This work presents a method to monitor students in a campus area for headcount in an area and activity of student. For high accuracy and real-time data, RFID readers are exploited and located at the fronts of every door of a room to identify students in the area. Ontology was created to represent room type and activity that can happen in a room. With the ontology, we can infer activity of students using an inference engine embedded in OAM framework based on RFID data and basic data from university. To make data easier for campus personnel in management, visualization systems are implemented. These are a visualization to display crowd in a campus area and a visualization tool for student activity.
A campus is where university or college buildings are situated. In campus, people are
most crowded in office days for many reasons such as doing educational activities,
group meeting, skill training, entertaining activities, etc. To make a smooth running,
management within campus is greatly required. The management in campus is such as
facility provision, security, and transportation service. In fact, making a good
management requires a lot of information [
In the past, some researches proposed on gaining the headcount and student
movement data. [
However, none of the above mentioned researches mentioned on student activities in the presence although they obtained students’ location. Students’ activities are one of the most required data for managing facilities and safety. In this work, our goal is to acquire students’ activities using identification from RFID with ontology based inference. To combine data of who, where and when, we can semantically infer activities of students using knowledge representing an ontology, inference engine and real-time data. 2
There are some existing researches focusing on activity recognition and headcount in campus detection. We review their works and summarize them in this section. Moreover, we briefly explain an Ontology Application Management (OAM) framework that we use as a core in inference an activity in this work.
Recently, a work on student movement and headcount detection [
Another publication is about activity recognition using ontology with context in a home [6]. This work applied object based sensors to read human action in a home. With the gained sensor data, inference engine was utilized to reason and result in guessing an activity of a person who interacts with the object. This work showed a good success on using logic and reasoning with ontology in recognizing activity in home. Unfortunately, home activities are limited to objects existing in home in which are different from other places. They cannot directly apply to another location. Moreover, the work focused on a single user in a home while an amount of persons in large place such as university campus could apparently be over a hundred participations.
Ontology Application Management (OAM) framework is an application development platform created to support in a development of an ontology-based application [7]. This framework is an integrated platform that supports both RDF data publishing from databases based on domain ontology and processing of the published data in ontology-based applications, i.e. semantic search and recommender system applications. It provides a user interface for ontology-data mapping, a semantic search engine and interface, and recommendation system engine.
The framework was implemented on top of existing Semantic Web data and application platforms that are D2RQ [8], Jena’s RDF data storage and Jena’s reasoning engine [9]. However, the provided interface can help users to skip a process to directly use of complex syntax of Jena’s reasoning engine.
In this work, we apply OAM for ontology-data mapping and rule generation. With the support from OAM, we can create a rule to determine an activity from instances in spreadsheet [10] without knowing syntax of JENA. 3
This work uses two sources of data. The first one is static data about students and course. The second one is data obtained from RFID to get identification, location and time of students. An ontology schema is designed to cover aspects of the data stored in database, and mapping between ontology and database are done. With the ontology, an inference engine can infer an activity according to a semantic of a given person, location and time. An overview of the proposed method is illustrated in Fig. 1. This part describes about general data from university database. These data include information of students, courses and buildings as shown in Table 1 and 2 respectively. These data are used as reference for real-time data. Real-time data are obtained from RFID (Radio-frequency identification) readers. To gain data, RFID data requires students to place a tag in their student ID card on an RFID reader. RFID readers are located in front of each room, and touching a card to a reader is treated as counting for class attention or authentication to access and exit some certain areas such as library and computer room. The setup of RFID for students to identify themselves and authentication are exemplified in Fig. 2.
Fig2. An RFID setting in front of rooms from the testing site
Upon reading, identification of student is gained and the datum is stored along with time and reader location. RFID data are exemplified in Table 2. In this paper, we design an ontology [11] to represent knowledge of activity in campus. Hozo ontology editor [12] was used as development tool. The ontology is designed to collect concepts and their relations relevant to students, course, building and activities. The main tree of the ontology is activity concept that has properties such as location, time and person. Sub-concepts of activity are, for example, class_lecturing, lab_studying, self-studying, reading_book, playing_sport and meeting. Some parts of the ontology are shown in Figure 3.
Fig 3. Some parts of Activity Ontology from Hozo editor
In the total, this ontology contains 35 concepts and 26 relations. Once the ontology was completed, it was exported as OWL format [13] file to be used in further process.
To assign instance to an ontology concept, the two data stored in databases we mentioned above are focused. Database schema is mapped to ontology schema using Ontology Application Management Tool (OAM tool) [7]. To instantiate, we map a database field to OWL class from the ontology according to mapping table shown in Table 3.
To estimate an activity from instance, inference engine is needed to give a conclusion on the basis of evidence and reasoning. In this work, we applied OAM [7] in which having Jena inference engine [9] plugged in. With the power of OAM, we can create a rule to determine an activity from instances in spreadsheet [10] without knowing syntax of JENA.
Activity result Taking_Lecture Reading Borrowing_book Having_lunch Meeting
From Table 4, the first rule is to check ‘if students have their RFID read at the lecture room that is for a course that matches to the course of that student within 15 minutes before the class starting, an activity of Taking_Lecture will be assigned to that student’. Next rule is ‘if student goes in library for more than 30 minutes, his/her activity will be Reading’.
The activity will be assigned to students’ activity field in database with time stamp. With an accumulation of activities, an activity profile of students can be created.
Since this work aims to gather activity data of students, displaying data is a must to university personnel to analyze and plan accordingly. We designed two visualizing aspects in this work.
The former is to visualize a graph based on area and time. This visualization is to
find areas where are crowded and a time in hour-based for planning several services
such as security monitoring and transportation service [
Fig 4. A visualization of crowd in campus from RFID data
The latter visualization is to display activity profile of a student. These data are an accumulation of inferred activities of each student. This visualization can display and sort data in many aspects, such as activity of students in a free time or activity count of students who have GPA more than 3.00. Fig. 5 shows the screen-captured visualization of activity in campus.
Fig 5. A visualization of activity profiles of students
This work presents a method to detect students’ location in a campus for headcount in an area and activity of student. RFID readers are exploited and located at the fronts of every door of a room to identify students in the area for precision. Ontology was created to represent room type and activity that can happen in a room. With the ontology, activity of students can be inferred using an inference engine embedded in OAM framework based on RFID data and basic data from university. To make data easier for campus management to university staffs, visualization systems are implemented as a visualization to display crowd in a campus area and visualization for student activity.
This work applies on RFID technology in obtaining student data. Thus, we can automatically and precisely identify students and their location. However, in activity recognition, a location and activity are related using ontology and inference techniques to assign activity to students. This method provides us a simple way to gain activity in general. Unfortunately, this method cannot detect a mischievous behavior such as reserve a tutoring room for sleeping and causing other students to lose chance to use the facility for good deeds. Moreover, Some activities such as reading a book in a library should be gathered more in details if a library data about borrowing book can be accessible. With library data, we can infer deeply that students go in library to study more in subjects that related to their current course or not.
To improve our method, we plan to apply more sensor types, such movement sensor and electricity current usage sensor, to detect more minor actions to compose into an activity. We expect this to help us in getting more data in details for mischievous behavior detection for campus management and security control. Furthermore, more basic data from other databases such as library database will be gathered and applied in our method. This should identify details of action that a student conducts. 6. Wongpatikaseree, K., et.al. : Activity Recognition using Context-Aware Infrastructure Ontology in Smart Home. In The Seventh International Conference on Knowledge, Information and Creativity Support Systems (KICSS2012) (2012). 7. Buranarach, M., Thein, Y.M., Supanithi,T. : OAM: An Ontology Application Management Framework for Simplifying Ontology-based Semantic Web Application Development. To appear in : International Journal of Software Engineering and Knowledge Engineering 8. Bizer, C., Seaborne, A.: D2RQ-Treating Non-RDF Databases as Virtual RDF Graphs. In:
Poster at the the 3rd International Semantic Web Conference (ISWC2004) (2004). 9. Apache jena: available online at https://jena.apache.org 10. Buranarach, M., Rattanasawad, T., Ruangrajitpakorn, T.: Ontology-based Framework to Support Recommendation Rule Management using Spreadsheet. In The Tenth International Conference on Knowledge, Information and Creativity Support Systems (KICSS2015) (2015). 11. Mizoguchi, K.: Tutorial on Ontological Engineering. Title: Tutorial on ontological engineering - Part 1: Introduction to Ontological Engineering Author: Mizoguchi, R. Reference: New Generation Computing, OhmSha&Springer, Vol.21, No.4, pp.365-384, 2003 12. Hozo Ontology Editor: available online at http://www.hozo.jp 13. McGuinness, D., Harmelen, F., OWL Web Ontology Language Overview: available online at https://www.w3.org/TR/2004/REC-owl-features-20040210/