Evaluation of Quality for Semi-Structured Database System

Published Online December 2019 in MECS

{data($p1//mname)}

BI PIN

RASHI

SASHI

PRIYA

• 6.    Find the project name and project id from the CSG Project1 and Project2

{data($p1//pname)}

{data($p1//pid)}

{data($p2//pname)}

{data($p2//pid)}

< table ID=”project”>

ABC XYZ DEF XYZ ABC

P1001 P1003 P1004 P1005 P1001

PQR

P1006

• 7.    Find the details of publications where MName=”Bipin” from project1 and also find the details of publication where MName=”Priya” from Project2.

and $p2//mname = "PRIYA" return

{$p1//(puid,ptopics)}

{$p2//(puid,ptopics)}

P001

RRR

P007

NNN

• VI. Framework for Evaluation of Quality

The quality evaluation in multidimensional data model of conceptual level is two-fold viewpoints. Set of criteria are associate with each viewpoint, which are further defined using proposed metrics. As stated earlier, the performance (quality of results) and productiveness (time, cost, effort) of information system development may be affected by the grade of conceptual data model. In quality evaluation, the two viewpoints are (1) Designer level viewpoint and (2) User level viewpoint .The criteria like transactional/query throughput and query performance are identified by the viewpoint of designer level and the criteria like effectiveness and analyzability are identified by the viewpoint of user level.

• A.    Transactional/Query Throughput (QthD)

Throughput measures the number of transactions executed per second. Generally, the speed of a database in a system measured by throughput. For execution of the full set of five queries in different order, a number of query users(S) are chosen, which described in section VII Case-I and Case-II. The throughput metric is computed as the total amount of works (Sx5), converted to hours from seconds (3600 seconds per hour) and divided the total elapsed time (T S )[Elapsed time is simply the amount of time that passes from the beginning of an event to its end.] required between the starting of first query and completion of the last one query.

Qt ℎ=

SX5X3600

• B.    Performance of the Query Execution (QEP)

Performance of query execution means the amount of time for execution of a specified query retuning with an appropriate resultant set i.e the time to make a successful one round trip.

• C.    Effectiveness (E)

A schema is said to be effectiveness    when it represents user requirements in a natural way as well as semantic way. Measurement of effectiveness used the concept of some conceptual data model which is sufficient for exposed of some specified user requirement analysis in the system.

QthD

• D.    Analyzability

The Analyzability is the measurement of the flexibility of a user in a database model.

• VII. Empirical Validations of Proposed Metrics

This portion of the article is focused on the proposed metrics and measurements empirical validation in order to prove their practical utility. The set of proposed metrics use the mechanism for guiding the grade of data models from a practical point of view that is known as the objective of empirical study. An experiment has been setup for analyzing the group of metrics and the proposed quality measurements like query throughput and query execution performance. The process of empirical validation also aims to recognize the metrics and measurement from the proposed set.

• A. Experimental Settings

The desire definition of the experiment using TCSS model can be encapsulated as:

In order to analyze the set of metrics for TCSS for the purpose of evaluating if they are useful with respect of the measurement of quality of a specified data model and operability in the context of query.

Query :

To examine the scalability of proposed TCSS X-Query implementation, trying to perform an experimental evaluation using ORDER XML and PART XML database. The size of ORDER xml is 5571 KB and the size of PART xml is 1000KB respectively.

• Cases :

According to the query and their types the database is organized into five basic types of queries: Selection, Retrieve, Union, Intersection and join; and the query subsets are categorized into Q1 to Q5. Trying to perform an experimental evaluation using ORDER XML database (data are taken anonymously)(Case I). The size of ORDER xml is 5571 KB(Case II).

• Case I:

Trying to perform an experimental evaluation using ORDER XML database (data are taken anonymously). The size of ORDER xml is 5571 KB.

Q1: find the order status and order date from CSG order.

Q2: find the details of order where ORDERKEY=”2” and order CUSTOMERKEY= “781”.

Q3: find the details of order priority and order comment where O_ORDERKEY=”992” and “358”.

Q4: find the order STATUS which have the same CUSTOMERKEY=”317” and ORDERKEY=”998”.

Q5: find the CUSTOMERKEY and ORDERSTATUS of all orders where all the ORDERID are same.

• Case II:

Trying to perform an experimental evaluation using PART XML database (data are taken anonymously). The size of PART xml is 1000KB.

• Q1: find the parts name and parts Brand from CSG Part.

• Q2: find the details of part where PARTKEY=”3” and Part RETAILPRICE=”903”.

Q3:find the details of PART BRAND and PART COMMENT where P_PARTKEY=”3” and “938”

• Q4: find the part CONTAINER of all part which have the      same      PARTKEY=”1960”      and

PARTBRAND=”BRAND#33”.

• Q5: find the PART SIZE and PART TYPE of all parts where all the P_BRAND = “BRAND#32”.

Table 2. Metrics and Measurement Value of Each Case(in Scheam level)

CASE I Query	QthD	QEP	E
Q1	10.1882	2073	0.098
Q2	9.5377	2177	0.1048
Q3	9.9475	2093	0.1005
Q4	10.050	2094.5	0.0995
Q5	9.824	2139	0.1018

CASE II Query	QthD	QEP	E
Q1	15.3191	800.5	0.0653
Q2	16.129	806.5	0.0620
Q3	15.8765	736.5	0.0630
Q4	15.852	781.5	0.0631
Q5	6.375	736	0.0611

Table 3. Collected Operation Time in ms.((TCSS-X-query)

CASE		Q1	Q2	Q3	Q4	Q5	AVG-TIME
CASE I	Comp.Time	1469	1582	1544	1473	1509	1515.4
	Eva.Time	2045	2089	2040	2082	2086	2068.4
CASE II	Comp.Time	1442	1547	1591	1460	1487	1505.4
	Eva.Time	841	840	738	731	760	782

Hypotheses: the following hypotheses are used for the experiments [28]:

• • Null hypothesis (H0): Into the set of metrics and Quality Measurement as well as feasible factor of data model have no significant relationship.

• • Alternate hypothesis (H1): Into the set of metrics and Quality Measurements as well as feasible factor of data model have significant relationship.

• B.    Experimental Steps

In order to obtain the results of the experiment it is categorized into 2 phases. In the first phase, it is checked that there are no relationship among the group of schema metrics. In the second phase, there are relationship into the set of average compile time and the average operation time, which has been evaluated to identify the group of metrics. The feasible factor of the conceptual data model has been significantly influenced by the group of metrics [28].

This is performed the independency test using non-parametric chi-square test. In both type of analysis the level of significance is set to α=0.10. Thus in both types of analysis if ρ value ( 2tailed ) <0.10 the null hypothesis H0 will be rejected .

Phase- I: There is no relationship between the set of schema level metrics, which are tested using non-parametric chi-square test. The result has been shown in below(spss-22):

Chi-Square Tests

Value df Asymp. Sig. (2-sided) Pearson Chi-Square Likelihood Ratio N of Valid Cases 20.000a 16.094 5 16 16 .220 .446 a. 25 cells (100.0%) have expected count less than 5. The minimum expected count is .20.

Fig.2. Phase I Chi-Square Test Query Throughput* Query Name

Chi-Square Tests

Value df Asymp. Sig. (2-sided) Pearson Chi-Square 20.000a 16 16 .220 .446 Likelihood Ratio 16.094 N of Valid Cases 5 a. 25 cells (100.0%) have expected count less than 5. The minimum expected count is .20.

Fig.3. Phase I Chi-Square Test QEP* Query Name

Chi-Square Tests

Value df Asymp. Sig. (2-sided) Pearson Chi-Square 20.000a 16 16 .220 .446 Likelihood Ratio 16.094 N of Valid Cases 5 a. 25 cells (100.0%) have expected count less than 5. The minimum expected count is .20.

Fig.4. Phase I Chi-Square Effectiveness * Query Name

The following Hypothesis is considered for the purpose. H01: Non significant relationship among all attributes. H11: Significant relationship among all attributes.

Reject H01, if p- value<0.10.

In Chi-Square Test all the obtained p- value are greater than α value 0.10. Hence it is significant that there is no significant relationship in all schema level metrics.

CASE I:

Hypothesis Test Summary

Chi-Square Tests

Value df Asymp. Sig. (2-sided) Pearson Chi-Square 20.000a 16 220 Likelihood Ratio 16.094 N of Valid Cases 5 16 .446 a. 25 cells (100.0%) have expected count less than 5. The minimum expected count is .20.

Fig.8. Phase II Chi-Square Test Evaluation Time*Query Name

Null Hypothesis             Test Sig. Decision

The distribution of QueryThroughpi^1                     Retain the

1 is the same across categories of k-r„J1|^         .406 null

QueryName.                   Wahis Test            hypothesis.

I n d e p e n d e nt

2 The distribution of QEP is the sameSamples across categories of QueryName. Kruskal-

Wallis T est

Retain the .406 null hypothesis.

The distribution of Effectiveness is dependent-            Retain the

3 the same across cateqories of LTmcJai           .406 null

Q u e ryN a m e.                       Wallis est              hyp oth esis.

Analyzing the above table, all the p-value obtained in Chi-Square Test is greater than α value 0.10 (its 0.22). it can be concluded that there exist a strong relation among compile time and evaluation time as the p-value>0.10, in each case. Hence, the proposed measure has significant influence on the Operability factor of conceptual level multidimensional data model.

Asymptotic significances are displayed. The significance level is .05.

Fig.5. CASE I Hypothesis Test Summary

CASE II:

Hypothesis Test Summary

	Null Hypothesis	Test	Sig.	Decision
■1	The distribution of QueryThroughpi^j1®^^1®^' is the same across categories of Q^ryName.              К Test		.406	Retain the null hypothesis.
2	Independent- The distribution of QEP is the sameSamples across categories of QueryName. Kruskal"               "          Wallis Test		.406	Retain the null hypothesis.
3	The distribution of Effectiveness is the same across categories of QueryName.	IndependentSamples Kruskal-Wallis Test	.406	Retain the null hypothesis.

Asymptotic significances are displayed. The significance level is .05.

Fig.6. CASE II Hypothesis Test Summary

Phase-II: In this phase, there is a relation between proposed set of compile time and evaluation time. The set of average compile time and the average operation time, which has been evaluated to identify the existence of any significant influence of the group of metrics. The feasible factor of the conceptual data model has been significantly influenced by the group of metrics [26] .The result has been shown in below:

Chi-Square Tests

Value df Asymp. Sig. (2-sided) Pearson Chi-Square 20.000a 16 220 Likelihood Ratio 16.094 16 .446 N of Valid Cases 5 a. 25 cells (100.0%) have expected count less than 5. The minimum expected count is .20.

Fig.7. Phase II Chi-Square Test Compile Time*Query Name

• VIII. Conclusion

In this article, a frame work of quality evaluation for conceptual level multidimensional data model has been discussed in general and for TCSS data model, in particular. The framework comprised of Direct and Indirect metrics. The set of proposed metrics use the mechanism for guiding the grade of data models from a practical point of view that is known as the objective of empirical study. An experiment has been setup for analyzing the group of metrics and the proposed quality measurements like query throughput and query execution performance. The process of empirical validation also aims to recognize the metrics and measurement from the proposed set. This article proposes a framework for quality evaluation of transactional calculus for semi-structured database system using TCSS X-Query with two different cases of data with different size.

This article also has been focused on empirical validation of the set of metrics and measurements to prove their practical utility. The several proposed metrics and operability factor of multidimensional conceptual data model significant influence is shown by empirical validation.