...lopment for public participatory GIS.pdf

...lopment for public participatory GIS.pdf tài liệu, giáo án, bài giảng , luận văn, luận án, đồ án, bài tập lớn về tất...

Requirement Analysis and Metric Development for

Public Participatory GIS

Nagesh Kolagani # , Palaniappan Ramu $ , Corne van Elzakker*, Vishram Naniwadekar € ,

Koshy Varghese §

# IIT Madras, Chennai 600 036, India nagesh333@gmail.com

$ IIT Madras, Chennai 600 036, India palramu@itm.ac.in

€ IT Consultant, 81, Vayupuri, Hyderabad 500 094, India vishram@alumni.iitm.ac.in

* ITC, University of Twente, The Netherlands c.vanelzakker@utwente.nl

§ IIT Madras, Chennai 600 036, India lakshman@iitm.ac.in

Abstract: During the last few decades, many rural communities in the developing countries are

facing a crisis due to over exploitation of natural resources Governmental and other agencies are carrying out various intervention programs to remedy the situation However, the achievements of many of these programs have been sub-optimal due to lack of participation by stakeholders in effective monitoring of their implementation A majority of the stakeholders do not have formal education and are unable to understand and participate in these programs effectively Public Participatory GIS (PP-GIS) can help such stakeholders to visualize, understand and participate better Hence, there is a need for a framework to identify, describe and analyze requirements of rural communities for such a PP-GIS solution While such frameworks are used extensively when developing commercial software, they are not being widely practiced when developing PP-GIS solutions In this paper, we propose a 'Quality Function Deployment' based framework for developing

a PP-GIS solution We demonstrate it using case study of a PP-GIS solution developed using ‘Free and Open Source Software’ modules for participatory monitoring of natural resource management programs A detailed requirement analysis was carried out with various user groups The qualitative user requirements gathered during these discussions were translated into quantifiable metrics These were then assigned weights based on their relative importance using 'Analytic Hierarchy Process' method A PP-GIS solution with following three components was developed to meet these requirements: 1) an Android mobile application for collecting field data, 2) a cloud based Quantum GIS Python plugin that converts field data into maps and 3) a GeoServer that serves these maps using 'Web Feature Service' to a Quantum GIS desktop client Usability trials of the PP-GIS solution were carried in an Indian village and it met most of the requirements well

Keywords: Public Participatory GIS; Requirement Analysis; Quality Function Deployment; Metric development; Analytic Hierarchy Process

1 INTRODUCTION

During the last few decades, many rural communities in the developing countries are facing a crisis due to over exploitation of natural resources For example, in India, introduction of deep tube well technology in the 1980s lead to large-scale extraction of groundwater from deep aquifers for cultivation of profitable water intensive crops; such a multi-fold increase in extraction resulted in severe groundwater crisis by 2002 To remedy the situation, Governmental and other agencies are carrying out various 'Natural Resource Management' (NRM) programs, such as construction of rainwater harvesting structures to increase recharge of aquifers However, the achievements of many

of these programs have been sub-optimal (Government of India, 2011) due to lack of participation by stakeholders in effective monitoring of their implementation When authors visited some of these villages in India, they noticed a majority of these NRM works were very inefficient (Figures 1 and 2)

This is due to the fact that a majority of the stakeholders do not have formal education and hence are unable to understand and participate in these programs effectively Public Participatory GIS (PP-GIS) can help such stakeholders to visualize, understand and participate better Free and open source software based mobile and cloud technologies, such as use of Android mobile platforms and internet browsers, have spread widely in the villages and towns respectively during the last decade and are making it possible to develop and deploy such PP-GIS solutions easily

There is a need for a framework to identify, describe and analyze requirements of rural communities for such a PP-GIS solution While such frameworks are used extensively when developing commercial software (Lamia, 1995), they are not being practiced when developing PP-GIS solutions

Figure 1: A well located farm pond with water Figure 2: A farm pond with no water

The PP-GIS solution needs to be evaluated and further enhanced iteratively based on user feedback until a satisfactory solution is arrived at To facilitate development of such a solution, the process of evaluation needs to be carried out using quantifiable and objective metrics Such metrics need to be developed from the requirements gathered using this framework The framework should be a generalized framework to facilitate its application in different parts of the world and in diverse circumstances (Geertman, 2002)

In this paper, a Quality Function Deployment (QFD) based framework (Akao, 1990; Dean, 1992) is proposed for developing PP-GIS solutions QFD is a framework developed in Japan in the 1960's to capture customer requirements and to develop products to meet these requirements QFD is used extensively for product development in software domain (Richardson, 1997) as well as in non-software domains (Hepler and Mazur, 2006) In this paper, it is demonstrated using case study of a PP-GIS based monitoring system for NRM A mobile and cloud based free and open source PP-GIS solution was developed to meet these requirements and was compared with existing solutions

2 METHODOLOGY

The main matrix in QFD is the 'House of Quality' (Figure 3) It is a conceptual map that helps in gathering customer requirements and translating them into quantifiable metrics It involves the following steps:

1 'Who': Identifying the customers (ci): This is important as very often there is more than one

type of customer, e.g community stakeholder and governmental regulatory agency, each with their own requirements

2 'What': Determining customer requirements (rj): A detailed survey and analysis of

requirements of customers will be carried out based on discussions with customers through methods such as 'Key Informant Interviews’ and 'Focus Group' discussions Literature survey and discussions with technical experts will help in carrying out the requirement analysis and

in refining the requirements identified

3 'What' versus 'Who': Determining customer-wise relative importance of requirements (wij):

Analytic Hierarchy Process (AHP) (Saaty, 1980) is a suitable method for identifying relative

importance of these requirements and assigning them weights (w ij) It is explained in detail in the next section 2.2

4 'Now' and 'What versus Now': Evaluating existing products (pk) for their ability to meet these

requirements: A 5-point scale (0 to 4) (nik) is used for this purpose starting from 0='does not

meet at all' to 4='meets completely'

5 'How': Restating customer requirements in terms of quantifiable metrics (ml): Each customer

requirement should have at least one metric (or more) to measure it

6 'What' versus 'How': Relating each metric to each requirement (hil): A 4-point scale (0 to 3)

can be used to measure how well a metric captures a requirement: 'None', 'Weak', 'Medium'

or 'Strong' The resulting 'hil' matrix is then normalized along each row i.e sum of values in

each row is made equal to 1

7 'Conflict': Identifying inter-dependance between metrics (if any): This identifies conflict between metrics, which can be either negative or positive A scale of -3 to 3 is used for this purpose, starting at -3=Strong negative relationship to 3=Strong positive relationship

8 'Score': Evaluating overall score (sk) of existing products using metrics: Evaluation of existing products (pk) was done earlier by customers using a 5-point scale It will be now repeated technically using value (x kl ) of each metric (m l ) for each product (p k ) An overall score (s k) is computed for each product for easy comparison This is calculated as a value between 0 to 1 using equation 1:

i multiply each metric values (x kl) with:

a how well a requirement is captured by this metric (hil) and

b how important that requirement (wi) is

ii sum these products

9 'How much': Determining metric-wise targets for the product under development: It is a set of

target values for each metric (tl) to be met by the new product

Figure 3: 'House of Quality' AHP is used for assigning weights to various requirements based on their relative importance to the customer It involves the following steps:

 Pair-wise comparison of requirements based on their relative importance is gathered from a customer using a questionnaire A scale of 9 to 1/9 is used for this purpose

 Consistency index (CI) is computed for each of these resulting pair-wise reciprocal comparison matrices:

where: λmax: Maximum real eigen value

n: Number of dimensions/metrics RI: Random Index (=1.32 for n=7)

Those with a value less than 0.15 will be accepted as consistent and considered for further analysis

 Principal eigen vectors are computed for each of these AHP matrices Mean or median of the components of these eigen vectors can be used as the weights of these requirements

Prototypes of new product designs need to be evaluated and further enhanced iteratively based on user feedback till a satisfactory solution is arrived at Following steps are carried out for this purpose:

1 Selecting existing products and proposed new product designs that meet these requirements

2 Field trials of these products and the proposed metrics

3 Enhancing new product(s) based on user feedback (back to step 2)

4 Evaluating and refining the proposed metrics (back to step 2)

The proposed QFD based framework for developing a PP-GIS solution is demonstrated using case study of a PP-GIS based monitoring system for NRM A village in India where Government was carrying out a NRM project was selected for this purpose As part of the project, the Government provides funds to the village to construct several NRM related community works, such as rainwater harvesting structures The Government constitutes a ‘Watershed Development Team’ consisting of few local stakeholders and few non-local technical experts to plan and implement these works with active involvement of stakeholders To ensure active participation of all stakeholders, the Government tries to make detailed information about each work available to them in an easy manner The current practice being followed is to paste detailed information about each work on a 'Village Notice Board' (‘VNB’) The information is organized as tables where each row represents an NRM work and each column describes certain aspect of that work: unique ID, type, description of location, implementation status, cost, etc Due to issues of low literacy among stakeholders, lack of sufficient details in the data provided, lack of privacy in accessing the data, etc., the ‘VNB’ approach is not being very effective and is not being amenable for improvement To remedy the situation, a ‘villageGIS’ software product was developed and deployed in this village and was found to be much more effective The QFD process adopted while developing the ‘villageGIS’ product is described in detail below

1 'Who':

Through literature survey and discussions with technical experts, users listed in Table 1 were identified

Table 1: List of users

1 CSH Community Stakeholders

To monitor community works in their village E.g Literate, semi-literate and illiterate; Farmers and farm workers; Male and female

2 CRP Community Resource Persons

To collect field data needed by CSH E.g Middle school children; Literate youth; Literate farmers and farm workers

3 CBO Capacity Building Organization

To train CSHs in data usage and CRPs in data collection

4 PIF Project Implementing Agency: Field staff

To help CRPs in locating works in the field and collecting data

5 PIO Project Implementing Agency: Office staff

To provide additional data about works and their surroundings E.g Official physical and financial data

E.g Background maps such as satellite raster imagery and farms GIS vector maps

6 PIM Project Implementing Agency: Monitoring wing

To monitor works and overall project progress in various villages

2 'What':

For each user group, a detailed requirement analysis was carried out using semi-structured ‘Key Informant Interviews’ with a few knowledgeable users The requirements gathered were refined through ‘Focus Group’ discussions with other users These are organized hierarchically and

presented in 1st column of Table 2 with appropriate indentation

3 'What' versus 'Who':

For each user group, AHP was carried out with a few knowledgeable users and weights (w ij) were computed for their requirements The weights are also organized hierarchically and presented in Table 2 under columns CSH to PIM (i.e columns 2 to 7) with appropriate indentation The sum of weights under any 1 requirement at any one level adds up to 1

Table 2: User group wise requirements and theirs weights

4 'Now' and 'What versus Now':

The current practice, 'VNB', was evaluated based on subjective user feedback on a scale of 0-4 for its overall ability to meet user requirements The results are summarized in column 2 of Table 3

5 'How':

For objective assessment, each requirement is measured by one or more of following metrics (column

3 of Table 3):

 Some of these metrics are administered to customers as questions with ‘Yes’ or ‘No’ as answers; such metrics are indicated as Q1, Q2, etc

 Some of these metrics are measured by administering 1 or more of following usability trials:

 M1: No of works accessed successfully in 30 minutes (while other parameters are fixed)

 M2: Time taken to count number of works in a specified area of specified type(s)

 M3: Time taken to compute total cost of works in a specified area of specified type(s)

 M4: Time taken to count number of works in a specified area of specified type(s) with catchment area less than a desired value

 M5: Time taken to count number of works in a specified area of specified type(s) with % beneficiaries from lower strata less than a desired value

 M6: Time taken to acces)s data of one work (while other parameters are fixed)

 M7: Time taken to collect data of one work (while other parameters are fixed)

6 'What' versus 'How':

How well that metric captures a requirement is specified by a value between 0 to 1 (column 3 of Table 3) E.g “M1(1.0)” against req 2(i) indicates metric M1 captures req 2(i) fully

7 'Conflict':

No significant conflicts were noticed between the metrics

8 'Score':

Current practice, ‘VNB', was evaluated again objectively using above metrics (column 4 of Table 3) Table 3: Metric-based evaluation of two products: ‘Village Notice Board’ and ‘villageGIS’

Product VNB (0-4)

VNB (x kl )

villageGIS

(x kl )

1 Detailed data about each community work

i Physical details

ii Financial details

iii Time, such as Start date-&-End date and current status (Proposed/In-Progress/Completed)) 4 Q7(1.0) 1 1

v Physical surroundings, such as Catchment area, Satellite imagery & other background maps 0 Q9(1.0) 0 1

vi Social context, such as socio-economic background of beneficiaries, farm & other background maps 0 Q10(1.0) 0 1

2 Search functionality

ii Sequential access: Locate works in a specified area of specified type(s) 1 M2(1.0) 0 1

iv Spatial query on attributes such as cost, socio-economic background of beneficiaries, etc 1 M4(0.5),M5(0.5) 0,0 1,1

5 Privacy

8 Technical support

9 'How much':

After discussion with customers, taking into account evaluation of existing product, metric-wise targets for the product under development were determined To meet these requirements, a free and open source mobile and cloud based PP-GIS solution, 'villageGIS' (Figure 4), was developed It has three components: 1) an Android mobile application for collecting field data, 2) a cloud based Quantum GIS Python plugin that converts field data into maps and 3) a GeoServer that serves maps using Web Feature Service (WFS) to a Quantum GIS desktop client This PP-GIS solution was evaluated in the case study village (Figures 5 to 10) to see how well it met the above requirements and refined further Final evaluation results are given in column 5 of Table 3 It was observed to meet most of the requirements very well

Figure 4: A mobile and cloud based PP-GIS solution: 'villageGIS'

Figure 5: Android mobile App Figure 6: Village youth gathering field data

To create an easy-to-understand map of all community works, village youth go to each work with an Android mobile app and collect its GPS coordinates and photographs using simple button clicks They audio record its measurements, beneficiary details, etc School children download data of all works into a computer and create a map in Quantum GIS using an easy-to-use single-click Python plugin

Figure 7: Quantum GIS Viewer Figure 8: School children preparing GIS maps

Figure 9: Quantum GIS maps of proposed and implemented NRM works can be used by stakeholders

to visualize and understand their problems and associated solutions Figure 10: This is eventually

expected to lead to locally relevant solutions such as tube well cooperatives

Detailed requirements and associated metrics were identified for a PP-GIS solution required to monitor NRM programs in Indian villages A mobile and cloud based PP-GIS solution, 'villageGIS', was developed to meet these requirements During field trials in a village, the PP-GIS solution was observed to meet most of the requirements well It can be extended successfully to other villages Successful case study shows that the proposed QFD based framework holds promise for use while developing PP-GIS solutions for other participatory programs in other parts of the world More case studies will be needed to evaluate its benefits and limitations

Above work was supported by grant no 5-4/2010-TE, TDET scheme, Department of Land Resources, Ministry of Rural Development, Government of India It was carried out in close collaboration with WASSAN, Hyderabad and MS Mobile Technologies Private Limited, Hyderabad We thank G

Krishnamurthy, N Chandra, Ms P Nagamma and other participants from the local community for

their active support in anchoring the field work

Akao, Y., (ed.) 1990 Quality function deployment Productivity press, Cambridge, MA

Dean, E B., 1992 Quality function deployment for large systems Proceedings of the 1992

International Engineering Management Conference, Eatontown, NJ, 25-28 October

Geertman, S., 2002 Participatory planning and GIS: a PSS to bridge the gap Environment and Planning B, 29, 21-36

Government of India, 2011 Common Guidelines for Watershed Development Projects-2008: Revised Edition-2011 National Rainfed Area Authority, Planning Commission, Government of India, New Delhi, India 59 pp

Hepler, C and Glenn M., 2006 Finding customer delights using QFD Proceedings of Quality Institute for Healthcare Annual Conference American Society for Quality Section 1

Lamia, W M., 1995 Integrating QFD with object oriented software design methodologies

Transactions from the seventh symposium on quality function deployment, Novi, MI, 11-13 june,

pp 417-434

Richardson, Ita, 1997 Quality function deployment - A software process tool? Third Annual

International QFD Symposium, 1st-2nd October Linkoping University, Linkoping, Sweden Saaty, Thomas L., 1980 The Analytic Hierarchy Process, McGraw-Hill, New York