Using Concepts in Grid Data Organization - Advanced Research on Computer Education, Simulation and Modeling - page 449

of the object class department, course and student respectively. Each student has a
unique stuNo. The attributes title, mark, address and hobby are optional. Attribute
hobby is multivalued, while stuName is required. There are two relationship types,
called dc and cs. The former is a binary relationship type between object classes de-
partment and course, while the latter a binary relationship type between course and
student. A department can have one or more (1:n) courses, and a course belongs to
one and only one (1:1) department. A course can have zero or more (0:n) students,
and a student can take 1 or more courses. The label cs on the edge between student
and mark indicates that mark is a single valued attribute of the relationship type cs.
That is, the attribute mark is an attribute of a student in a course. From these con-
straints, we can derive that {course; student}→mark.
3 Using Concepts in Searching Processing
Here we outline how the concepts captured by CRD-FS mode can be used to optimize
linked pattern searching evaluation with three linked pattern queries.
Searching 1: Find the stuName values of student elements having stuNo value equals
to “c1001”. The XPath expression is: //student[@stuNo=”c1001”]/stuName.
Using the CRD-FS mode, we know that stuName is a single valued attribute of the
student object class and stuNo is the identifier of the student, so stuNo→stuName. To
prcess the searching, we only need to find the first student element in the grid file
with attribute stuNo equal to”c1001”, and then find the value of its subelement stu-
Name. However, if we use a WSDL or grid mode of the grid data, we would not
know that stuNo is the identifier of student or that stuName is a single valued attribute
of student, so we would need to traverse the whole grid file.
Additionally Wu et al. have proposed an algorithm that concentrates on searching
for content or values with concept information as compared to structure-focused
searching processing. We will discuss content search in more details in Section 5.
Searching 2: Find the average marks of all the students.
To answer this searching the processor needs to know that stuNo is the identifier of
object class student, and mark is a single valued attribute of the relationship type
between course and student.
However, a WSDL cannot express the concepts that stuNo is the identifier of stu-
dent object, and also cannot express that mark is a single valued attribute of the rela-
tionship between student and course, that is {course; student}→mark. Without this
information, there is no way to know whether the XSearching searching with an ag-
gregation function (or linked pattern searching) is written correctly or not.
Searching 3: For each student, find all courses taken by the student with the marks the
student achieved in the course. Consider for example the searching , where a rectangle
represents an object class and a circle represents an attribute. On the left hand side the
searching is specified, and on the right hand side the output format is given. The lines
between the searching part and the output part show the correspondences between the
objects in the searching and those in the output. The searching is asking for the marks
of students in courses and in the output courses must be nested within student rather
than students within courses.