1 Introduction

Emotional tendency analysis is a process that evaluates the emotions presented in the textual data. Emotional tendency analysis identifies the emotions such as affection, fear, rage, sorrow, and happiness of the users [1]. Emotional tendency analysis is mostly used for tourism recommendation systems. A construction intensity-based tendency analysis for the tourism development process is recommended [2]. The main aim of the approach is to evaluate the preferences of users which are gathered from feedback. The construction intensity-based approach identifies the intensities that create a major impact in providing services to the customers [3, 4]. The tendency analysis technique increases the accuracy of analysis and enhances the feasibility range of the systems. A multi-dimensional tourist satisfaction method is used for the tourism department [5]. The tourist satisfaction method is mainly used to evaluate the behavioral patterns and preferences of tourist customers. The satisfaction method improves the overall feasibility and significance range of the tourism systems. The satisfaction method measures the exact emotional tendency of the customers which reduces the latency in the computation process [6].

Tourist preference-based emotional tendency analysis is widely used to improve the performance and satisfaction range of tourism systems. Tourist preference detection is a crucial task to perform which produces relevant data for further enhancement processes [7]. A tourist preference-based emotional tendency analysis method is used for recommendation systems (RS) for tourism applications. It is a personality-aware RS model that personalizes the feedback based on the emotions of the customers. It is used as a model which uses heterogeneous groups which identify the preferences and interests of the users [8, 9]. Both spatial and temporal features of the emotions are detected from the content. The features provide the exact personalities of the customers which minimizes the computational complexity in RS [10]. A convolutional neural network (CNN) and long short-term memory (LSTM)-based sentiment analysis technique are used for tourism systems [9]. The CNN algorithm uses a feature extraction method to extract the necessary features from the emotional aspects of the customers. The actual emotional tendencies of the customers are evaluated for destination recommendation services [11].

The optimization algorithm is mostly used to identify the best solution to solve a problem in the systems. Optimization algorithms are commonly used to reduce the computational complexity ratio in tourism systems [12]. Optimization algorithm-based emotional tendency analysis is used for tourism systems. An optimization algorithm based on multinomial logit (MNL) is implemented for the emotional analysis process [12]. The MNL-based method identifies important emotional characteristics that are relevant to the analysis process. The identified characteristics produce feasible information that minimizes the latency in the computation process [13]. The heterogeneous preferences of users are evaluated which enhances the functional aspect and accuracy of recommendation services. A classification algorithm-based mining method is used for the emotional tendency analysis process [14]. The classification algorithm classifies the exact emotional abilities of the users that provide preference for the recommendation process [15]. The classification algorithm provides necessary information to tourism development systems. A mining technique is used here to classify the types of emotions that enhance the structural capacity of the systems [16]. The contributions are:

  • Proposing, designing, and validating a novel control method by assimilating LSTM and fuzzy control algorithm for traveler emotion tendency analysis

  • Reducing the repetition in data analysis by reducing the discreteness based on availability and filtering the adaptable information

  • Performing a data-dependent analysis along with a comparative study using external data sources, metrics, and methods for assessing the proposed method’s efficacy

Gezhi et al. [17] proposed a conceptual model for environmentally responsible behavior (ERB) in the tourism department. The proposed model is mainly used here to analyze the positive emotions and self-efficacy range in tourism systems. It is commonly used to create a major impact that analyses the good emotions of the tourist during the traveling period. The proposed model provides a feasible solution to solve issues in the environmental management of tourism destinations. Choachaicharoenkul et al. [18] developed a multi-objective trip planning method based on user preference and restaurant selection for tourist trip design problems (TTDP). The main aim of the method is to maximize the feasibility and significance range of the systems. It uses a greedy algorithm to minimize the total distance between the destinations. It is a multi-objective that is used to detect the TTDP issues in the tourism system. The developed method enhances the performance level of the systems. Chen et al. [19] introduced a latent class choice approach for mobility-as-a-service (MaaS). The actual goal of the approach is to evaluate the preference of tourists which provides feasible services to the users. It examines the exact preference of the tourists which minimizes the latency in the computation process. It produces relevant information for service bundles for the systems. The introduced approach improves the recommendation and feasibility range of the applications. Kovačić et al. [20] designed a new mobility preference for smart technologies in tourist regions. The exact preference of the tourists is evaluated based on the need and positive emotion of the tourist. A quantitative analysis technique is used here to analyze the relationship between variables and characteristics of the systems. The designed method enhances the satisfaction level of tourists during the travel period. Kim et al. [21] proposed a tour-based stated preference (SP) for the tourist preference process. A bundle configuration is used here which evaluates the MaaS that produces optimal modes for the travel process. Integrated information is implemented to improve the preference range of the systems. It is mostly used for transportation sustainability which provides various modes for the systems. The proposed model reduces the overall complexity of designing tourism models. Mancini et al. [22] developed a user preference-based optimal selection method for sports mega-events. A kernel search is used in the method to evaluate the necessary data for the tourism architecture process. It addresses the issues that cause damage to the tourism services which maximizes the efficiency range of the systems. The developed method provides excellent performance to the tourism department. Ruiz-Meza et al. [23] introduced a new tourist trip design using a heterogeneous preference for the environmental impact creation process. The actual goal of the design is to provide an effective tourism strategy for the decision-making process. It facilitates planning by improving the accuracy of the decision-making process. The introduced design minimizes the emissions and benefits of the tourism environment. The rest of the references are tabulated in Table 1.

Table 1 Summary of the references
Full size table

The methods/ approaches discussed above focus on briefing and representing the emotional and behavioral tendencies of users as seen in [17] and [27]. Some other works present ontological and machine-learning kind of methods for analyzing the tourist contexts as in [30] and [24]. The highlighted methods improve assessments that are single faced such as the feasibility and contextual analysis. Different from these methods, the proposed method focuses on relating repeated and non-repeated data for user preference and availability analysis. This improves the precise study and assessment of individual land group emotions. Such a process improves the data accumulation, need, presentation, and survey with recommendations for tourist-related responses. Therefore, this article presents a differentiable algorithm assimilated tourism data analysis method for emotional tendency validations.

2 Dataset Description

The data used in this article are acquired from [31] which presents brief information on “sentiment analysis” from a renowned travel recommendation and planning website. The tourist review (4000 Nos.) is used for emotional tendency analysis in particular for tourist attractions. The classifications are pursued for data repetition and its continuity between successive years for 10 fields. Based on the fields, the emotion tendency extraction and representation are given in Fig. 1.

Fig. 1
figure 1

Emotion tendency illustration

Full size image

The data acquired from distinguishing categories experienced by the consumers are used for validating preferences and availability for emotion analysis. User preferences and data availability are the key terms for emotion and analysis over the timeline. Such emotional tendencies are either repeated/ recommendable based on current reviews (Fig. 1).

3 Organized Combinational Control Method

The proposed OC2 method is designed to improve the validation tendencies of the consumer's travel plan and optimal recommendations using LSTM and a fuzzy control algorithm. The emotional tendency analysis of the travelers is validated for filtering the non-repeated data from the previous travel/tour history for controlling data discreteness through LSTM and FC. The large volume of data observed from the travelers/consumers such as their interests, travel plans, visit sites, value for money, and location preferences is briefly analyzed for which the outdated/trivial data from the consumer's emotional learning is to be thwarted through non-recurrent analysis. OC2M is one such method that makes use of LSTM and FC for the classification of observed data. The proposed OC2M’s process flow is presented in Fig. 2.

Fig. 2
figure 2

OC2M’s process flow

Full size image

The process of OC2M for the emotional tendency of travelers analysis based on observed data from traveler’s plans, interests, preferences, etc. for classifying the repeated and discreteness in data through LSTM. The observed data is analyzed for classification using LSTM based on places and emotional connectivity. The LSTM output is used for identifying the repeated data and discreteness through correlation from the previous history observed. The process of OC2M is analyzed to provide optimal traveling recommendations, where the non-repeated data is initially filtered. The traveler’s data input is expressed as

$${\text{Trav}}_{d_i } = \frac{1}{{tp}}\left\| {\sum_{z = 1}^{tp} {{\text{Trav}}_a } \left( z \right)*{\text{Trav}}_b \left( z \right)*{\text{Trav}}_c (z)} \right\|,$$
(1)

where

$${\text{Trav}}_{abc}\left(z\right)=\frac{1}{\pi }\underset{0}{\overset{1}{\int }}\frac{{\text{Trav}}_{abc}\left(z\right)}{tp}d.tp,$$
(2)

where the variables \({\text{Trav}}_{a}\left(z\right)\), \({\text{Trav}}_{b}\left(z\right),\) and \({\text{Trav}}_{c}\left(z\right)\) used to represent the traveler's travel plan, interests, and preferences at different times. These observed data are required for emotional tendency analysis of tourism consumers \(z\) for the previously visited place's connectivity and user interests. If \(a\), \(b,\) and \(c\) change for consumer's interests, preference, and emotional connectivity at different periods \(tp\) and hence,

$$z=\frac{1}{\pi }\sum_{tp}^{1}{\text{Trav}}_{a+b+c}\left(z\right) d.tp.$$
(3)

Based on Eq. (3), the first initial non-repeated data are filtered for all the users; this information is observed from the previous travel/tour history based on users \(a\), \(b,\) and \(c\) in different time intervals and is represented as \(\left(N\left(FP\right)\times tp\right)\). Here, \(N\left(FP\right)\) represents the number of filtering processes pursued for sequential and discrete data observed instances. The data filtering process is illustrated in Fig. 3.

Fig. 3
figure 3

Data filtering process

Full size image

The data filtering is performed using LSTM for its a and b in extracting \(di\) (Repetition). In the process \(d\) and \({d}_{i}\) are correlated for \(\text{Tra}{V}_{a}\), \(\text{Tra}{V}_{b}\), and \(\text{Tra}{V}_{c}\) for detecting \(N\left(FP\right)\) and \(\frac{1}{N\left(FP\right)}\). This process requires 3 combinations \(\left(a,b\right), \left(b,c\right),\) and \(\left(c,a\right)\) for validating the \(d.{t}_{p}\) across different \(Tra{V}_{{d}_{i}}\). This filtering excludes \(\frac{1}{N\left(FP\right)}\) for the different \(\left(Z,{t}_{p}\right)\) (Refer to Fig. 3). The filtering output analysis is presented in Fig. 4.

Fig. 4
figure 4

Filtering output analysis

Full size image

The filtered output is validated for the different categories expressed in Fig. 1. The above representation is just the data replication observed from the input dataset. The filtering is pursued for the \(\text{Tra}{V}_{{d}_{i}}\) and \(\text{Tra}{V}_{{d}_{z}}\) such that the categories are cumulatively augmented. This means the data are added along the different categories in the above analysis. The LSTM operates on this measure for increasing \(Tra{V}_{{d}_{i}}\) and \(\left(N\left(FP\right)\right)\). (Refer to Fig. 4). Filtering is pursued to reduce the repeated data and data discreteness present in \({\text{Trav}}_{{d}_{i}}\). Non-repeated data occurred due to the maximum repeated data and discreteness observed in this analysis at any time. This normalization is computed using a high pass filter to support such analysis is given as

$$\left.\begin{array}{c}{\text{Trav}}_{a}\left(z,tp\right)=\frac{{\text{Trav}}_{a}\left(tp\right)}{tp}*\frac{\Delta }{2} {RD}_{i}\left(N\left(FP\right)\times tp-d\right)\\ And,\\ {\text{Trav}}_{b}\left(z,tp\right)=\frac{{\text{Trav}}_{b}\left(tp\right)}{tp}*\frac{\Delta }{2} {RD}_{j}\left(N\left(FP\right)\times tp-d\right)\end{array}\right\},$$
(4)

where

$${RD}_{i}=\left|\frac{{\text{Trav}}_{{d}_{i}}\left(tp\right)-d}{2}\right|RD\left(tp\right)$$
(5a)

and

$${RD}_{j}=\left|\frac{{\text{Trav}}_{{d}_{i}}\left(tp\right)-d}{2}\right|RD{\left(tp\right)}_{N\_RD-1}.$$
(5b)

Based on the above equations, the variables \({RD}_{i}\) and \({RD}_{j}\) are the classification of maximum and minimum repetition in sequential data observation. The factor \({\complement }_{emd}(tp)\) denotes the consumer's emotional data based on \({RD}_{i}\) and \({RD}_{j}\). Based on the non-repeated data \(N\_RD\) and discreteness \(d\) occurred from the sequence of data using LSTM and fuzzy control. The variable \(\Delta\) means the capacity of the filters based on the user interests, preferences, and emotional connectivity for tendency analysis. Now, the normalized filtering of repeated data from sequential observation is defined as in Eqs. (6a and b)

$$\exists \left[RD\left(tp\right)\right]=\frac{\frac{\Delta }{2}\left(N\left(FP\right)\times tp-d\right)}{{tp}^{2}} \left[{RD}_{i}-{RD}_{j}\right]$$
(6a)
$$=\frac{1}{tp}\left[\underset{0}{\overset{1}{\int }}\frac{{RD}_{i}\left[\left(N\left(FP\right)\times tp\right)-{2}^{d}\right]}{tp}d.tp-\underset{0}{\overset{1}{\int }}\frac{{RD}_{j}\left[\left(N\left(FP\right)\times tp\right)-{2}^{d}\right]}{tp}d.tp\right].$$
(6b)

Equations (6a and b), the normalized less data repetition \(\exists \left[RD\left(tp\right)\right]\) is observed after filtering the non-repeated data from the previous tour/travel history to reduce the data discreteness. From this \(\exists \left[RD\left(tp\right)\right]\), the consumer’s emotional tendency identification based on the user’s interests, preferences, and emotional connectivity with the place/plan is observed from the previous histories for further classification of valid and invalid adaptable data for providing optimal recommendations. Equations (7) and (8) used to compute the valid adaptable data \(\left({V}_{AD}\right)\) and invalid adaptable data \(\left(IN\_{V}_{AD}\right)\) are computed as

$${V}_{AD}=\frac{1}{2\pi (N\left(FP\right)\times tp)}\left|\sum_{rm=1}^{tp}\left({a}_{i/j}-{b}_{i/j}\right)d\right|, \forall {RD}_{j}={RD}_{i}+1$$
(7)

and

$$IN\_{V}_{AD}=-\sum_{i={rm}_{\text{min}}}^{{rm}_{\text{max}}}{V}_{AD}\text{log}{\left({V}_{AD}\right)}_{\left(N\left(FP\right)\times tp\right)},$$
(8)

where \(rm\) denotes the recommendations for the travelers provided using a fuzzy control algorithm. Here,\(\,{rm}_{\text{max}}\) and \({rm}_{\text{min}}\) are the maximum and minimum recommendations observed. The log normalization of valid adaptable data also contains \(IN\_{V}_{AD}\) from the analysis is as in Eq. (9)

$$IN\_{V}_{AD}\left[\exists \left[RD\left(tp\right)\right]\right]=\frac{{V}_{AD}}{\text{log}\left[\frac{{rm}_{\text{max}}-{rm}_{\text{min}}}{tp}\right]}.$$
(9)

This log normalization is performed for filtering the adaptable data with the future tour/travel plan using \(\exists \left[RD\left(tp\right)\right]\) and \(IN\_{V}_{AD}\) it differs for the different time instances. The classification and filtration process is performed based on the place or plan connectivity with the users using LSTM and fuzzy control. The classification outputs before and after normalization are presented in Table 2.

Table 2 Classification outputs for 5 categories
Full size table

The before and after normalization for the 5 categories are analyzed in Table 2. Depending on the available \({V}_{AD}\) and \(IN \_{V}_{AD}\), the normalization is pursued. This normalization aids to classify \({V}_{AD}\) and \(IN \_{V}_{AD}\) across different \(\left(a,b\right) \left(b,c\right)\) and \(\left(c,a\right)\) combinations. Pursuing the \(\text{Tra}{V}_{{d}_{i}}\) data, the \({V}_{AD}\) is classified by \(\text{Tra}{V}_{{d}_{z}}\) such that the \({d}_{i}\in N\left(FP\right)\) neglecting \(\frac{1}{N\left(FP\right)}\). This is required for improving the LSTM discreteness across multiple replications (Table 2). This classification helps to identify and segregate the false recommendations and true recommendations for all travelers/consumers. In this classification, the user's interests, travel plans, and preferences are independently analyzed at each level to address the emotional connectivity of the travelers with the place or plan. The analysis of the emotional tendency of tourism consumers is becoming unmanageable due to the increasing population. Amid the challenges in this manuscript, the place connectivity and liable recommendations are the appropriate factors satisfying the travelers/users of different classes. Hence, regardless of the tourism consumers, liability in adaptable data based on the emotional connectivity of the places or plan is a prominent consideration here for tourism consumers. The proposed OC2M focuses on this consideration by providing optimal recommendations for travel plans through a fuzzy control algorithm. In this proposal, the emotional tendency is administrable for users and their tourism plan. The fuzzy process is illustrated in Fig. 5.

Fig. 5
figure 5

Fuzzy process

Full size image

The fuzzy process relies on \({\text{C}}_{\text{end}} (tp)\) input for categorizing \(\left(Z,tp\right)\) and \(\left(d.tp\right)\) independently. The first classification under \(\left(Z,tp\right)\) extracts all continuous (year-based, timeline) output grouped as \({V}_{AD}\). In this grouping pass, along with \(r{m}_{\text{max}}=\text{high},\) the normalization is performed. The \(IN \_{V}_{AD}\) classification under \((d.tp)\) and \(r{m}_{max}=min\) are discarded as the data become inconsistent. In this process, \(\left(Z,tp\right)\) and \(\left(d.tp\right)\) are the fuzzification process for mitigating recurrent data. The fuzzy process, thus, segregates the \({V}_{AD}\) from the actual input \({C}_{\text{end}} \left(tp\right)\) (Fig. 5). The tourism consumer analyzes their emotional tendency to identify repeated data and discreteness in mobile applications. OC2M classifies the valid and invalid adaptable data based on the tourism consumer’s interests, preferences, and emotional connectivity with the location or plan from previous histories. In this model, the data repetition and discreteness of the user’s data are easily computed for achieving maximum recommendations and validation tendencies for the varying tourism consumers. Further, this model aims to provide discreteness-less and repeated data-less information to maximize the consumer's emotional tendencies. The proposal classifies valid and invalid adaptable data concurrently for supporting such analysis. The emotional tendency analysis of travelers is different for non-repeated data and repeated data, to handle large volumes of data. The first emotional tendency of travelers is represented as

$$\left.\begin{array}{c}\underset{i\in \mathit{tp}}{\text{Max}}\,{\text{Trav}}_{{d}_{i}} \forall\, \frac{{RD}_{i}}{{RD}_{j}}={N\_RD}_{i}\\ \underset{j\in {\mathit{RD}}_{i}}{\text{Min}}d.tp \,\forall\, N\_RD,\text{ where}\, d.tp={tp}_{RD}-{tp}_{N\_RD} \\ \\ \text{and}\, \underset{i\in \mathit{tp}}{\text{Min}}\,{\complement }_{emd}(tp)\, \forall\, i\in \left(\frac{{RD}_{i}}{{RD}_{j}}+d\right)\end{array}\right\}.$$
(10)

In Eq. (10), the emotional tendency analysis of the \(n\) number of tourism users \(U\) at different time instances is processed. In the next sequential emotional data analysis, the data repetition and discreteness take place. The objective of minimizing the data discreteness is represented using the constraint \({\complement }_{emd}(tp) \forall i\in \left(\frac{{RD}_{i}}{{RD}_{j}}+d\right)\). The number of emotional data in the given time \((T)\) , which is expressed as \({\text{Trav}}_{{d}_{i}} \times T\) and the repeated data is \({U}_{n}\times {\text{Trav}}_{{d}_{i}}\). From the overall user’s emotional data analysis based on \({\text{Trav}}_{{d}_{i}} \times T\) and \({U}_{n}\times {\text{Trav}}_{{d}_{i}}\) is the adaptable data for a detailed analysis. Repeated data and discreteness are feasible using LSTM. From this instance, the classification of valid adaptable data and invalid adaptable data is essential to identify non-repeated data. The capacity of the filters is computed using fuzzy control; the additional analysis time needed for data repetition and discreteness is the assisting factor for improving optimal recommendations. The filtering of adaptable data is pursued for all the tourism consumers are validated using consumer’s emotional learning. Post, depending upon the adaptable data classification, the validation tendencies are the improving factor. From the filtering process, preference and recommendation are the prevailing instances for analyzing different users' emotional data. Table 3 presents the minimum and maximum recommendations for the three emotions observed.

Table 3 Minimum and maximum recommendations
Full size table

In Table 3 the minimum and maximum recommendations for the varying travelers and emotions are analyzed. The cumulative data inputs are used for analyzing \({C}_{\text{end}}\left(tp\right)\) and filtering process. This case is valid for \(N(FP)\) such that \(\left(a,b\right), \left(b,c\right),\) and \(\left(c,a\right)\) are optimal across different min and max differentiations. Therefore, the filtered inputs determine the required \(\text{Tra}{V}_{a}, \text{Tra}{V}_{b},\) and \(\text{Tra}{V}_{c}\) such that \(\frac{1}{N\left(FP\right)}\) is reduced. As this is reduced, the \(\exists \left[RD\left(tp\right)\right]\) is used for \({V}_{AD}\) provided \(IN \_{V}_{AD}\) are mitigated.

4 Repeated Data Analysis

In this sequential tourism consumers data analysis of emotional tendency based on LSTM and fuzzy control algorithm, the filtering of non-repeated data from the previous travel/tour history for \(\left({\text{Trav}}_{{d}_{i}} \times T\right)\) for all \({U}_{n}\) based on places and plan connectivity is the deciding factor. The probability of sequential data analysis of tourism consumers \({\rho }_{{{\complement }_{emd}}_{s}}\) continuously is given as

$${\rho }_{{{\complement }_{emd}}_{s}}={\left( 1-{\rho }_{RD}+{\rho }_{d}\right)}^{tp-1} \forall i+j\in tp,$$
(11)

where

$${\rho }_{RD}=\left(1-\frac{{\text{Trav}}_{{d}_{i}}\in N\left(FP\right)\times tp}{\frac{{rm}_{\text{max}}-{rm}_{\text{min}}}{tp}}\right)$$
(12)

and

$${\rho }_{d}=\sum_{tp=1}^{a,b,c}\left(\frac{{RD}_{i}}{{RD}_{j}}+d\right).$$
(13)

As per Eqs. (11, 12, and 13), the sequential analysis of the emotional tendency of travelers is based on the probability of repeated data and the probability of data discreteness such that there is outdated/trivial data. The filtering process depends on the user’s interests, preferences, and emotional connectivity for analyzing the data repetition and discreteness probabilities at the same time as tendency analysis. Hence, the FC process filters the adaptable data in computing as in Eq. (10). Therefore, the adaptable data with future tour plans are computed as

$$AD\left(N\left(FP\right)\right)=\frac{{\text{Trav}}_{{d}_{i}}}{\left|{\rho }_{RD}-{\rho }_{d}\right|} .{\rho }_{{{\complement }_{emd}}_{s}} \forall i+j\in T.$$
(14)

However, the optimal recommendation for emotional tendency analysis is valid for both \({\text{Trav}}_{{d}_{i}} \times T\) and \({U}_{n}\times {\text{Trav}}_{{d}_{i}}\) is to increase consumer's travel plans. The filtering process of non-repeated data at \(T\) instances to reduce the data discreteness \({U}_{n}\times {\text{Trav}}_{{d}_{i}}>{Trav}_{{d}_{i}} \times T\), the analysis is descriptive using the consumer’s emotional learning. Therefore, the identifiable constraints \({\rho }_{RD}\) and \({\rho }_{d}\) are less to satisfy Eq. (10). The analysis for the repeated and non-repeated data is presented in Fig. 6.

Fig. 6
figure 6

Repeated and non-repeated data analysis

Full size image

The \(IN\_{V}_{AD}\) and \({V}_{AD}\) data analysis is performed under three variations, i.e., categories, cumulative inputs (K), and \(d(\%)\). The LSTM is used for reducing the repeated data across various \(d\) such that the available data are used for \(\left(d.tp\right)\). In the fuzzification process, the recommendations are used for \(\rho \_{C}_{{\text{end}}_{\text{S}}}\) such that \(d\) is filled for further analysis. Therefore, the validations for \({V}_{AD}\) are comparatively less than \(IN \_{V}_{AD}\). This is optimal other than \(d\) variations due to \({\rho }_{RD}\) reduction such that \(m{ax}_{i}\in tp\) and \(m{in}_{j}\in RD\) are the validating conditions (Fig. 6). The contrary solution of this sequential data analysis is the prolonging \({\rho }_{d}\) and, hence, the maximum non-repeated data resulting in outdated/trivial data. The non-repeated data from previous travel histories is \({U}_{n}>tp\) and hence the consumer’s emotional tendency has been varied for each user. Based on the FC process filtration, the repeated and discrete data are identified here using the proposed model and LSTM. The probability of non-repeated data \(\left({\rho }_{{N\_RD}_{T}}\right)\) is given as

$${\rho }_{{N\_RD}_{T}}=\frac{{\rho }_{{{\complement }_{emd}}_{s}}. AD\left(N\left(FP\right)\right).\left[{\rho }_{RD}+{\rho }_{d}-\left(\frac{{RD}_{i}}{{RD}_{j}}\right)\right]}{rm.tp}.$$
(15)

In Eq. (15), the outdated or trivial data observed from the sequential tourism data are discrete. The consumer’s emotional tendency analysis as in the above equation requires high analysis time, thereby reducing the data discreteness. These constraints are addressable using LSTM and fuzzy control to mitigate the data discreteness through non-recurrent analysis. The output for classifying adaptable data relies on the consumer’s emotional learning. It aids in supporting such data analysis for both sequential and discrete instances. In this, both instances analyses are compared with the previous travel histories using a fuzzy control algorithm. Hence, the constraint for providing optimal recommendation is different, that follows LSTM output through the classification of valid and invalid adaptable data. The classification is prescribed for both the sequential and non-repeated data analysis by estimating the future tour/travel plan for users. The first FC process filter relies on the maximum recommendation. The data discreteness is observed from the sequential emotional data analysis of tourism consumers, where valid and invalid adaptable data are balanced and, hence, the recommendation is constant for all users as in Eq. (10). The recommendation changes for all the users based on travel plans, interests, and preferences to determine the emotional tendency for identifying repeated data and discreteness as in the above equations. The available tourism user’s emotional tendency analysis is validated for improving optimal recommendation and reducing discreteness.

5 Performance Assessment

The performance assessment is performed using recommendation assessment, data discreteness, connectivity analysis, analysis time, and data repetition. The X-axis is varied for the number of travelers between 100 and 1200 and the max. data are acquired from plan, stay, expense, travel, food, entertainment, misc. The existing TSP [21], TPEDTR [24], and eTOUR [28] are considered in the background section for the proposed method’s efficacy verification.

In Fig. 7, the emotional tendency of travelers is analyzed based on their user’s interests, travel plans, and preferences using the proposed model and LSTM to improve the optimal recommendation ratio. In the LSTM process, repeated data and data discreteness are identified to support such analysis using LSTM, and the fuzzy control algorithm relies on adaptable data. The emotional tendency analysis for all the users is validated for filtering the non-repeated data occurrence from the previous travel/tour history for reducing data discreteness using LSTM and FC. The outdated data is addressed for the future tour/travel plan with the constraint \(\left(N\left(FP\right)\times tp\right)\) is satisfied for filtering non-repeated data from the sequence of instances using FC. The FC process continuously filters the adaptable data to prevent trivial data, if these data are detected from any instance; the recommendation is modified for the travelers. Therefore, the emotional tendency is analyzed along with the consumer's emotional learning to improve the recommendation efficiency.

Fig. 7
figure 7

Recommendation efficiency

Full size image

The data discreteness is identified from the instance for classifying repeated and non-repeated data. The previous tour/travel histories for detailed emotional tendency analysis of travelers are represented as discreteness in Fig. 8. Based on the analysis, the previous and current data are compared to identify variations and repetitions in data that rely on LSTM and FC to increase emotional connectivity with the place or plan. In this process, non-repeated data is trained for future tours or travel plans to prevent outdated data from the instance. This proposed model functions two processes such as LSTM and fuzzy control for the available users for emotional tendency analysis and this assessment is easy for satisfying less data discreteness. In the sequential analysis of the traveler’s emotional tendency analysis, the analysis time and repetition increase, and the optimal traveling recommendations are provided based on the valid adaptable data preceded using Eqs. (7, 8, 9, and 10) estimation. From the instance, less data discreteness is achieved compared to the other factors.

Fig. 8
figure 8

Data discreteness

Full size image

The sequential traveler’s emotional tendency analysis is pursued to avoid the repeated data and trivial data from the instance using OC2M; the LSTM first filters the non-repeated data for improving the user’s interests as illustrated in Fig. 9. These observed data are required for sequential emotional tendency analysis of tourism consumers \(z\) from the previously visited places connectivity and their preferences. This proposed scheme satisfies high connectivity analysis with the plan and places are analyzed for maximizing the recommendation to those particular places. In this connectivity analysis, the large volume of data observed from the users is analyzed at different instances or intervals. The convergence between the user's information is identified for reducing repetition until discreteness is identified in that instance. Hence, the recommendation is a change based on users' emotional and place connectivity; the proposed model and LSTM perform high connectivity analysis.

Fig. 9
figure 9

Connectivity analysis

Full size image

In this proposed model, the traveler’s emotional tendency analysis time is less compared to the other factors. Based on the user’s personal interests, travel plans, and value for money are analyzed for providing optimal recommendations for traveling. The consumer’s emotional learning process influences the outdated or trivial data from the instance to support such analysis. This proposed scheme is used to provide authentication for all the user requests and service provider’s responses while augmenting the information requirement. The observed sequential data are analyzed for filtering the data using LSTM based on places and emotional connectivity. The LSTM output is used for identifying the repeated data and discreteness through correlation from the previous history observed to improve the traveler’s interests. Filtering is pursued to reduce the repetition and discreteness present in \({Trav}_{{d}_{i}}\). Non-repeated data occurrence due to the maximum repeated data and discreteness observed from this continuous analysis at different instances in which the less analysis time is satisfied by the proposed model is as presented in Fig. 10.

Fig. 10
figure 10

Analysis time

Full size image

Based on users, \(a\), \(b\) and \(c\) are identified in different time intervals is represented as \(\left(N\left(FP\right)\times tp\right)\) for reducing discreteness in data. Here, \(N\left(FP\right)\) used to pursue the sequential and discrete data analysis for future travel plans is represented in Fig. 11. Based on the analysis, the \({RD}_{i}\) and \({RD}_{j}\) are classified from the sequential data observation and filters the adaptable data using LSTM and FC. The \({RD}_{i}\) and \({RD}_{j}\) in consumer’s emotional data are observed and analyzed in different instances for reducing discreteness. From this \(\exists \left[RD\left(tp\right)\right]\), the sequential consumer’s emotional tendency analysis is performed based on the user’s interests; preferences, and emotional connectivity with the place/plan identified from the previous histories. In this process, the future travel plan is recommended based on the classification of valid and invalid adaptable data through fuzzy control. In this proposed model, repeated and outdated data are avoided to improve the traveler’s interests. The classification and filtration process is performed based on the place or emotional connectivity of the users and is identified using LSTM and fuzzy control. This classification helps to improve the optimal recommendations for all travelers/consumers. Therefore, less repetition is achieved, thereby increasing user interest. Table 4 summarizes the comparative analysis results for the varying travelers and the data categories.

Fig. 11
figure 11

Data repetition

Full size image
Table 4 Comparative analysis summary
Full size table

6 Conclusion

To improve the emotional tendency analysis of travelers, this article introduced the organized combinational control method. This method integrates the LSTM and fuzzy control algorithms for improving preference connectivity based on adaptable and inadaptable data. Before this process, the LSTM operations were valid in identifying discrete and repetition preferences for reducing computing times and similar emotions. The fuzzy control distinguishes the data adaptability using previous travel history such that the matching emotions are extracted across various timelines. Therefore, filtering data are required for identifying adaptability that eventually increases the recommendation efficiency. The process is repeated until maximum discreteness is utilized such that the impact is less feasible for various other travel data inputs. The further emotion tendency assessment is performed based on the above data across various connectivity analyses and validation tendencies. This feature reduces the discreteness in data availability for emotional tendency analysis from different tourists and data categories.