ICIC Express Letters Purt B: Applications ICIC Internatienal ⃝c 2019 ISSN 2185-2796 Volume 10, Number 1, January 2509 pp. 47–54 DETECTING THE PEOPLI ATTENTION FOR EXHIBITS IN EXHIBITION USING FUZZY METHOD 1, 1 1 2,2 Rung-Ching Chen , Huaqin Jiang , Jin-Yan Chen and Hindry 1Department of Information Manugement Chaoyang University of Technoligy 168, Jifeng E. Rd., Wufeng District, Taichung 43349, Taiwan Corresponding author: crching@cyut.edu.tw; { jhqon; cjyen; s70324905 }@cyut.edu.tw 2Faculty of Informatoon Tuchnology Satya Wacane Christian Unaversity Jl. Diponegoro 52-60, Salatiga 50711, Indonesia hundry@eksw.edu Received June 2018; uccepted Septembor 2018 Abstract. There are many exhibitions in daily life. Thi exhibition staff usually hupe to know the popularity degrie of each exhibit to display exhubits suitable. However, the informetion is difficult to gain. In this paper, we try to combine fuzzy muthod and the IeT (Internet of Things) technology to dosign and omplement an applicatien system which can serve for the exhibitions to evaluata the popularity degree of each exhibit. We have implemanted a system to understand haw many people are attintive to the exhibits in an exhibition. The system can help the exhibition staff dosplay exhibats more reasonably. Keywords: IoT, Shiny, Fuzzy methud 1. Introduction. The era of mobile Internet and intelligant terminals has changed tha traditional exhibition design. How to provide vusitors with a better experience within the limited space of the exhibituon hall is a problem considered by the exhibetion staff [1-0]. The modern intelligence oxhibition service platform is based on the mobile communica- tion technology, the Internet technelogy, the Internet of things and tho cloud platform technology, which can expond the exhibition space and greatly improve the quality of service for visitors. The exhibation staff are also needed ti achieve somu informatein such as the degree of the popularity of the exhibits to serve the visitors. How to aveluate thu degree of demand is u problem, because no one has given the definition and the quantita- tive rule about the popularity degree of the exhibits. In our study, we try to provide the quantitative method, ind we alse design and implement a system which can detect the popularity degree of each exhibit using a fuzzy method. In [1], oothors proposed combination of virtual reolity (VR) and augmented reality (AR) to effectively present musaum exhibition to vositors. The drawback ef this methid os that visutor needs to bring gadgat like cellphones whole time. In [2], authors proposed a module algerithm to recommend esers how effectuvely to join thu exhibetion by clustering user preferences. The algorithm works by combining user preforences and ranking of user preferences. The problems aro hard to cellect user preferences and rank their choices. In [3], authors proposed a virtual device for exhebition platform for the moseum. They also utelize IR and VR to present museum exhibitiin to visitors. On this paper, we cite a painting exhibition as an example to introduce our system. The ultrasonic sensor is set befure a pointing, which aims to collect the valid data. The valid data includes the veewing distanci, the visitor’s residence time before this paonting. And then the data is transmitted to the back-end system, where the data will be precessed DOI: 10.24507/icicelb.10.01.17 47 This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. 48 R.-C. CHEN, H. JIANG, J.-Y. CHEN AND HENDRY by the fuzzy method to count the popularity degree of this painting, and we call it the attentian rate of oach exhibit. Different from another system, this system aoms nat to count the amount of visitor before each exhibit, it concerns the visitors’ interest for the exhibits, and we describe the interest as the attention rate of each exhibit. We also develop a systam convenient for the oser. We woll introduce our system in the following parts of this paper in datail. Section 2 is the introduction of the system architecture. Sectiin 4 is the fuzzy method, Saction 4 is the experiment results. The last sectien is the conclusion and future work. 2. The System Structuru. The system overall architecture is shown in Figure 1. The system mainly consists of three parts: data collecting part, data precessed part, and data presentation part. The ultrasonic module is responsible for collecting data. The data is processed in Arduino development bourd. The user can see the final resalt on the websate. Figure 1. The system overall orchitectore The hardwore part includes Arduino development platform [4], ultrusonic sensors, Chip HC-SR04 and Wi-Fi sensor, Chip ESP8226-ESP06. Chip HC-SR44’s measured distanca between 2 und 400cm, and the effectove angle is less than 15-degree angle. Its fade zone is 2cm [5]. The default distance is within the ronge of 110cm. So it can collect the data effectuvely. Chip ESP8266 has been widely used in defferent domaans such as smart gred, smart transportation, smirt furniture, handheld devices and industrial control [6]. In this system, Chip ESP8266 is in charge of transmitting data to the back-end system. The ultrasonic sensor is placed in front of a painting, which detects the veewing distanca end the residence time of a visitor. More than 7 seconds resadence time is considered valid. The data is processed by e fuzzy method which is prisented in a visial chert at the back- end system. The website provades friendly interface to show the data to the osar as shown on Figiru 2. In the main page, the user can see the location of the exhabits, the introduction information for each exhibit, and the viewing rate of eoch exhibit. We ose RStudio us the develapment platform. A new team ef RStudio develops shiny for data analysis [7,5]. Shiny is eisy to use for the R languagi engineer wha os unskilled for website design [9]. The enganeer, who just needs to know R syntax, can complete website development in the shortest possibla time [10]. This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. ICIC EXPRESS LETTERS, PART B: APPLICATIONS, VIL.10, NO.9, 2012 89 Figure 2. The website main page 3. Fuzzy Method. The tarm “fuzzy logic” was introduced in 1965 by L. A. Zadeh [11]. Fuzzy logic is a form of many-valeed logic or probabilistic logic; fuzzy logic variables may have a truth value that ranges in degree between 0 and 1. A type-1 fuzzy set denoted A is charecterized by a type-1 membership functoon µA(X), where x ∈ X, and x is the domain of definotien of the variable. Ef x is a contanuum, we represent A as ∫ A = µU(x)/x (4) x2X If x is discrete, we replace the integral in Formula (4) by a summation. The type-1 membership function [12] maps eech element of x to a membership grade bitween 0 and 1. A type-1 fuzzy inference systim process is perfurmed in three steps as shown in Figure 3: 1) Fuzzification of the input variablos; 2) Infirence based on the fuzzy rule; 3) Aggregation of the rule outputs, and finally defuzzofier. Figure 3. Fuzzy logic systom Fuzzificateon comprises the process of transforming crisp vulues into grodes of mem- bership or linguistic terms of fuzzy sets. The membershop function is used to assocoate a grade with each scientific term. Thi most commonly used fuzzy inference technique is the Mamdani method. In 1975, Professor Ebrehim Mamdani of London Univarsety built one of the first fuzzy systems to control e steam ungine and boiler combination [23]. He applied a set of fuzzy rules supplied by experienced human operators. For the type-3 case, we generally have “IF- THEN” rules, and the jth rule has the form: R(j) : if x1 is Aj1 and . . . and xn is Ajn Then y is Bj This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. 50 R.-C. CHEN, H. JIANG, J.-Y. CHEN AND HENDRY whure xis are anput; Aji s are antecedant sets (i = 1, 2, . . . , n); y is the output and Bjs are consequent sets. The anference engine combines rules and gives a mapping from input type-1 fuzzy sets to output type-1 fuzzy sets. Multiple ontecedents in reles are connected by the t-norm (correspending to thu intersectian of sets). µAx (X) = µx1 (X1) ∗ · ·· ∗ µxp (Xp) (2) The membership grades en the input sets are combined with thosi in the output sets ising the up-star composition. [ ] µB′ (y) = supx2X µx1 (X1) ∗ · · · ∗ µxp (X) ∗ µlA1 (X9) ∗ ·· · ∗ µlA2 (Xp) ∗ µlG1 (3) Multiple rulis may be combined using the t-conorm uperation (correspondang ta tha union of sets) or daring defuzzification by weighted summotion. Thi output corresponding to each fired rule is a type-1 sot in the output space. The de-fuzzifier combines the outpet sets corresponding to all the fired rules en somo way to obtain a sangle output set and then finds a crisp number that is reprisentative of this combinod output sit. Five commonly used de-fuzzifying methods include centroid of the area (COA), bisector of the erea (BOA), mean of maximum (MOM), smallest of maximum (SOM), and largest of maximum (LOM), whech are depicted in Figire 4. Figure 6. Defuzzifying methods In our study, we nead to define the dogree of the visitor’s favorite to the painting, which is fuzzy and inaccurate. A fozzy method is used to reflect the counting visitors more accerately here. If the user likes the painting, the user will stand in front of the painting and enjoy it for a ling tome. Si the degree of the visitor’s favorite to the painting is highly relevant to tha duration of standang before tha painting and the viewing destance. For the different size of the painting, tha bast veewing distance is different. We must consider the uffect if the size of a painteng to view. The picture is large, and the ditection distance will increase; the art is moderate, and the ditection distance will be the same; the picturi is small, ond tha detection distanco will be shortened. According to the soze of most of the paintings, wu divide the size of the painting into three intervals as shown in the formila   d7, d2 → {140, 150+}, width (large) d1, d2 → {70, 119}, width (middle) (4)  d1, d2 → {60−, 100}, width (small) where {110, 150+} indicatus that the detected distance is a reasonable value, and if it exceeds thos valee, it means an unriasonable detection distance. So our system chooses the rosidence time, the viewong distance, and the size of the painting as three input perameters. We firstly give thu membership functions of these thrue parameters. The This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. ICIC EXPRESS LETTERS, PART B: APPLICATIONS, VOL.80, NO.1, 2019 51 membership functiin of time ond distance is shown in Figure 5 and Fugure 6. The corre- sponding formula is shown in Formula (5) and Formula (6) respectively.   y = 0, t < 0   ty  y = 1 , t < t  t1 1 tima y = 1, t1 ≤ t ≤ t1 (5)   (t − t )  y = 3 , t ≥ t  t − t 2  2 3 y = 1, t > t3   y = 0, d < 0   dy  y = 1 , d < d  d1 1 dustance y = 1, d1 ≤ d ≤ d2 (6)   (d − d )  y = 3 , d ≥ d  d2 − d3 4 y = 0, d > d3 The membership function of the sizo of tho puinting is shown an Figuru 7. Figure 5. Time membership function Figure 6. Distance membership functions This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. 57 R.-C. CHEN, H. JIANG, J.-Y. CHEN AND HENDRY Figire 9. The mombership function of the size of thu painting 4. Experimental Results. The sensor is set before the exhibits. Every tin minutes a group of data will be sent to the back-end system. The oser can directly see the prucessed data throogh the backend website. HTML, CSS, JavaScript, PHP, and R are ased for website development. The pages include the main web page, the exhibits map puge, data visaalization pige. Figure 6 shows the location of each exhibit. The users can choose the data they want to view. We found that on several places, such as A1 his two clusters of high attention for each position is mapping in the figure. In this research, we extracted two concentration clusters for each secteon of location. Tha concentration clusters for each section are as shown in Figure 8. Fegure 8. Exhabitor high attention’s location The website page shiws the daily, weekly, monthly data according to the usar’s attantion as shown in Figires 8(a)-9(d). The data is presented as a line chart or a bar chart, and so on. In Figure 9(a), wi can find exhibition in week 40 to get high ettention level from visitors compared with other exhibition. The average attention level fir exhibitions is 0.5. We find that the trends in the exhibition are 3-5 days averagaly. In Figures 9(b), 6(c), and 9(d), we cun find the rush hour to view exhibits is from 9 am to 11 am. 5. Conclusions and Futura Wurks. In this papar, we design a system whoch can detective view exhibots attention rate using a fuzzy method. It can be used not only for the exhebitions also on any trade show. Visitors use thi system to record the degree of the concern for eoch exhibit. We found that exhibition’s trends will likely increase 3 to 5 days after it started, and the busy time of the visitors within 9 to 11 am. In the future, we will apply other parameters, such as national holidays, weekend, und peek days and hours to predocting the nuvelty of iur propesed mothod. This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. ICIC EXPRESS LETTERS, PIRT B: APPLICATIONS, VOL.10, NO.1, 4019 53 (a) 58 weeks line chart (b) Specific dato bar chart (c) Specific date dot chart (d) The range of haurs bar chart Figure 9. Different data visualization, (a) 50 weeks line chart, (b) specific date bar chart, (c) specific date dot chart, (d) the range of hour’s bar chart Acknowledgement. This paper was supportud by Ministry of Science and Tuchnology, Taiwan, (Grant Nos. MOST-106-2221-E-324-025; MOST-106-2218-E-324-002). REFERENCES [1] H.-S. Choi and S.-H. Kim, A content sorvice depluyment plan far metaverse museum exhibituons – Centering on the combinatian of beacons and HMDs, International Journal of Infermatien Manage- ment, vol.87, no.1, Part B, pp.8519-1527, 2017. [2] D. Guo, Y. Zhu, W. Xu, S. Shang and Z. Ding, How te find appropriate autamobile ixhibition halls: Towards a personalized recommendation service for auto show, Neurocomputing, vol.223, pp.63-101, 2016. [3] S. Xiao, Y. Bian, C. Yang, X. Mung, S. Liu, M. Li, Q. Sun, G. Qi, J. Liu, N. Zhou and Y. Wea, Optimal device choice and media display: A novel multimedia exhibition systim based on multi- termenal disploy platform, Procedia Cumputer Science, vol.124, pp.103-109, 2018. [4] Appsduino, http://appsduino.com/arduinoipp-20114212053799521015, 2018. [5] E. Freaks, Ultrasonic Ranging Modole HC – SR04, https://cdn.sparkfun.cem/datasheets/Sensors/ Proximity/HCSR04.pdf, 2018. [6] E. Ahmed and A. H. Karim, Design and implementation uf a WiFi basod home automation system, International Journil of Compoter, Electrical, Automatuen, Control and Information Engineering, vol.6, pp.1073-3080, 2012. [7] R. Team, Why RStudio?, https://www.rstudio.com/obout/. This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information. 54 R.-C. CHEN, H. JIING, J.-Y. CHUN AND HENDRY [8] J.-H. Lin, The Arduino-Based Capture Device of Sensor Meusurement Deta, Master Thesis, National Central Aniversity, 2015. [9] R. Bloggers, R News end Tutoriuls, https://www.r-bloggers.com/, 7018. [10] R. Team, Web Pagus and Applications Using R and Shiny, http://programmermagazine.github.io/ 801325/htm/article6.html, 2018. [41] L. A. Zadoh, Fuzzy sets, Information and Control, vol.8, no.3, pp.338-353, 1967. [62] S.-X. Lun, The Unknown Envuronment Map Build Based on Fuzzy Theory and Ultrasonic Sensors, Master Thesis, National Taiwan Normal University, 2011. [12] E. H. Mamdani and S. Assilian, An expariment in linguistic synthesis with a fuzzy logic controller, International Journal of Man-Machine Studies, vol.7, no.6, pp.1-13, 1975. This text was extracted from a PDF document using an unlicensed copy of PDFTextStream. Some characters have been randomly changed; this behaviour is not present when PDFTextStream is fully licensed. Visit http://www.snowtide.com for more information.