Jing-ya Hao and Ying Wu
[1] Z. G. Okem and D. Pekkurnaz, Healthcare users’ experiencewith E-health: Benefits, drawbacks, and the future, inTransforming health in Turkey: An evaluation of two decadesof reform, (Abingdon: Taylor and Francis, 2025), 141. [2] Y. Xu, Y. C. Liu, H. S. Xu, and H. Tan, AI-driven UX/UI design:Empirical research and applications in FinTech, AcademiaNexus Journal, 3(1), 2024. [3] S. Frans, M. R. T. Dewa Dominica, I. K. Lucky, S. Lilik, andY. U. Eva, Application of the user centered design method toevaluate the relationship between user experience, user interfaceand customer satisfaction on banking mobile application, JurnalInformasi Dan Teknologi, 6(1), 2024, 7–13. [4] E. Astuti, I. Harsono, S. Uhai, H. N. Muthmainah, and A.Y. Vandika, Application of artificial intelligence technology incustomer service in the hospitality industry in Indonesia: Aliterature review on improving efficiency and user experience,Sciences Du Nord Nature Science and Technology, 1(1), 2024,28–36. [5] X. Yang, L. Zhang, and Z. Feng, Personalized tourismrecommendations and the E-tourism user experience, Journalof Travel Research, 63(5), 2024, 1183–1200. [6] O. D. Segun-Falade, O. S. Osundare, W. E. Kedi, P. A.Okeleke, T. I. Ijomah, and O. Y. Abdul-Azeez, Developingcross-platform software applications to enhance compatibilityacross devices and systems, Computer Science & IT ResearchJournal, 5(8), 2024, 2040–2061. [7] Z. Gao, J. H. Cheah, X. J. Lim, S. I. Ng, T. H. Cham,and C. L. Yee, Can travel apps improve tourists’ intentions?Investigating the drivers of Chinese gen Y users’ experience,Journal of Vacation Marketing, 30(3), 2024, 505–534. [8] J. S. Kim, M. Kim, and T. H. Baek, Enhancing user experiencewith a generative AI chatbot, International Journal of Human–Computer Interaction, 41(1), 2025, 651663. [9] B. B. Feng, L. L. Zhang, and J. Yin, Simulation ofpersonalized recommendation algorithm for user feedbackdata of social networks, Computer Simulation, 41(11), 2024,375–379. [11], and T = 9 in this paper. The modified competi-tive weight MCWi is normalised to get an improved weightωIi , ωIi = MCWiNei=1 MCWi = MCWi.At this point, the basic weight ωiB and the improvedweight ωiI of the experience dimension are obtained, both ofwhich are the weights after their respective normalisation.The final experience dimension weight is obtained bysynthesising and normalising the two:ωi=ωiB · ωiIi=Nei=1 ωiB.ωiI(7)2.3 Improve the Analysis of Emotional ExperienceFactors at the Level of APP InterfaceOperationOn the basis of understanding the user’s trust and theweight of experience dimension, analyse the factors thatimprove the emotional experience at the level of APPinterface operation. The emotional design of products isa design concept based on people’s inner emotional needsand spiritual needs, and ultimately creates products thatare physically and mentally pleasant, so that life is full offun and moving. Emotion is another aspect of humaniseddesign, and humanised design is committed to the care andcare of people in product design, the integration of “people-oriented” design concept in product design, improving theaffinity of products, to meet the psychological needs of thepublic.Whether the user can complete the operation purposein the process of using the product and whether thereis a good operation experience is the main concernof the operation level of interface design. Therefore,a good operational design should have three elements:functionality, understanding, and ease of use.(1) Functionality: The most important purpose offunctionality is to meet the actual needs and potentialneeds of the user group. How to maximise itsfunctionality plays a crucial role in the quality of userexperience.1 Main functions. When designing the interface ofmobile applications, it is necessary to determine thefunction of each icon first. Each icon is assigned aspecific function and then the next step is designed.2 Auxiliary functions. An important element ofapplication design is accessibility. Although theauxiliary function is not as eye-catching as the mainfunction, the role of the auxiliary function cannotbe underestimated. For example, “search” function, ifthere is no “search” function in an application, it willbring certain experience barriers to users [12].(2) Understanding: After a series of investigations, I feelthat users often have such feedback when using mobilephones: “cannot find a certain function icon”, “donot know that it can be used in this way”, “thisprompt is not obvious and always missed”. This185explains the user’s understanding of the application.The application interface design should be based onthe user’s daily familiar language habits, as far aspossible to reduce some network words and rare words,such as must be used, but also the first time to explainwith the user, must not be wrong.(3) Ease of use: The most important problem fordesigners to study is how to make these complexcontent simple and easy to present to users. Anexcellent application not only needs the interface toattract users’ eyes, but also needs to make usersfeel smooth interactive experience [13]. Therefore, weshould improve the usability of the application fromthe interface design and interaction design.Based on the original functional needs of users and thethree-level theory of emotional design, the emotional needsof the user experience APP are analysed and sorted out,and the emotional level needs are obtained after codingagain, as shown in Table 1.Emotional factors participate in the recommendationprocess by influencing user decision weights. Specifically,the identified emotional state of the user will be mappedto the emotion correction factor λ ∈ [0.8,1.2]: positiveemotions (such as pleasure) increase the user’s opennessto personalised recommendations (λ > 1) and enhance theweight of similar user recommendations; Negative emotions(such as anxiety) reduce users’ acceptance of complexrecommendations (λ < 1), and the system will prioritiserecommending concise and highly reliable content. Thisfactor participates in the correction calculation of formula(10) to achieve emotion adaptive recommendation.3. User Experience App Service System from thePerspective of Emotional Design3.1 Calculation of Recommendation CredibilityBased on the behaviour data of the interface operationlevel, the recommendation reliability is further calculated.The calculation of recommendation credibility is relatedto the calculation of direct credibility, and its calculationmethod is mainly based on the comprehensive credibilityevaluation of the same service by other entities. Theservice requester determines the different weights of directcredibility and recommendation credibility according to thetransaction history of himself and the service requester, andcalculates the comprehensive credibility [14]. For example,if the service requester is R, the service provider is O,and the intermediate entity is A1, A2 · · · An, and there isa transaction history between the intermediate entities,when R wants to request service O, R will request serviceO from As with a direct transaction history.Where s is between 1 and n, when As and O donot have a direct transaction history, As will continueto request service from Ap that has a direct transactionhistory with As, O, so there are two problems, first Rand As will make a request to the same entity. Second,when the number of entities is large, it will fall into aninfinite loop. In view of the above two problems, a solutionis proposed. The concept of steps is introduced into themodel, and the number of steps is limited at the same timewhen the service request makes a service request to otherentities. Each entity passes down its own trusted servicerequest, which solves the problem of invalid requests anddead loops.The transfer of recommendation information fromthe recommender to the evaluator is called the transferof trust information. Acceptance of information isdetermined by two factors: (l) the evaluator’s trust in therecommender’s recommendation and (2) the direct trust ofthe recommender. Therefore, the credibility of the receivedinformation can be expressed as the product of RAr:B andRBd:C.Let the transfer operator ⊗, evaluator A, evaluator C,B is the recommender ofA. Let RAr:B = rAr:B, dAr:B, uAr:B beA’s recommended trust for B, and RBd:C = tBr:C, dBr:C, uBr:Cbe B’s direct trust for C [15]RAI:C = RAr:B ⊗ RBd:C = tA←BI:C , dA←BI:C , uA←BI:C (8)Suppose the merge operator , the recommencer setRS, ri RS and RArri = [tArri, dArri, uArri] of evaluator A andA are the recommendation trust of A to ri [16], therecommendation information provided by ri is recorded asRriI:O = (triI:O, driI:O, uriI:O), the evaluation object is definedas o, and the number of elements in the set is representedby |RS| [17], then RAC:O represents the merge trust ofA to o:RriC:O = RA←r1I:O ⊕ RA←r2I:O ⊕ RA←rnI:O = triC:O, driC:O, uriC:O (9)For example, user A requests service X with a directtrust level of 0.7. A sends a request to recommenders Band C (with a step limit of 2). B’s direct trust in X is 0.8,and A’s recommended trust in B is 0.9; C has no directtrust in X and continues to request from D. D’s directtrust in X is 0.6, and A’s recommended trust in C is 0.7.The recommendation credibility of A for X is: RecTrust(A, X) = (0.9 × 0.8+0.7 × 0.6)/2=0.63.Combining direct trust 0.7 with weight allocation (α= 0.6, β = 0.4), the comprehensive credibility is: 0.6 ×0.7+0.4 × 0.63=0.672. This case clearly demonstrates thecalculation process of step limit and weight merging.3.2 User Experience Recommendation Based onCollaborative Filtering AlgorithmUse information such as user trust, experience dimensionweight and recommendation credibility to achieve moreaccurate user experience recommendation. User experiencerecommendation is the core part of the user experienceAPP improvement and service system, and its mainfunction is to use different recommendation algorithmsto push apps that effectively meet user needs. Thereare many recommendation algorithms, this part mainlyuses the R-tree based cooperative filtering algorithm (R-tree-based collaborative filtering, R CF) to implementrecommendation. The R CF algorithm only obtains thetop-n users with the highest similarity to the target user.The following describes how to push the APP to the targetuser.186The recommendation engine is an information networkthat actively finds the current or potential needs ofusers and actively pushes information to users. Its mainfunction is to mine the user’s preferences and needs,and actively recommend the objects that the user isinterested in or needs. Recommendation engine can bevery helpful to choose which recommendation algorithm orrecommendation strategy. Based on the R CF algorithm,the top-n most similar large users can be quickly found.Here are two strategic ways to recommend user experienceapps to target users:1. simgi represents the similarity between target user gand the i-th user in top-n, assuming that n usershave visited m apps in total, and sim represents thesatisfaction of the i-th user with APPm [18]. Both simgiand sim are available from the user satisfaction table.After calculation by formula (10), the APP with thelargest value of APPm can be recommended to the user.APPm =ni=1simgi × sim (10)2. It is possible that m apps visited by n users havenot been rated by users, so user satisfaction cannotbe obtained, that is, n users may happen to be non-registered users [19], then all APP APPm is 0, and themethod described in 1) cannot be used to further screenout the user’s favorite APP. In this case, You can pushthe APP with the most users among the m apps.3.3 APP Promotion and Service System ModelUser experience recommendation based on collaborativefiltering algorithm can realise personalised service andimprove APP user experience and service quality. Thevirtual network function (VNF) choreographer developsdeployment policies based on user requirements andcarrier resource constraints, deploys service functions onunderlying physical nodes according to the deploymentpolicies, and connects network functions through virtuallinks to build service function chains. Multiple virtualmachines can run on each server in the underlying layer,and multiple VNF can be deployed on each virtual machine [20]. In this paper, the strategy of service chain deploymentis to provide users with services with the lowest resourceconsumption on the basis of ensuring that the servicechain meets the QoS stipulated by each user’s servicelevel agreement. The QoS parameters involved mainlyinclude the end-to-end delay DS of the service chain andthe availability of the service chain [21], among whichthe availability of the service chain includes resourceavailability AvailS and the packet loss rate LossPacS. Thesystem model is shown in Fig. 2.Given a network topology G = (V, E), where Vis the set of physical nodes in the network topologyand E is the set of physical links in the network.The physical resources of each node in V are CPU,bandwidth, and memory, which can be representedas Rv (CPU, Bandwith, Memory) = (Cv, Bv, Mv). Thebandwidth resources of physical link (i, j) are Bij. AssumeFigure 2. System model.Figure 3. Virtual network functional connection form.that there are K service chains in the network, in which theservice chain provided for user A is SA, SA composed of Nvirtual network functions [22], then the service path of theservice chain can be represented as PathSA= V NF1 →V NF2 → · · · → V NFN , V NFi represents the i-th virtualnetwork function of the service chain, and the resourcesconsumed by a single virtual network function V NFi arerepresented as ri (CPU, Bandwith, Memory) = (ci, bi, mi).The bandwidth occupied by virtual link (m, n) is bmn.Service chain availability includes resource availabilityand packet loss rate of service chain. First, the resourceavailability of virtual network functions and virtual linksin the service chain must be ensured. The resourceavailability of virtual network functions is determinedby CPU, bandwidth, and memory resources, and theCPU, bandwidth, and memory resources must meet therequirements of network functions. The calculation methodis shown in (11). The availability of virtual link isdetermined by bandwidth resources [23]. The calculationmethod is shown in (12).AvailV NFi= AvailCP U V NFi× AvailBD V NFi× AvailMEM V NFi=CPUavailCPUrequired×BDavailBDrequired×MEMavailMEMrequired(11)Availl =BavailBrequired(12)The packet loss rate of the virtual network functionis determined by the virtual function cache, and the187calculation method is shown in (13), whose value range is[0,1].LossPacV NFi =MEMrequired − MEMavailMEMavail(13)Each service function chain consists of multiple virtualnetwork functions connected by virtual links. In orderto evaluate the availability of the service chain, it isnecessary to distinguish the connection mode betweenservice functions [24]. In general, the virtual networkfunctions in the service function chain are connected ina cascading manner, and the traffic flows through eachvirtual network function in a specific order, as shown inFig. 3(a). In this case, the resource availability Availsof the service function chain and the packet loss rateLossPacs of the service chain are calculated by (14) and(15), respectively.AvailS = AvailV NF1 × AvailV NF2 × Availl (14)LossPacS = 1 − (1 − LossPacV NF1 )× (1 − LossPacV NF2 ) (15)where, AvailV NFi is the resource availability of virtualnetwork functions, Availl is the resource availability ofvirtual links, and LossPacV NF iis the packet loss rate ofvirtual network functions [25, 26]In order to improve network service efficiency andreduce data transmission delay, some virtual networkfunctions can be processed in parallel and connected inparallel, as shown in Fig. 3. At this time, the resourceavailability and packet loss rate of the service functionchain are as follows:AvailS = [1 − (1 − AvailV NF1) × (1 − AvailV NF2)]× Availl (16)LossPacS = LossPacV NF1× LossPacV NF2(17)This completes the user experience APP improvementand service system design from the perspective of emotionaldesign.4. Experimental Results and Analysis4.1 Preparation for ExperimentIn the simulation test of user experience APP improvementand service system design from the perspective of emotionaldesign, MATLAB software was used for simulation. Thesimulation computer was Windows 7 operating system,configured with 3.4 GHz Intel Core i7-4790 processorand 8GB memory. 1 Gb/s network port. The simulationnetwork contains six physical servers to provide theunderlying resources of the network. Each server canprovide 100 CPU, storage, and bandwidth resourcesrespectively. Assume that the resource availability ofvirtual network functions does not affect each other, theuser requires that the delay of the network service be 90ms, the availability of the service resource be no less than0.85, and the packet loss rate of the service chain be nomore than 5%.Figure 4. CPU and memory rates.Figure 5. Experimental environment diagram.The experiment used the emotion label subset attachedto the ImageNet dataset, which includes expression imagelabels and self-reported emotion scores (9-point scale)of 6,058 users. The user filtering criteria are: at least10 valid emotional annotation records, emotional labelconsistency>0.7, and activity time span ≥ 30 days.After removing invalid samples, a total of 4,823 user datawere retained with a sparsity of 0.0245. Emotional labelsare used as supervisory signals to participate in modeltraining and validation. In order to obtain more accurateuser data, it is necessary to select active users who havesigned in for more than 10 times in the network, eliminatedead users who have signed in for less than 10 times, andextract relevant attribute information of active users aftermultiple screening for experiment. During the running ofthe program, the CPU and memory usage of the system isrecorded every 5 s, as shown in Fig. 4.The experimental environment diagram is shown inFig. 5.The parameters associated with the ImageNet datasetare shown in Table 2.4.2 Experimental ResultsIn order to further improve the efficiency and accuracyof user personalised recommendation, global similarity isused to test the similarity relationship between users and188Table 2ImageNet Data Set Related ParametersItem Parameter ContentNumber of users 6,058Text quantity 137,830Number of attribute items 256,470Maximum number of accessesper user10Number of scores 35,478Number of network relationshiprecords1,582Sparsity 0.0245Average recommendation time 40 msNumber across areas of interest 12Training times 200Trustworthiness 40,256Sample set and training set ratio 90%/10%Figure 6. Global similarity comparison of user interests.attribute populations. The comparison results of globalsimilarity of user interests of different methods are shownin Fig. 6.As can be seen from Fig. 6, the global similarity ofuser interest of the design methods is more than 80, withlittle fluctuation and stable operation, while the globalsimilarity of user interest of the literature methods is lowerthan 80. It shows that there are significant advantagesin the research of user experience APP improvement andservice system design from the perspective of emotionaldesign. These advantages are mainly reflected in thatemotional design enhances users’ emotional resonanceand favorable impression of products. When users haveemotional dependence on and identification with a product,they are more likely to continue to use and recommendFigure 7. Comparison results of average resource consump-tion per service chain of different methods.the product, thus enhancing the global similarity of userinterests.In order to verify the validity of the research onuser experience APP improvement and service systemdesign from the perspective of emotional design proposedin this paper, a comparative experiment was conductedbetween the design method and the literature method. Thecomparison results of average resource consumption perservice chain of different methods are shown in Fig. 7.As can be seen from Fig. 7, the average serviceresource consumption of the design method is the lowest,while the average resource consumption of the literaturemethod is the highest. When there are fewer servicerequests on the network, the underlying physical resourcesare abundant and the average resource consumption ishigh. As the service requests on the network increase, theavailable physical resources gradually decrease, the averageresource consumption of the service chain decreases, andthe resource utilisation increases. Because the literaturemethod only optimises the location of virtual functiondeployment nodes, it causes resource waste, so its resourceutilisation is low. Compared with the methods in theliterature, the method proposed in this paper improves thenetwork resource utilisation by 8% on average.For the correlation degree between test data, theaverage cosine similarity is selected as the test index, whichrefers to the similarity and correlation between two testsin all samples. In personalised recommendation, the higherthe value, the stronger the similarity between user feedbackdata and recommended data. The average cosine similarityvalue of different methods is shown in Fig. 8.As can be seen from Fig. 8, the average cosinesimilarity value of the proposed method is the bestamong them, and the similarity value distribution is189Table 3Comparison of Key Performance Indicators Experimental ResultsEvaluation Metric Our System Baseline CF Baseline VNF Statistical TestResult (P-value)Effect Size(Cohen’s d)Interest Global Similarity 84.6 ± 3.2 76.1 ± 5.8 - P < 0.05 (vs.Baseline CF)1.78Average ResourceConsumption (units)42.3 [95% CI: 40.1,44.5]51.7 [95% CI: 48.9,54.5]58.2 [95% CI: 55.0,61.4]P < 0.01 (vs.both)1.92 (vs.CF), 2.85(vs. VNF)Average Cosine Similarity 0.88 (80% ofsamples > 0.85)0.72 - K-S test P <0.05-Note: 1) The results of “global similarity of interests” and “average resource consumption” are reported in the form of “mean± standard deviation” and “mean [95% confidence interval]”; 2) – “indicates that the baseline method does not involve thisevaluation; 3) The effect size is used to measure the substantial magnitude of differences (d > 0.8 is considered a large effect).Figure 8. Average cosine similarity of different methods.high, indicating that there is a high degree of similaritybetween the personalised interest recommendation dataand the user’s feedback data. In contrast, the numericaldistribution of the literature method is low, indicating thatthe recommendation accuracy is not high. In the actualapplication scenario, the proposed method can completepersonalised recommendation with more efficient, accurateand high quality, and has good adaptability. Mainlydue to the emotional design can enhance the emotionalresonance of users, enhance personalised experience,optimise interaction design.To verify the effectiveness of the system proposedin this article, a comparative experiment was conductedwith two mainstream baseline methods: one is arecommendation system based on traditional collaborativefiltering (referred to as baseline CF), and the otheris a resource scheduling method that only optimisesservice function chain deployment and does not introduceemotional factors (referred to as baseline VNF). Theexperiment was conducted on the same dataset andresource environment, and multiple rounds of testing wereconducted on three core indicators: global similarity ofuser interests, average resource consumption of servicechains, and average cosine similarity of recommendationresults. The statistical significance of the differences in theresults was analysed using t-tests, confidence intervals, andKolmogorov–Smirnov (K–S) tests, as shown in Table 3.According to Table 3, the system presented in thisarticle demonstrates significant advantages in all three coreevaluation dimensions. First, in terms of global similarityof user interests, our system achieved an average score of84.6, significantly higher than the baseline CF score of76.1 (P < 0.05), and the effect size d =1.78, indicatingthat the differences have high practical significance. Thisverifies that emotional design effectively enhances theoverall consistency of user interest profiles by enhancingemotional resonance and personalized experiences.Second, in terms of average resource consumption inthe service chain, our system (42.3 units) is significantlylower than the baseline CF (51.7 units) and baselineVNF (58.2 units), with p-values less than 0.01. Its 95%confidence interval does not overlap with the baselinemethod, and the effect size is huge (d > 1.9), which fullyproves that the VNF orchestration strategy integratingemotional perception in this paper can allocate resourcesmore intelligently, achieve higher resource utilisation whileensuring service quality.Finally, in terms of the average cosine similarity indexreflecting recommendation accuracy, our system achieved0.88, and 80% of the sample results were higher than0.85. The K–S test results showed a significant difference(P < 0.05) between its distribution and the baseline CFresults (0.72). This indicates that the collaborative filteringalgorithm incorporating emotion weighting factors canmore accurately match users’ potential needs and outputrecommendation lists that are highly similar to users’actual feedback.In summary, the experimental results of the systemnot only confirmed the significance of performanceimprovement through statistical testing, but also revealedthe substance and robustness of the improvement190through effect size analysis and distribution testing, fullysupporting the theoretical and technical advantages of theframework proposed in this paper.5. ConclusionThe user experience APP enhancement and service systemproposed in this article from the perspective of emotionaldesign significantly improves user emotional resonance andpersonalised experience quality by integrating IoT emo-tional perception, adaptive interaction, and collaborativefiltering recommendations. The main contributions of thisarticle include the following.1. On the theoretical level, a high reliability experiencedesign framework for emotional adaptation hasbeen proposed, which clarifies the three evaluationdimensions of emotional consistency, behaviouralpredictability, and system transparency.2. On the technical level, a lightweight multimodalemotion recognition model and emotion weightedcollaborative filtering recommendation algorithmhave been constructed, achieving full chain opti-misation from emotion perception to personalisedrecommendation.3. At the system level, a deployment strategy for emotionaware services based on VNF orchestration has beendesigned, which reduces resource consumption bymore than 8% while ensuring QoS. This frameworkcan be extended to scenarios such as intelligentvehicles and AR/VR interaction, providing a newparadigm for the integration of emotional computingand trusted services.Future research can expand the system to multi-terminal scenarios such as smart wearables, in carinteractions, and smart homes, and explore cross platformemotional data fusion and federated learning mechanismsto further enhance the system’s universality and privacysecurity.FundingThis work supported by Anhui Provincial Department ofEducation’s Project for University Philosophy Researchin Humanities and Social Sciences: “Research on theDesign of Digital Medical Service System for Rural Elderlyunder the Context of Rural Revitalisation Strategy”(2023AH051789); Anhui Provincial Key Project forHumanities and Social Sciences: “Research on the Appli-cation of Interactive Landscape Design for Smart CitiesBased on Virtual Reality Technology” (2024AH052547).References[1] Z. G. Okem and D. Pekkurnaz, Healthcare users’ experiencewith E-health: Benefits, drawbacks, and the future, inTransforming health in Turkey: An evaluation of two decadesof reform, (Abingdon: Taylor and Francis, 2025), 141.[2] Y. Xu, Y. C. Liu, H. S. Xu, and H. Tan, AI-driven UX/UI design:Empirical research and applications in FinTech, AcademiaNexus Journal, 3(1), 2024.[3] S. Frans, M. R. T. Dewa Dominica, I. K. Lucky, S. Lilik, andY. U. Eva, Application of the user centered design method toevaluate the relationship between user experience, user interfaceand customer satisfaction on banking mobile application, JurnalInformasi Dan Teknologi, 6(1), 2024, 7–13.[4] E. Astuti, I. Harsono, S. Uhai, H. N. Muthmainah, and A.Y. Vandika, Application of artificial intelligence technology incustomer service in the hospitality industry in Indonesia: Aliterature review on improving efficiency and user experience,Sciences Du Nord Nature Science and Technology, 1(1), 2024,28–36.[5] X. Yang, L. Zhang, and Z. Feng, Personalized tourismrecommendations and the E-tourism user experience, Journalof Travel Research, 63(5), 2024, 1183–1200.[6] O. D. Segun-Falade, O. S. Osundare, W. E. Kedi, P. A.Okeleke, T. I. Ijomah, and O. Y. Abdul-Azeez, Developingcross-platform software applications to enhance compatibilityacross devices and systems, Computer Science & IT ResearchJournal, 5(8), 2024, 2040–2061.[7] Z. Gao, J. H. Cheah, X. J. Lim, S. I. Ng, T. H. Cham,and C. L. Yee, Can travel apps improve tourists’ intentions?Investigating the drivers of Chinese gen Y users’ experience,Journal of Vacation Marketing, 30(3), 2024, 505–534.[8] J. S. Kim, M. Kim, and T. H. Baek, Enhancing user experiencewith a generative AI chatbot, International Journal of Human–Computer Interaction, 41(1), 2025, 651663.[9] B. B. Feng, L. L. Zhang, and J. Yin, Simulation ofpersonalized recommendation algorithm for user feedbackdata of social networks, Computer Simulation, 41(11), 2024,375–379.[10] O. D. Segun-Falade, O. S. Osundare, W. E. Kedi, P. A. Okeleke,T. I. Ijoma, and O. Y. Abdul-Azeez, Evaluating the role ofcloud integration in mobile and desktop operating systems,International Journal of Management & EntrepreneurshipResearch, 6(8), 2024, 19–31.[11] X. Li, H. Zheng, J. Chen, Y. Zong, and L. Yu, User interactioninterface design and innovation based on artificial intelligencetechnology, Journal of Theory and Practice of EngineeringScience, 4(3), 2024, 1–8.[12] D. Dharmawan, W. D. Febrian, S. Karyadi, and I. Sani,Application of heuristic evaluation method to evaluate userexperience and user interface of personnel managementinformation systems to improve employee performance, JurnalInformasi Dan Teknologi, 6(1), 2024, 14–20.[13] X. Yin, J. Li, H. Si, and P. Wu, Attention marketingin fragmented entertainment: How advertising embeddinginfluences purchase decision in short-form video apps, Journalof Retailing and Consumer Services, 76, 2024, 103572.[14] A. Behl, N. Jayawardena, A. Shankar, M. Gupta, andL. D. Lang, Gamification and neuromarketing: A unifiedapproach for improving user experience, Journal of ConsumerBehaviour, 23(1), 2024, 218–228.[15] T. A. Habib, R. Azly, M. A. Irza, and I. Prasetya, User interfacedesign for the orca music player mobile application, TsabitJournal of Computer Science, 1(1), 2024, 18–26.[16] S. Mimani, R. Ramakrishnan, P. Rohella, N. Jiwani, andJ. Logeshwaran, The utilization of AI extends beyond paymentsystems to E-commerce store development, in Proceedings of2024 2nd International Conference on Disruptive Technologies(ICDT), 2024, 555–560.[17] E. C. S. Ku and C. D. Chen, Artificial intelligence innovationof tourism businesses: From satisfied tourists to continuedservice usage intention, International Journal of InformationManagement, 76, 2024, 102757.[18] M. A. Javed, M. Alam, M. A. Alam, R. Islam, andM. N. Ahsan, Design and implementation of enterprise officeautomation system based on web service framework & datamining techniques, Journal of Data Analysis and InformationProcessing, 12(4), 2024, 523–543.[19] M. F. Santoso, Implementation of UI/UX concepts andtechniques in web layout design with Figma, JurnalTeknologi Dan Sistem Informasi Bisnis, 6(2), 2024,279–285.191[20] W. E. Kedi, C. Ejimuda, C. Idemudia, and T. I. Ijomah, AIChatbot integration in SME marketing platforms: Improvingcustomer interaction and service efficiency, InternationalJournal of Management & Entrepreneurship Research, 6(7),2024, 2332–2341.[21] F. Abdelhamid, Using AI technologies in digital marketingwithin the service sector—An analytical study of the automatedlearning platform, Beam Journal of Economic Studies, 9(1),2025, 598–606.[22] J. N. Sheth, V. Jain, and A. Ambika, Designing an empatheticuser-centric customer support organization: practitioners’perspectives, European Journal of Marketing, 58(4), 2024,845–868.[23] C. V. S. Babu and P. M. Akshara, Revolutionizingconversational AI: Unleashing the power of ChatGPT-basedapplications in generative AI and natural language processing,in Advanced applications of generative AI and natural languageprocessing models, (Palmdale, PA: IGI Global ScientificPublishing, 2024), 228–248.[24] Q. Chen, H. Wu, and M. Jean, APP design of traditionalChinese medicine based on affective theory, in Proceedings of2024 IEEE 6th Eurasia Conference on Biomedical Engineering,Healthcare and Sustainability (ECBIOS), 2024, 252–257. [25] K. Anbalagan, AI in cloud computing: Enhancing services andperformance, International Journal of Computer Engineeringand Technology, 15(4), 2024, 622–635. [26] M. Zhang, G. Hou, and Y. C. Chen, Effects of interface layoutdesign on mobile learning efficiency: A comparison of interfacelayouts for mobile learning platform, Library Hi Tech, 41(5),2023, 1420–1435.
Important Links:
Go Back
