaptive Communications Systems,vol.13,no.4,pp.329–341, [23] J. Xu and J. Xu, “Research on the design of online travel service recommendation system based on data analysis,” in Proceedings of the International Conference on Cognitive Based Information Processing and Applications (CIPA 2021), pp. 1042–1046, Springer, Huainan, China, 2022 August. [24] X. Zheng, Y. Luo, L. Sun, Q. Yu, J. Zhang, and S. Chen, “A novel multi-objective and multi-constraint route recom- mendation method based on crowd sensing,” Applied Sci- ences, vol. 11, no. 21, p. 10497, 2021. [25] L. Feng, “Design of tourism intelligent recommendation modelof mounttaiscenicareabased on knowledgegraph,”in Proceedings of the 2020 International Conference on E-Com- merce and Internet Technology (ECIT), pp. 241–244, IEEE, Zhangjiajie, China, 2020, April. [26] C. Cao, Y. Tang, D. Huang, W. Gan, and C. Zhang, “IIBE: an improved identity-based encryption algorithm for WSN Se- curity,” Security and Communication Networks, vol. 2021, Article ID 8527068, 8 pages, 2021. [27] C. Yuan and M. Uehara, “An optimal route recommendation method for a multi-purpose travel route recommendation system,” in Proceedings of the International Conference on Broadband and Wireless Computing, Communication and Applications, pp. 398–408, Springer, Fukuoka, Japan, Octomber 2019. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Discrete Dynamics in Nature and Society Hindawi Publishing Corporation

Research on the Recommendation Algorithm of Rural Tourism Routes Based on the Fusion Model of Multiple Data Sources

Loading next page...
 
/lp/hindawi-publishing-corporation/research-on-the-recommendation-algorithm-of-rural-tourism-routes-based-tbvwaR3i1f

References (30)

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2022 Hong Li et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1026-0226
eISSN
1607-887X
DOI
10.1155/2022/2262148
Publisher site
See Article on Publisher Site

Abstract

Hindawi Discrete Dynamics in Nature and Society Volume 2022, Article ID 2262148, 10 pages https://doi.org/10.1155/2022/2262148 Research Article Research on the Recommendation Algorithm of Rural Tourism Routes Based on the Fusion Model of Multiple Data Sources 1 1 2 Hong Li, Man Qiao, and Shuai Peng Henan University of Animal Husbandry & Economy, Zhengzhou 450046, Henan, China Guangdong University of Finance and Economics, Guangzhou 510320, Guangdong, China Correspondence should be addressed to Shuai Peng; 20191021@gdufe.edu.cn Received 23 November 2021; Accepted 21 March 2022; Published 14 April 2022 Academic Editor: Wei Zhang Copyright©2022HongLietal.+isisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Rural tourism has become an important force in implementing the rural revitalisation strategy and accelerating rural economic development. +e hectic pace of life has made more and more city dwellers yearn for rural life, and travelling in the countryside has become their weekend choice. However, the current level of rural tourism informationization is low, the publicity is in- sufficient, the tourists’ awareness is low, and the source of customers is seriously insufficient. To this end, this paper proposes a relatively novel multidata source fusion tourism recommendation algorithm, which adopts the idea of tensor orthogonal de- composition and fuses multisource data models to predict the target domain’s for rating. +e integrated consideration of multiple data sources under the do-it-yourself approach assists the target domain to discover the target user neighbourhood users more quickly and to discover the user’s interest degree more accurately. It is worth pointing out that the recommendation algorithm proposedinthispaperunderthefusionofmultipledatasourcesisnotnecessarilyapplicabletodatasourceswithweakcorrelation, such as travel data sources and music data sources, which are relatively weakly correlated, and the algorithm is slightly weak in making predictions of user preferences. operation [3]. However, in the traditional matrix decom- 1. Introduction position process, the data structure information is often lost With the national economy entering a new normal, tourism in the matrix decomposition process due to data sparsity, hasusheredinagoldenperiodofrapiddevelopment[1],and making the results distorted. Tensor is a way of storing rural tourism has become an important part of China’s multidimensional data, and the concept of tensor decom- tourism industry. +e busy pace of life has made more and position is based on the idea of matrix completion, which more city dwellers yearn for rural life, and travelling in the aims to fill in the missing (or unobservable) parts of the countryside has become their weekend choice. However, at target matrix [4]. In simple terms, this means that matrix A present, the level of information technology for rural is used to approximately evaluate matrix B (there is some tourism is low, the publicity is not strong enough, tourists inherent correlation between A and B). +e invisible parts of the matrix B are filled in by the matrix A. are less aware of it, and the source of visitors is seriously insufficient. In this context, rural tourism urgently needs +is study, through the platform of rural tourism information technology to increase publicity and improve products on the malefactor research, found that the content service levels [2]. oftheseruraltourismproductsismainlyconcentratedinthe Inpracticalrecommendationsystems,themostcommon better economy and tourism industry of more developed method of fusing multiple data sources is matrix decom- areas, while the economy in relatively backward areas is not position. Decomposed user feature matrix U and item covered [5] or there is also incomplete information, not feature matrix V are obtained by training the loss function, enough to meet the rising demand of users. For example, the and finally the scoring matrix is reduced by matrix inverse interface of searching for any area on the Nongjiale platform 2 Discrete Dynamics in Nature and Society +rough the study of the above rural tourism wisdom simply shows a little of the same scenery and cuisine in- formation, there is no user search function on the rural products, we found that these rural tourism wisdom products are mainly concentrated in the more developed tourism service platform, and the product description is simple, which cannot hook the people’s desire to travel [6]. tourism industry and the better economic regions, while the +e development of rural tourism products has a very far- relatively more economically backward regions rural tour- reaching significance in promoting the economic develop- ism wisdom products either do not have complete infor- ment of rural areas, combining field research, literature mation or are not covered, and the scope of operation is analysis, and data from existing rural tourism platforms to narrow, the services provided are limited and not well construct a rural tourism product model and a user model. known, and their service content cannot meet the needs of +e aim is to improve the quality and efficiency of users’ users [11]. For example, the interface of searching for any access to useful information, so that rural tourism products area on the WeChat platform of Nongjiale is just to show a littlesceneryanddishinformation,thereisnousersearchon are more accurately submitted to users. Combining users’ personalised characteristics to recommend the rural tourism the WeChat miniprogram of rural tourism service platform, and the product introduction is simple, which cannot hook products they really need to have has become a valuable and challenging research topic. the people’s desire to travel. +rough literature research, the academic world has also seen moreresearchonintelligentruraltourism;forexample, 2. Related Work [12] published in the China Tourism News about the new While China’s economy is growing rapidly and people’s mission given to tourism in the new era, [13] conducted an living standards aresteadily rising, the tourismindustry, as a in-depth analysis of the problems in the development of sunrise industry, is receiving increasing attention from the rural tourism information technology and believes that the government and enterprises [7], and tourism has become an factors that inhibit the development of rural tourism in- formation technology are publicity, information manage- importantdrivertostimulateconsumptionandinducerapid upgrading and transformation of the industry. Just as ment level, information infrastructure construction, etc., [14] argues that there is a certain difference between the China’s tourism industry is moving towards mass tourism and global tourism, the development of information tech- demand and supply of tourists in rural tourism and that nology and mobile Internet applications has given rise to a measures should be provided to reduce the difference be- new concept of smart tourism for the tourism experience. tween the two, [15] designed an ecological service system for +e combination of tourism and information technology rural tourism from three aspects: service design process, constitutes smart tourism, which is a necessary path for the interactive experience, and branding, and [16] argues that, current development of tourism [8]. with the help of mobile Internet, a seamless connection Tourism websites, tourism APPs, and tourism WeChat between tourism enterprises and tourists can be achieved applets are current manifestations of smart tourism, and and a bridge for efficient communication between them can be built. with the rapid development of smart phones and mobile networks, many smart tourism products have emerged, such Although smart tourism is a unique concept in China, similar projects with smart tourism have emerged inter- as Ctrip, founded in 1999, whose mobile APP was launched in 2010, and its acquisition of the UK-based airfare search nationally earlier than in China, with USA, Singapore, platform Skyscanner Limited was in November 2016, which Korea, England, and other countries being among the more means that Ctrip has started to enter the road of inter- representative ones. +e smart wrist system online in the nationalisation [3, 9]. Where to Go was established in 2005, United States in 2005 opened the beginning of smart and in 2010 it launched its APP, the same as Ctrip; it is a tourism; the feedback system equipped with radio frequency comprehensive travel APP that provides a collection of technology positioning device one by one Mountainwatch wasthefirsttobeusedbytheColoradoSteamboatSkiResort business travel management, hotels, airline tickets, holiday booking, and travel information. In October 2015, the two in the United States, which can provide tourists with real- time consumption and ski routes. Touchwood, a service travelgiantAPPs announcedamerger. In2006, theMaHive travel website was launched and became popular with users, platformin Seoul,South Korea, is aimed at self-guided,rural travellers, who can use the platform to perceive tourist and its mobile APP was launched in 2011. Just as smart tourism was being widely promoted and applied, China’s information [17, 18]. rural tourism industry also stepped into a path of rapid development [10]. Previous rural tourism products can no 3. Construction of User Ontology Model for longer meet the tourism needs of the people, and there is an Rural Tourism Platform urgent need to introduce intelligent rural tourism into rural tourism. +at is why MPPs such as Meiju Countryside, Find Personalised recommendation for users is the ultimate goal the Yard, Go Farm, He Xiang You, Meet the Countryside, of an intelligent recommendation system, and the con- and Down to the Country Guest have emerged, and cor- struction of a good user ontology model is a prerequisite for responding WeChat platforms such as Countryside Tourism implementing intelligent recommendations. +e user on- Service Platform, Countryside Tourism Merchant, Nongjia tology model in this study requires the acquisition of data Platform, and Shanghai Nongjia Platform have appeared on relatedtotheuser’sinterestsinthefieldofruraltourism,and WeChat mini-programs. then combining the user’s personal information to Discrete Dynamics in Nature and Society 3 automatically build their interest models. +e user model is eventually build a model that can be recognised by our computer. +e construction process of the user personalised an abstract representation of the personal information and interest information of the system users. +e basic personal interest model [19] is shown in Figure 1. information in the user model is relatively stable, while the user’s interest information fluctuates greatly with time and 3.1. Countryside Tourism Platform User Information. +e product attributes, so how to accurately express the user’s format and quality of the acquired data directly affect the interest and facilitate the calculation becomes the key to the quality of the user model. Currently, there are two common implementation of the user model. techniques used to obtain user interest information, Considering the probabilistic topic representation, the specifically: core idea is to understand each text message as a mixture of multiple topical features,where each topic is thedistribution (1) Displayfeedback technology: the main wayto record probability of the corresponding feature [22]. user preferences is through the user’s evaluation of In the rural tourism platform study, the rural tourism the corresponding product. +is requires the user to product ontology is used to represent the user’s interests, activelyparticipateandactivelyevaluate theproduct, which is relatively simple in structure and can be imple- which takes up more of the user’s time and does not mented by the keyword vector space representation method. allow for good access to the user’s personal interest +e user model consists of the user’s interest set, the user’s information when the user’s participation is not interest attribute set, the user’s attention and weight for each high. interestintheinterestattributevalues,and3parts,whichare (2) Implicit feedback technology: it is mainly through the user’s interest set, i.e., the product’s key attribute set; the the system to view and analyse the user’s behavioural user’s interest attribute set, i.e., the product attribute value data to obtain personalised information about the set. user’s interests. Implicit feedback technology does not require active participation of the user, mainly throughtheplatform’sback-endsystemtoobtainthe 3.4. Dynamic User Interest Model. +e user model contains corresponding information of the user, such as the two types of information: static, such as basic information number of visits to a product, the length of stay of about the user, such as gender, occupation, and knowledge each visit (because of the number of visits to background; and dynamic, which changes over time. In this products of interest to the user, the length of stay study, we mainly consider the part of dynamic updates. may be longer), the number of searches, etc. Users’ interests change with the environment and psycho- logical factors, so the common methods used to update dynamicusermodelsaretheforgettingfunctionmethodand 3.2. Personal Information of Rural Tourism Platform Users. the time window method. +e forgetting function method is Usually, the user’s basic personal situation will make the similar to the law of memory forgetting in that, without user’s interests relatively stable [20]; for example, users with external stimuli, a user’s interest in something will decay young children have a greater interest in parent-child rural over time [23]. In this study, the user interest model is tourism tours, and low-income people will generally choose considered to change with the combination of time for- rural tourism products with lower or free costs. getting and frequency of access. User interest forgetting In this paper, the authors conclude from the corre- function is used to dynamically update the user interest sponding literature, research on existing user information model based on the forgetting factor, which is calculated on rural tourism platforms, and field surveys that the per- using sonal information affecting users’ choice of rural tourism products mainly includes income status, age range, address t − T min f(t) � exp − k 􏼠 􏼡, T ≤ t ≤ T 􏼁, (1) min max of residence, user’s gender, nature of work, knowledge T − T max min composition, and education level received. where T is minimum forgetting interval and represents min the forgetting process, i.e., the buffer period, and the value is 3.3. User Model Construction. Rural tourism websites cur- the difference between the interest Ts and the interest ref- rently on the market have functions such as searching and erence time; T ismaximumforgettinginterval, indicating max recommending [21], but the following problems still exist: the decay cycle, i.e., the time required for the interest to decrease to its original value; t is user access interval; Tis the (1) +e accuracy rate of search and recommendation is timeofthelastvisitonthecurrentdate(indays); kisinterest not high. decay rate, where the value of k is proportional to the decay (2) +e correlation between recommended attractions is rate, here defined as 1 (the specific value can be adjusted not great. +rough analysis, it is found that the root according to the actual needs of the user). cause of the problems is the lack of modelling of user Equation (1) considers the case of a forgotten decay information. cycle, i.e., (T ≤ t ≤ T ), when t > T or t < T is not min max max min In this study, the user model is constructed by using the involved, as defined in this study. relevant information generated by the platform users in the When t < T , f() �1, it means that the interest is not min process of browsing and purchasing products to decreasing. 4 Discrete Dynamics in Nature and Society Keywords based Probabilistic Implicit Explicit inward agent model feedback feedback space model User User interest User interest information information User interest acquisition information representation User interest model learning User Modeling User interest representation of user interest modeling interest information model User interest model update Figure 1: Selected photographic records of participants. When t > T , �0, it means that the user has lost the essentially evaluates the preference information of most max interest and is removed from the user model. users and then makes a recommendation for a particular +eexponentialforgettingfunctionallowsusers’interest user. Once the set of users’ rating vectors is constructed, the weights to decay according to the length of the time interval. target user’s rating vectors from other rows are used to +is forgetting function takes into account the laws of predict the target user’s rating of the product (filling in the human memory and treats interest as a special kind of missing values of the target user’s rating in one row). A memory. tensor is amultilinear vector space, where first-ordertensors +e dynamic user interest weighting formula that can be considered as vectors and second-order tensors can combines thefrequency of visits tothe dynamicuser interest berepresentedbymatrices.Tensorslargerthansecondorder model can consider not only the one factor of time, but also are uniformly called higher order tensors. +e rank of a the frequency of visits to the interest. In this study, the tensor is defined as if a tensor can be expressed as an outer dynamic user interest model [24] is constructed by com- product of N vectors, and then the rank of the tensor is said bining the time forgetting factor and the frequency of access to be N. +e rank of a tensor means a tensor of rank one, to the interest, as shown in (1) which can be expressed by the outer product of vectors [25]. Figure 2 shows a third-order tensor, whose outer product is 1, ⎧ ⎪ of the form x .x .x . ⎪ 1 2 3 +e cp decomposition of a tensor is based on the basic concept of a rank one tensor, which is a decomposition of a v t, N � (2) 􏼁 exp􏼒− 􏼓∗ fi(t), i i ⎪ multiorder tensor into the form of an outer product of Ni ⎪ multiple rank one tensors. For example, 0, r r r X � 􏽘 x , x , . . . , x , (3) 1 2 N where N indicates the frequency of the user’s visit to the ith i r�1 interest point; v (t, N ) indicates the weight of the user’s i i where K denotes the number of rank one quantities. interest in the ith interest point at time t. +e initial value of In the description of this paper, we need to consider the user’s interest in all interests is 0, the weight of the user’s predicting thescoringdata inthetarget data sourcefrom the interest in “parent-child” recreation at the current moment feature vectors of the secondary data source, and here we is V , and according to the time cycle of rural tourism, the enhance the correlation between multiple data sources with minimum forgetting interval is 30 days and the maximum the help of tensor decomposition techniques between the forgetting interval is one year, i.e., 365 days. target and secondary data sources. 4. Tensor Decomposition-Based Fusion 4.2. Basic Ideas of the Algorithm. It is assumed that the user Model for Multiple Data Sources has had relevant Internet operations on multiple data 4.1. Tensor Models. +e reason for applying the idea of sources and has generated a rating matrix in the corre- tensor decomposition to the collaborative filtering recom- sponding data source. +e set of data sources is mendation algorithm is that collaborative filtering R � 􏼈r , r . . . r . . .􏼉, and the user’s rating matrix is i 1 2 i Discrete Dynamics in Nature and Society 5 Figure 2: +ird-order tensor. ′ 1 λ λ λ represented by the set q, where X � 􏽮x , . . . , x 􏽯 rep- 2 M 2 T 2 B 2 1 1p t min(M, T, B) � ‖R − [[M, T, B]]‖ + ‖M‖ + ‖T‖ + ‖B‖ , F F F 2 2 2 2 resents the user’s rating of item Y � y , . . . , y . 􏽮 􏽯 t D D (4) Here, X, Y denote the corresponding user ratings of items in different data sources, which may be in the same where [[M, T, B]] � 􏽐 M T B , a regularization ° ° r�1 D D D r r r data source or in different data sources where the user term is added to prevent overfitting of the training results, λ ratings of items overlap, e.g., the user ratings of items in the represents the regularization parameter, and the model is data source and in the data source at the same time in the trained using stochastic gradient descent. corresponding data source, and there is also the possibility To ensure that the local optimum result can be achieved that the user ratings of items in each data source are un- in finite time, it needs to be shown that partial derivatives related. +e data sources are also independent of each other existfortheCdatasourcesintheirrespectivedirections.+e with no overlap [26]. +e data sources are considered to be proof procedure is as follows: the target data sources, and other relevant data sources are considered to be the secondary data sources. zf (5) � 􏼐X − [[M, T, B]] 􏼑(BΘT) + λ M. (1) (1) M Assuming that there is partial overlap and interpopu- zM lation of users between data sources, the corresponding Similarly it can be shown that the partial derivatives of ratings made by users in different domains are abstracted and c in the respective directions are into n-order tensors according to the idea of tensor, and the tensorsinthisexampleallbelongtorankonetensor,andthe zf � 􏼐X − [[M, T, B]] 􏼑(MΘB) + λ T, (2) (2) T vector model diagram is shown in Figure 3. zT (6) zf � 􏼐X − [[M, T, B]] 􏼑(MΘT) + λ B. (3) (3) B 4.3. Model Building. +e multisource tourism information zB fusion model proposed in this paper is based on tensor decomposition, and the user’s rating matrix in the target data source is regarded as a reorganization of the approx- 4.4. Travel Recommendation Algorithm Design. Since the imation matrix of the secondary data source through the user’s scoring matrix in a multidata source environment idea of matrix complementation, and then the elements of may not be a regular tensor, it is not possible to use the the approximation matrix are used as the unobservable part tensor decomposition model directly, so here it is nec- of the target matrix for rating estimation [27]. essary to introduce an invertible transformation to ensure +e main objective of the multidata source fusion travel that the ones in different domains can be transformed into recommendation algorithm is to construct a global rating the same dimensional information matrix; i.e., the matrix matrix model, where the matrix fuses the global rating product of the [M, T, B] score components is expressed in matrices of users in different data sources. +e matrix fusion the form of process is shown in Figure 4. Fromtheabovemodel,weabstractthethreedatasources Y � M 􏽘 B + E , K k (7) into a third-order tensor model, assuming a global data domain of R, the third-order tensor is described as R � [[M, T, B]] � 􏽐 M ∘ T ∘ B , where M, T, B where A,B denote the tensor matrix vectors in the auxiliary r�1 D D D r r r correspond to the user ratings in the three data sources of data sources. 􏽐 � diag(C ) denotes the diagonal matrix of k k movies, travel, and data, respectively. According to the R × R of the target data sources. E denotes the residual tensor decomposition model, the fusion of multiple data terms of the model training. +e global scoring matrix for sources can be normalised into a minimisation solution multiple data sources can be obtained by minimizing the problem. +e solution formula is as follows: objective function: 6 Discrete Dynamics in Nature and Society Book Movie Join Filed User Traval User Domain Figure 3: Data fusion model. 4 3 4 2 3 1 3 4 1 2 2 2 3 2 1 2 3 2 2 1 2 1 4 3 3 4 1 1 3 4 4 3 3 2 1 4 3 Figure 4: Matrix fusion process. � � � �2 � � 1� � λ λ λ T 2 2 2 � � M T B � � (8) U ≈ min(M, T, B) � Y − M 􏽘 B + ‖M‖ + ‖T‖ + ‖B‖ . � K � F F F � � 2 2 2 2 � � Here it is assumed that X � 􏼈X , X , . . . , X 􏼉 is a Guangzhou, Guilin, Hangzhou, and Haikou) were crawled D 1 2 K scoring matrix for K different domains, where X has N × from a domestic travel website using the dynamic agent M dimensions, N denotes the number of users under the technique of the PythonScrapy crawler framework, which current dimension, and M denotes the number of items in contains 8720 users and 23,305 rating records of the at- the K th data source. tractions by users; each user can make a range of ratings for +e vectors under different data sources can be itera- the attractions, along with 1,500 travel tips, corresponding tively trained and fused under the same data source. With types of attractions, the number of visitors to the attractions the help of a unified data source model, the similarity of the under different seasons, etc. +e information is mined to target data source domain is calculated and the nearest analyse the relevant characteristics, including gender, age, neighbours are selected based on this to obtain global occupation, travel time, as well as the route, type, and rating recommendation results [9]. of the attraction, and stored in the database for subsequent When the global user rating matrix is obtained the target analysis. user can be selected to calculate the nearest neighbours for +e types of attributes and the corresponding number of predictedratingsandobtainthefinalsetofrecommendation ratings for the attractions are shown in Table 1. results. +e rating prediction formula is as follows: +euserratingsforthescenicspotsareshowninTable2. 􏽐 sim(u, u) ∗ R − R 􏼁 u∈U(u) u u 􏼌 􏼌 P � R + , (9) 5.2. Experimental Results and Discussion. +e experimental u,i 􏼌 􏼌 􏼌 􏼌 􏽐 sim u, u 􏼁 􏼌 􏼌 u∈U(u) dataset was divided into a training set and a test set in the ratio of 8:2, and the results were validated by evaluating the where U(u) represents the global data rating matrix, data from the training set and using the data from the test sim(u, u)representsthesimilarityofusersontheglobaldata set. +e test dataset contains over 17 attraction types (his- domain, and R represents the average user rating. torical, scenic, natural, human, etc.) [11]. 5. Analysis of Experimental Results 5.2.1. Effect of Weighting Parameters. +e rating similarity 5.1. Experimental Data Sets. In order to obtain sufficient and attribute similarity of attractions are combined through experimental data to verify the feasibility of the algorithm, (9) to construct a global similarity of attractions, and the the experimental dataset was processed as follows: firstly, proportion in which these parameters are adapted is a fo- 1039 attractions in six Chinese cities (Beijing, Shanghai, cused point of investigation for this section of the 1 5 2 1 3 4 4 3 1 2 1 Discrete Dynamics in Nature and Society 7 Table 1: Types of attractions. 0.60 POI_ID POI_NAME Rating Type 1 +e Great Wall 456 History, scenery 0.55 2 Palace Museum 432 History, humanity 3 XiHu 112 Scenery, park 4 Nanshan Temple 321 Buddhism 0.50 5 Tiananmen Square 222 Politics, park 6 Fenjiezhou Island 532 Landscape, park 0.45 Table 2: User-attraction rating scale. 0.40 User_ID POI_ID Rating Times stamp 189 212 3 865454 231 331 2 823213 0.0 0.2 0.4 0.6 0.8 1.0 123 213 4 872443 Alpha 412 23 3 865231 k = 20 k = 80 11 321 5 865454 k = 40 k = 100 6 221 3 843233 k = 60 Figure 5: Effect of weighting parameters. experiment. From (9), μ + μ � 1, to ensure a single con- a p trollableexperimentalvariable;hereweproposeahypothesis of α � μ ,andthen μ � 1 − α,sothatasinglevariablecanbe a p controlled to observe the effect of the weighting parameters 0.75 on the experimental results. In this experiment, the uniform parameter α was taken in the range [0, 1], and the perfor- 0.74 mance of the algorithm was observed by adjusting the different neighbourhood users selected. As can be seen from Figure 5, the horizontal coordinate 0.73 representstherangeof theparameter values,andthevertical coordinate represents the MAE values, which change dif- ferently between different numbers of neighbours k as the 0.72 parameter goes from 0 to 1. +e MAE of the algorithm is optimal when the weight parameter a �0.6 and the number 0.71 of neighbours k �60. +e MAE value decreases at the be- ginning as the weight parameter a increases and starts to 05 10 20 30 400 60 70 increase when it exceeds α �0.6. +is is mainly because the Number of neighbor algorithm gradually ignores the evaluation of attraction attributes when the weight parameter exceeds 0.6. In order alpha = 0 to find the optimal set of neighbours for the target item, we alpha = 0.6 alpha = 1 set the range of neighbours for the target item to 60 and use three different weight control parameters a �0.061 to ob- Figure 6: Effect of number of residences. serve the influence of different number of neighbours on the recommendation result, which was observed by using three number of unrated scenic items. +e experimental results different weight control parameters; a �0.061. are shown in Figure 7. As shown in Figure 6, the overall performance of the algorithm’s ME value is low after fusing the optimal weight From Figure 7, it can be seen that the MAE values of the algorithms in this paper are relatively low when crossing the parameters, and the MAE of the algorithm gradually de- other two algorithms. Traditional algorithms IBCF and creases as the number of project neighbours increases and ITEM-CFhavedifficultyinobtainingthenearestneighbours gradually stabilizes when the number of neighbours exceeds of the target items due to the lack of basis for calculating the 50. +e experimental results show that when the number of similarity matrix due to the absence of a large amount of neighbours is chosen around 40, the algorithm can achieve rating data. In this paper, the algorithm uses a combination the optimal recommendation result. Comparing different of attraction scores and project attributes to calculate the algorithms, this section compares the algorithms in this global similarity in the absence of project scores, combined paper (RACF with the traditional user-based collaborative with the inherent project attributes to assist in the calcu- filtering algorithm (IBCF) and the improved item-based lation of global similarity, which alleviates the problem of collaborative filtering algorithm IITEM-CF). 800 users were selected in the experimental dataset, which contained a large data sparsity to a certain extent. MAE MAE 8 Discrete Dynamics in Nature and Society 0.95 4000 0.90 3500 0.85 0.80 0.75 0.70 0.65 0.60 0 20 40 60 80 movie travel book IRACF Figure 8: Distribution of test data statistics. IBCF ITEM-CF Figure 7: Comparison results of the same algorithm. (5) Without distinguishing the data sources, the global data is considered, and the movie data source, book data source, and tourism data source are regarded as 5.3. Real Life Examples. Based on the tourism data of the overall target data domain, and the similarity is domestic Ctrip, we used the dynamic agent technology calculated for the target user in the target data do- of Python scratch crawler framework to capture 1039 main and the nearest neighbours are selected, and scenic spots in 6 cities in China (Beijing, Shanghai, themissingratingitemsofthetargetuserarefilledto Guangzhou, Guilin, Hangzhou, and Haikou) from a make the user’s rating of the attractions in the domestic tourism website. It contains 8720 users and tourism data source. 23305 scoring records of scenic spots. Each user can store the scenic spots within the range of [1, 5]. At the +e data set division in the experiments is uniformly same time, there are 1500 tourism strategies, corre- constructed in the form of proportional division; i.e., the sponding scenic spot types, number of scenic spots ratio of the training set to the test set is 8:2, and the al- visited in different seasons, etc. gorithmmodelistrainedbasedonthedatainthetrainingset In total, the experiments in this section involve three to predict the users’ scores for the unrated items in the test data sources: the movie data sources, book data sources, and set. Firstly, as shown in the figure, we conducted a statistical travel data sources, using movies and books as secondary analysis of the scores in the secondary data sources to verify data sources to make predictions on the target data source, the correlation between the data sources, and the experi- the travel domain. +e experiments in this paper are divided mental results are shown in Figure 8. into the following tasks. Figure 8 shows the number of users who have rated books in the different data sources. Since the number of (1) +e movie data source and the book data source, users selected is fixed (5000), we can observe that there is a respectively, are used as auxiliary data sources to certain coverage of users’ behaviour in the different data calculate the similarity between users, to predict sources. +e main reason for this phenomenon is that if a users who did not make a rating, to calculate the user has marked books on the topic of history and hu- nearest neighbours, to calculate the correlation co- manities several times in the book data source, he will also efficient using the modified cosine similarity, and pay more attention to movies related to history and hu- thus to predict the items that populate the user who manities in the movie field, and by sending similar users in did not make a rating. thesecondarydatasource,ithelpsthetargetdatasourcefield (2) Based on the similarity between users calculated to discover similar users in the target field faster. +is is also from the auxiliary data sources, predictions were a prerequisite for fusion modelling of multiple data sources. made to the target according to the source of the Figure 9 shows the performance of the algorithm in the items that were not rated. case of different data sources. (3) Fusion of the target data source and the auxiliary In Figure 9, the horizontal coordinate indicates the data source was carried out. +e algorithm in this number of neighbours of the selected target user, and the paper is used to score the target data source. vertical coordinate indicates the value of the system average (4) Without differentiating data sources, the global data absolute error MAE. From the above figure, it can be seen areconsidered, andthe movie datasource, book data that the algorithm’s prediction in the single data source source, and travel data source are considered as the environment is difficult to find suitable neighbours in ef- overall target data domain. fective time due to the sparsity of the data, resulting in low MAE Discrete Dynamics in Nature and Society 9 the traditional culture of unique ethnic minorities in 0.94 Yunnan (no. 20BMZ164). 0.93 References 0.92 [1] M. Wang, “Research on convergence of rural tourism and related industries based on the perspective of holistic tourism: practice analysis from China,” Ekoloji, vol. 28, no. 107, 0.91 pp. 4295–4302, 2019. [2] P. Yochum, L. Chang, T. Gu, and M. Zhu, “Linked open data 0.90 in location-based recommendation system on tourism do- main: a survey,” IEEE Access, vol. 8, pp. 16409–16439, 2020. 0.89 [3] Y. Yi, C. Zhao, and J. Fu, “Research on the integration and development of modern agriculture and rural tourism based onAHP—takeYangjia towninMianyangcityasanexample,” 52 10 15 205 Open Journal of Social Sciences, vol. 7, no. 7, pp. 76–87, 2019. Neighbor [4] D. Wu, C. Zhang, L. Ji, R. Ran, H. Wu, and Y. Xu, “Forest fire travel book recognition based on feature extraction from multi-view travel movie images,” Traitement du Signal, vol. 38, no. 3, pp. 775–783, travel book movie [5] H. Zhang, Y. Duan, and Z. Han, “Research on spatial patterns Figure9:Performanceofthealgorithminthispaperwiththesame and sustainable development of rural tourism destinations in data source. the yellow river basin of China,” Land, vol. 10, no. 8, p. 849, [6] J.ZhangandL.Dong,“Imagemonitoringandmanagementof recommendation accuracy. As new data sources are added, the hot tourism destination based on data mining technology in algorithm’s MAE value starts to decrease compared to the big data environment,” Microprocessors and Microsystems, single data source environment, and the algorithm’s MAE vol. 80, Article ID 103515, 2021. value is lowest when fusing global data sources and reaches the [7] G. Alexandridis, A. Chrysanthi, G. E. Tsekouras, and optimal result when the number of neighbours is 25. G. Caridakis, “Personalized and content adaptive cultural heritage path recommendation: an application to the Gournia 6. Conclusions and Çatalhoy ¨ uk ¨ archaeological sites,” User Modeling and User-Adapted Interaction, vol. 29, no. 1, pp. 201–238, 2019. +is paper proposes a relatively novel travel recommendation [8] Q. Li, S. Li, S. Zhang, J. Hu, and J. Hu, “A review of text algorithm with fusion of multiple data sources. +e algorithm corpus-based tourism big data mining,” Applied Sciences, adopts the idea of tensor orthogonal decomposition and fuses vol. 9, no. 16, p. 3300, 2019. multipledatamodelstopredicttheratingofthetargetdomain. [9] F. Randelli and F. Martellozzo, “Is rural tourism-induced built-up growth a threat for the sustainability of rural areas? +eintegratedconsiderationofmultipledatasourcesunderthe +e case study of Tuscany,” Land Use Policy, vol. 86, use of the target domain is to assist the target domain to pp. 387–398, 2019. discoverthetargetuserneighbourhoodusersmorequicklyand [10] H. Li, D. Zeng, L. Chen, Q. Chen, M. Wang, and C. Zhang, more accurately to discover the user’s interest degree. It is “Immune multipath reliable transmission with fault tolerance worth pointing out that the recommendation algorithm pro- in wireless sensor networks,” in Proceedings of the Interna- posed in this paper under the fusion of multiple data sources is tional Conference on Bio-Inspired Computing: ?eories and not necessarily applicable to data sources with weak correla- Applications, pp. 513–517, Springer, Singapore, 2016 October. tion. +e algorithm has the lowest MAE value when fusing [11] I. Ibrahim, K. Zakariya, and N. A. Wahab, “Satellite image global data sources and reaches the optimal result when the analysis along the kuala selangor to sabak bernam rural number of neighbours is 25. tourism routes,” in Proceedings of the IOP Conference Series: Earth and Environmental Science, vol. 117, no. 1, Article ID 012013, United Kingdom, 2018. Data Availability [12] G. Bambi, S. Iacobelli, G. Rossi, P. Pellegrini, and M. Barbari, “Rural tourism to promote territories along the ancient roads +e dataset used in this paper is available from the corre- of communication: case study of the rediscovery of the st. sponding author upon request. Francis’s ways between fv,” European Countryside, vol. 11, no. 3, pp. 462–474, 2019. Conflicts of Interest [13] R. Beridze, I. Kordzaia, M. Diasamidze, and N. Beridze, “Sustainable rural tourism development recovery from covid- +e authors declare that they have no conflicts of interest 19 (in adjara region),” Globalization and Business, vol. 10, regarding this work. pp. 287–294, 2020. [14] S. Gorm ¨ u¨¸,s E. Atmi¸,s H. B. Gun ¨ ¸en, s N. K. Ozkazanç, and Acknowledgments M. Artar, “+e importance of mapping natural and cultural routes in rural tourism: bartın case,” International Journal of +is work was supported by the National Social Science Agricultural and Natural Sciences, vol. 10, no. 1, pp. 32–38, Fund Item, Research on Ecological Aesthetics in line with 2017. MAE 10 Discrete Dynamics in Nature and Society [15] I. Triˇsic,´ S. Stetic,´ D. Privitera, and A. Nedelcu, “Wine routes in Vojvodina Province, Northern Serbia: a tool for sustainable tourism development,” Sustainability, vol. 12, no. 1, p. 82, [16] T. Wijijayanti, Y. Agustina, A. Winarno, L. Istanti, and B. Dharma, “Rural tourism: a local economic development,” Australasian Accounting, Business and Finance Journal, vol. 14, no. 1, pp. 5–13, 2020. [17] C. J. A. Mitchell and M. Shannon, “Establishing the routes to ruralin-migrantproprietorshipinaCanadiantourismregion: a mobilities perspective,” Population, Space and Place, vol. 24, no. 3, p. e2095, 2018. [18] L.Wang,C.Zhang,Q.Chenetal.,“Acommunicationstrategy of proactive nodes based on loop theorem in wireless sensor networks,” in Proceedings of the 2018 Ninth International Conference on Intelligent Control and Information Processing (ICICIP), pp. 160–167, IEEE, Wanzhou, China, 2018, November. [19] M. A. Uwaisy, Z. K. A. Baizal, and M. Y. Reditya, “Recom- mendation of scheduling tourism routes using tabu search method (case study bandung),” Procedia Computer Science, vol. 157, pp. 150–159, 2019. [20] S. Li, S. Takahashi, K. Yamada, M. Takagi, and J. Sasaki, “Analysis of SNS photo data taken by foreign tourists to Japan and a proposed adaptive tourism recommendation system,” in Proceedings of the 2017 International Conference on Progress in Informatics and Computing (PIC), pp. 323–327, IEEE, Nanjing, China, 2017, December. [21] W. Zheng, Z. Liao, and Z. Lin, “Navigating through the complex transport system: a heuristic approach for city tourism recommendation,” Tourism Management, vol. 81, Article ID 104162, 2020. [22] L.Zhang,“Intelligenttourismrouterecommendationmethod based on big data,” International Journal of Autonomous and Adaptive Communications Systems,vol.13,no.4,pp.329–341, [23] J. Xu and J. Xu, “Research on the design of online travel service recommendation system based on data analysis,” in Proceedings of the International Conference on Cognitive Based Information Processing and Applications (CIPA 2021), pp. 1042–1046, Springer, Huainan, China, 2022 August. [24] X. Zheng, Y. Luo, L. Sun, Q. Yu, J. Zhang, and S. Chen, “A novel multi-objective and multi-constraint route recom- mendation method based on crowd sensing,” Applied Sci- ences, vol. 11, no. 21, p. 10497, 2021. [25] L. Feng, “Design of tourism intelligent recommendation modelof mounttaiscenicareabased on knowledgegraph,”in Proceedings of the 2020 International Conference on E-Com- merce and Internet Technology (ECIT), pp. 241–244, IEEE, Zhangjiajie, China, 2020, April. [26] C. Cao, Y. Tang, D. Huang, W. Gan, and C. Zhang, “IIBE: an improved identity-based encryption algorithm for WSN Se- curity,” Security and Communication Networks, vol. 2021, Article ID 8527068, 8 pages, 2021. [27] C. Yuan and M. Uehara, “An optimal route recommendation method for a multi-purpose travel route recommendation system,” in Proceedings of the International Conference on Broadband and Wireless Computing, Communication and Applications, pp. 398–408, Springer, Fukuoka, Japan, Octomber 2019.

Journal

Discrete Dynamics in Nature and SocietyHindawi Publishing Corporation

Published: Apr 14, 2022

There are no references for this article.