You are on page 1of 4

2017 Fifth International Conference on Information and Communication Technology (ICoICT)

Worth Eat: an Intelligent Application for Restaurant


Recommendation based on Customer Preference
(Case Study: Five Types of Restaurant in Tangerang
Selatan Region, Indonesia)

Ditdit Nugeraha Utama, Luqman Isyraqi Lazuardi, Hersy Ayu Qadrya, Bella Marisela Caroline, Tris Renanda,
Atthiya Prima Sari
Laboratory Optimization Models and Systems for Decision Support, Information Systems Department,
UIN Syarif Hidayatullah
Jakarta, Indonesia
dit.utama@uinjkt.ac.id

Abstract—In a metropolitan, to find a restaurant in rush hour the systems. And also, [3] implemented the concept of
is a challenging activity to do. It is going to be easy, when intelligence in database application.
someone accesses an application and it recommend the best
restaurant then he / she can visit to. Through using combination In addition, in transportation field, [4] have recommended
between methods fuzzy-logic and bubble-sort that were realized an intelligent applications for testing the roadway compaction
in customer preference case, an intelligent application for by measuring 100% of a roadway. The applications have
restaurant recommendation was technologically constructed. The capacity to enhance a compaction uniformity of soils and
method fuzzy logic used to parameterize three selected asphalt pavements. [5] studied concerning the estimation of
parameters; interest, location, and rating. And, the bubble-sort is safety-related cooperative intelligent transportation systems in
a method to optimize the recommendation based on full-factorial terms of the expected change in accident occurrence and
concept. Tangerang Selatan (Indonesia) area and twelve consequence probability. They developed a tool to do so. In
restaurants indicating five types of restaurant were practically energy and building field, [6] constructed an intelligent
chosen as a case study of this research, and seventy seven building application called building management systems
respondents were realistically involved to justify the selected (BMS) through operating sensor technology and
parameters. instrumentation for collecting high quality data. The systems
provide an opportunity energy efficiency of building use.
Keywords— intelligent application; restaurant recommendation;
customer preference; bubble-sort; fuzzy-logic The paper presents a constructed application for
recommending the fittest restaurant for customer based on
I. INTRODUCTION customer’s preference. The introduction section is followed by
For a busy people, to find a restaurant in peak time is a others: related works, research methodology, results and
challenging activity to conduct. A traffic congestion and tight discussion, and conclusion and further works. And then, the
time are two of several conditions that have to be faced area surrounding Tangerang Selatan (Indonesia) has been taken
practically in finding the restaurant. Moreover, concerning as a case study of the research.
food, it is about subjectively and individually taste and passion.
For tackling this situation, an intelligent application that is able II. RELATED WORKS
to recommend a proper restaurant is necessitated then.
Studies relating to “restaurant” as an object and topic of
The studies concerning an intelligent application have been research have been researched by many researchers. [7] talked
already studied by several researchers. [1] have answered the about revenue management of restaurant. To analyze it, he
lack of development for intelligent application in mobile proposed a new matrix of performance indicators and applied it
devices. They technically offered a development environment to see effectiveness measurement of restaurant strategy. [8]
for intelligent application for mobile devices. It was able to developed a model for decision support. The model is able to
address two lacks of development; mobile device limitation be used by tourist for making decision regarding restaurant
(e.g. memory and processing power) and diversity in the selection by using social information (specifically for the
mobile device market. [2] studied regarding intelligent systems TripAdvisor.com users). [9] assessed the effect of customers’
based on semantic web services (SWS). They discussed several perceived authenticity and emotions on purchase purposes. The
empirical issues related to the development process of such results exposed that unfamiliar food names and ingredients
systems and also proposed a systematic way for constructing

ISBN: 978-1-5090-4911-0 (c) 2017 IEEE


increase customers’ perceptions of authenticity and include procedure bubblesort(a as array)
positive emotions expressively. for i from 1 to N
for j from 0 to N - 1
Furthermore, [10] examined the impact of the program on
if a[j] > a[j + 1]
differences in consumer awareness and use of nutritional
information in restaurant. It assessed participant noticing of swap(a[j], a[j + 1])
nutrition information, influence of nutrition information on endif
menu selection, and nutrient intake. The research was endfor
conducted in Canada. [11] observed a higher-order structural endfor
model exploring business innovation, owners’ end procedure
entrepreneurship, and human capital. The findings improved ________________________________________
Code 1. Pseudocode of Bubble Sort
theories about restaurant entrepreneurship and performance and
presented the important consequences for industry to expand a
successful and sustainable restaurant sector. And [12] IV. RESULT AND DISCUSSION
investigated restaurant employees’ food allergy (FA)
knowledge and identified FA training. This study produced a A. The Constructed Application
greater understanding of them that can be gained by restaurant Usecase diagram (Fig. 1.) depicts a communication pattern
industry. between actor and system. Here, there are two human actors
identified; Customer and Restaurant. The actor
III. RESEARCH METHODOLOGY Customer interconnects to four types of usecase (activity
Customer judgment based fuzzy process was applied to group); Registering, Getting Recommendation,
parameterize three selected parameters. Here, seventy seven Ordering, and Rating. Indeed, the key part of the
customers’ justifications were technically recorded and application is in usecase Getting Recommendation.
transformed then to become parameter coefficients by using Here, through this usecase, the application gives the best
fuzzy-logic concept [13]. The three selected parameters (i.e. recommendation for customer. While, the actor Restaurant
distance, rating, and interest) represent three characteristics mainly interacts to the application for managing content, it is
illustrating recommender systems; location based, content facilitated via the usecase Managing Content.
based, and collaborative filtering [14] respectively; where each
parameter illustrates geographic convenience [15], users
attention similarity [16], and product interest discovery [17]. Registering

And the customer’s interest of restaurant category is ideally


indicating the term “customer preference”. Customer

Getting Recommendation
To analyze and design the constructed application, the
Restaurant
method object oriented [18] was used. Three diagrams usecase,
class, and sequence were scientifically provided in this paper. Ordering
They describe a communication pattern between system and Managing Content
actor, an interconnection among entities, and step by step Rating
action that interconnect all involved objects respectively.
Moreover, for recommending the fittest restaurant for Fig. 1. The usecase diagram of constructed application
customer, the algorithm bubble-sort [19] was operated (it is
represented in Code 1). The fittest value is obtained based on The configuration of interconnections among entities is
the calculation of the similarity among the parameters belong visibly depicted in the class diagram (Fig. 2.). It is a high level
to customer and restaurant. The calculation is coming from class diagram that is simply configured without attributes and
[20]; where the equations (1) and (2) were used to get operation. Mainly, the application has four classes;
respectively maximum and minimum relative value of Restaurant, Customer, Interest, and
similarity. Here, SMax represents the maximum similarity MembershipFunction. The class Restaurant has other
relative value describing when the biggest value is the most four aggregated classes; Category,
optimal one (for both parameters restaurant interest and rating); RestaurantLocation, Rate, and Recommendation.
and SMin symbolizes the minimum similarity relative value While, the class Customer has one aggregated class
depicting when the lowest value is the most optimal one (for (CustomerLocation), where the cardinality 1 and 1..*
parameter distance).
between classes Customer and CustomerLocation
theoretically symbolizes that the customer has mobile position.
SMax = ValueCurrent / ValueMax (1) In addition, bubble-sort method that is technically operated
to grade the recommendation is implicitly identified in the class
Recommendation as an operation. Here, the class
SMin = ValueMin / ValueCurrent (2) Recommendation communicates with the class
MembershipFunction to determine a set of fuzzy
language. The fuzzy membership function itself (with three
types of fuzzy language; low, middle and high) is depicted in Moreover, the Fig. 5. shows one screenshot sample of the
Fig. 3. Here, all grading values (input crisp value) from seventy constructed application. Here, the application directly
seven respondents for scoring each selected parameter were recommend the fittest restaurant after a customer access the
technically converted to become crisp output value (via fuzzy application. And then, in a straight line, the customer is able to
logic concept). order a menu and select the sit place that customer want to
select.
Interest Category

0..1 0..1
V. DISCUSSION
1 1

CustomerLocation Customer Restaurant 1 1 RestaurantLocation [8] developed a decision support model by using social
1..* 1 1 0..*
1
information for tourist. They used two main methods; fuzzy
1
sets and Bonferroni mean to denote online reviews and
1

MembershipFunction
1
Recommendation Rate
consider interdependence among criteria respectively. Their
1..* 1 model also conducts a similarity measurement to handle sparse
data in fuzzy environment. Here, similarly we used method
Fig. 2. The class diagram of constructed application fuzzy logic; however, we used it to parameterize the
parameters that humanly evaluated by seventy seven
customers. For similarity measurement, we operated an optimal
relative value that we realized for each selected parameter. And
the highest value was assumed as the most similar (the fittest).
We used the other optimization method bubble-sort as well to
prioritize and select the best decision alternative.

VI. CONCLUSION AND FURTHER WORKS


An intelligent recommender application was methodically
created. It operates three types of parameter (distance, rating,
and interest) to look for a similarity value (between customer
Fig. 3. Triangular fuzzy membership function and restaurant). One of selected parameters (parameter interest)
is hypothetically characterizing the “customer preference”.
Sequentially, the usecase Getting Recommendation Additionally, the application is able to recommend the fittest
is done based on sequence diagram in Fig. 4. To conduct the restaurant for customer who is looking for a restaurant to
usecase, seven objects are involved. Those objects are used to select. A combination between methods fuzzy-logic and
update the information into or read the information from. The bubble-sort were officially adopted as main concept in
final message (information) coming from this usecase is a constructing the application.
restaurant recommendation that customer is able to choose.
The other parameters have still opportunity to be
Customer CustomerLocation Interest RestaurantLocation Rate Category Reccomendation considered, to improve the constructed application;
particularly, the parameters connecting to economic aspect, e.g.
1 : updating location() budget, product price, etc. The parameters describing a
customer’s eating behavior is vastly possible to be taken into
2 : Selecting Interest()
account as well, such as eating time behavior, personal
3 : ReadLocation() experience of eating, etc. Also with parameters relating to
environmental attribute are interesting to be considered, e.g. the
4 : Rating()
quality of restaurant’s waste management, environment
5 : Read Category()
awareness, etc.
6 : Getting Recommendation()

7 : Recommendation()
REFERENCES
[1] L. Hall, A. Gordon, L. Newall, and R. James, “A development
Fig. 4. The sequence diagram of the usecase Getting environment for intelligent applications on mobile devices,” Expert
Recommendation Systems and Applications, vol. 27 no. 3, pp. 481-492, 2004.
[2] H. Barros, A. Silva, E. Costa, I.I. Bittencourt, O. Holanda, and L. Sales,
“Steps, techniques, and technologies for the development of intelligent
applications based on semantic web services: A case study in e-learning
systems,” Engineering Applications of Artificial Intelligence, vol. 24 no.
8, pp. 1355-1367, 2011.
[3] B. Ruparel, “Designing and implementing an intelligent database
application: A case study,” Expert Systems with Applications, vol. 3 no.
4, pp. 411-430, 1991.
[4] C.M. Savan, K.W. Ng, and K. Ksaibati, “Benefit-cost analysis and
Fig. 5. Screenshot of recommendation and order pages application of intelligent compaction for transportation,” Transportation
Geotechnics, vol. 9, pp. 57-68, 2016.
[5] U.C. Ehlers, E.O. Ryeng, E. McCormack, F. Khan, and S. Ehlers, [13] L.A. Zadeh, “Fuzzy logic = computing with words,” IEEE Transactions
“Assessing the safety effects of cooperative intelligent transportation of Fuzzy Systems, vol.4 no.2, pp. 103-111, 1996.
systems: A bowtie analysis approach,” Accident Analysis and [14] R. Burke, Hybrid web recommender systems, in P. Brusilovsky,
Prevention, vol. 99 no. A, pp. 125-141, 2017. A.Kobsa, and W. Nejdl (Eds.): The adaptive Web, Berlin Heidelberg:
[6] H.R. Shaker and S.L. Molnar, “A new data-driven controllability Springer-Verlag, 2007.
measure with application in intelligent buildings,” Energy and [15] Y.M. Li, C.L. Chou, and L.F. Lin, “A social recommender mechanism
Buildings, vol. 138, pp. 526-529, 2017. for location-based group commerce,” Information Sciences, vol. 274, pp.
[7] C.Z. Heo, “New performance indicators for restaurant revenue 125-142, 2014.
management: ProPASH and ProPASM,” International Journal of [16] L. Boratto, S. Carta, G. Fenu, and R. Saia, “Semantics-aware content-
Hospitality Management, vol. 61, pp. 1-3, 2017. based recommender systems: Design and architecture guidelines,” in
[8] H. Zhang, P. Ji, J. Wang, and X. Chen, “A novel decision support model Press.
for satisfactory restaurant utilizing social information: A case study of [17] E.Q. Silva, C.G.C Junior, L.M.L Pascoal, T.C. Rosa, “An evolutionary
TripAdvisor.com,” Tourism Management, vol. 59, pp. 281-297, 2017. approach for combining results of recommender systems techniques
[9] J.H. Kim, H. Youn, and Y. Rao, “Customer responses to food-related based on collaborative filtering,” Expert Systems with Applications, vol.
attributes in ethnic restaurant,” International Journal of Hospitality 53, pp. 204-218, 2016.
Management, vol. 61, pp. 129-139, 2017. [18] L. Mathiassen, A. Munk-Madsen, P.A. Nielsen, and J. Stage, Object-
[10] C.M. White, H.G. Lillico, L. Vanderlee, and D. Hammond, “A voluntary Oriented Analysis and Design, Marko Publishing ApS, Aalborg
nutrition labeling program in restaurants: Consumer awareness, use of Denmark, 2000.
nutrition information, and food selection,” Preventive Medicine Reports, [19] D. Knuth, The Art of Computer Programming: Sorting and Searching, 2
vol. 4, pp. 474-480, 2016. ed., vol. 3, Addison-Wesley, 1998.
[11] C. Lee, R. Hallak, and S.R. Sardeshmukh, “Innovation, [20] D.N. Utama, M.D. Saputra, L.N. Wafiroh, M.A.A. Putra, and P. Lestari,
entrepreneurship, and restaurant performance: A higher-order structural “F-multicriteria based decision support system for road repair and
model,” Tourism Management, vol. 53, pp.215-228, 2016. maintenance (case study: three areas in Tangerang Selatan, Province
[12] Y.M. Lee and E. Sozen, “Food allergy knowledge and training among Banten, Indonesia),” International Journal of Management and Applied
restaurant employees,” International Journal of Hospitality Management, Science, vol. 2 no. 10, pp.171-175, 2016.
vol. 57, pp. 52-59, 2016.

You might also like