Data Quality Management Maturity Measurement of Government-Owned Property Transaction in BMKG

Government-Owned Property (GOP) management, including the bookkeeping of GOP transaction, is part of GOP Officer responsibility to ensure the quality of transaction data. This responsibility also applies toGOP Officer in Indonesian Agency for Meteorological, Climatological and Geophysics Badan Meteorologi, Klimatologi, dan Geofisika (BMKG). GOP data as the source for the Central Government Financial Report is expected to be well-maintained. It must be presented as accurate as possible, although there are still inaccurate data presented in the latest BMKG GOP Report. This qualitative research using document study and some interview sessions aims to measure how well the Data Quality Management (DQM) maturity of GOP transaction in BMKG using Loshin's Data Quality Maturitymodel. Thus, the result of maturity assessment is analyzed to recommend and implement DQM activities from the Data Management Body of Knowledge (DMBOK). The purpose is to improve GOP DQM. The research shows that the level of DQM maturity is at a repeatable level to defined level. Moreover, 52 maturity characteristics need to be followed through with DQM activities.


I. INTRODUCTION
N OWADAYS, data have been considered as the main capital in an organization as well as financial and human capital. Organizations need to pay particular attention to data capital, as data give added value to the organization [1]. The statement is in line with the results of a survey conducted by Ref. [2] in 179 large companies. They concluded that business with Data-Driven Decision (DDD) making produced the highest productivity about 5-6% higher than expected. Reference [3] showed that the implementation of data governance affected the data quality especially Ensuring data quality is an important step to improve business results. The business analysis based on bad data will result in business losses. The data quality also influences the level of users satisfaction and stakeholders [4]. Maintaining high data quality level is essential for the organization, whether it is to improve the productivity of its employees or to give better services to the customers. To achieve a good data quality level, key stakeholders have to understand the importance. The organizations need to have a data steward and apply appropriate technology as well [5]. Data Quality Management (DQM) is a concept and practice to improve data and information quality including the quality of organization's policies and guidelines, the measurement, analysis, cleansing and correction, data process improvement, and data quality education [6].
The government of Indonesia has also been aware of the importance of maintaining their data including the data of their property as part of policies-making in various fields. Their concerns are shown by several Central Government Regulations and the Minister of Finance Regulation. Those are expected to serve as guidelines for Ministries/Agencies/Local Governments in managing their assets. Indonesian Meteorology, Climatology, and Geophysics Agency Badan Meteorologi, Klimatologi, dan Geofisika (BMKG) as one of the non-ministerial government agencies has been implementing the governmental act in meteorology, climatology, air quality, and geophysics. These duties include managing their state property as stated in 2008 Indonesian President Act Number 61 about Agency of Meteorology, Climatology, and Geophysic. According to 2014 Indonesian Goverment Act Number 27 about Goverment-Owned Property Management. There are several activities to maintain the property such as plan-making related to the financial [13], procurement, and investment in assets.
After the maturity level determined, an analysis is conducted to provide recommendations to improve the quality of data management strategies based on DQM activities in Data Management Body of Knowledge (DMBOK) which are considered as the best-theory approaches to data quality improvement [14]. Data quality activities in DMBOK are selected due to their continuous activities to ensure the desired level of data quality of the organization in each DQM cycle.

A. Data Quality Management (DQM)
DMBOK mentions that DQM is a vital support process in organizational change management. Data quality is closely related to the quality of information. Low data quality causes inaccurate information and leads to business performance degradation [6]. DQM has an impact in decision support system as well as the value of the decision [15]. Therefore, poor DQM impacts poor operational activities and strategic decision-making.
A governmental organization needs to implement data quality management initiatives since this methodology has proven to improve business decision-making. It also improves the organizational data integrity, controls business cost, reduces the risk of fraudulent activities, and maintains customer relationship [16].
There is a limitation to improve data quality if correction is done at the error data and does not seek the cause of the error. These limitation leads to a continuous correction process. Therefore, a framework for DQM is required to improve data quality more effectively and efficiently [17,18].

B. Data Quality Maturity Model
A performance management approach to data quality is used to illustrate how DQM is related to all activities in an organization depending on the information. Since the information is based on data and to improve data quality, organization needs to understand how far the maturity that fits the needs can provide a criterion to analyze their capability [11,19]. One way to evaluate and solve this problem is to assess the level of maturity associated with the quality of the data. Then, it determines the target level of maturity that meets the requirements of the organization the best.
A tool called Data Quality Maturity model can be used to categorize the level of maturity of an organization in handling design, implementation, production, problem-solving, and others [11]. The same approach applies to the Data Quality Maturity model which 60 I n P r e s s measures and visualizes how DQM aligns with all information activities in an organization. Table I shows some previous research which measured data quality.

C. Data Quality Maturity Model by David Loshin
Loshin has a model that can be used to measure the maturity of a quality data management. It is called as Data Quality Maturity model. This model is an adaptation of the maturity model developed by the Software Engineering Institute in Carnegie Mellon University. The framework is used to measure the maturity of DQM based on eight components. Those are: 1) Data Quality Expectations. This domain measures the expectations related to the quality of data that are explicit or implicit in various directives and policies of the organization. Determination of data quality expectations includes relevant measures in the dimensions of data quality, metrics to evaluate compatibility in each dimension, and processes for evaluating compatibility in each dimension. 2) Dimensions of Data Quality. This domain emphasizes the classification of data quality expectancy components and provides steps to evaluate compatibility with the measurement of the expected quality of the data. 3) Policies. Various types and sources of data cause complexity in data management. The created policies to manage data management include data certification, privacy, data flow, and reliable data sources for the organization. 4) Procedures. Data quality procedures describe the operational aspects of a system to validate the existence and effectiveness of data management activities. 5) Governance. DQM should incorporate participatory, collaborative, and oversight management of all individuals within the organization. To realize it, it requires a data of governance structure that manages oversight and a set of data stewardship processes across all individual organizations. 6) Standards. Data standardization simplifies and adapts to external and internal information exchange standards. Standardization related to data quality is data definition, data meaning, and data exchange. 7) Technology. The implementation of a data quality framework involves the participation of individuals in organizations that are expected to use technology with the intention of adhering to data quality protocols and processes. It also supports data quality service levels through a reference set of data and validates/verifies the compatibility of data values with the expectation. 8) Performance Management. Specific processes for governance, stewardship, identification of data quality expectations, and determining the suitability of data quality expectations require performance management schemes to monitor overall organizational data quality.
The characteristics contained in the eight components can be seen in Tables A1-A2 (see Appendix). These eight components can be used as a measurement tool to determine how far the management of data quality within an organization is. The measurement produces values which are mapped according to their maturity level. There are five levels of maturity starting from the initial level which data practices and policies are still ad hoc to the highest that processes and practices assessed in a sustainable, upgraded, and optimized manner. The levels of maturity are as follows.
1) Initial. The process used for data quality assurance is mostly ad hoc with the most effort to respond to data quality issues. 2) Repeatable. There is some management in the organization and simple information-sharing activities. There are some process disciplines, mostly it is adopted from good practice and tries to imitate the practice in the same situation. 3) Defined. At this level, the team that handles data quality begins to document things like data governance policies, processes to define expectations of data quality, technology components, data quality, and report of validation processes. 4) Managed. DQM includes business impact analysis, defines expectations of data quality, and measures compliance with those expectations. 5) Optimized. Performance measurement across the organization can be used to identify opportunities for systemically improving data quality.

D. Data Management Body of Knowledge (DMBOK) in Data Quality Management (DQM) Activities
DMBOK has defined 12 activities that can be used to improve the quality of data adjusting to business objectives. Those activities include: 1) Develop and promote data quality awareness 2) Define data quality requirements 3) Profile, analyze, and assess data quality 4) Define data quality metrics 5) Define data quality business rules 6) Test and validate data quality requirements 7) Set and evaluate data quality service levels 8) Continuously measure and monitor data quality 9) Manage data quality issues 10) Clean and correct data quality defects  This research shows an approach to data quality orientation that facilitated and enhanced the quality of managerial decision-making in the context of redesigned business processes. Data quality was considered as a factor in business process success. It was conceptualized using a rule-based approach. 7 Reference [25] Accuracy The research proposed an integrated framework that organizations could adopt a part of the financial and management control process to provide a mechanism to calculate data problems. It determined potential solutions and monitored costs and benefits. It also improved and maintained data quality. 11) Design and implement operational DQM procedures 12) Monitor operational DQM procedures and performance Activities that are best practices in DQM based on DMBOK are used to response BMKG challenges in improving the maturity level of their DQM. DMBOK approach is a continuous approach so that the process of data quality improvement, planning, dissemination, supervision, and the action can be repeatedly done when data issue arises.

A. Research Stages
This research started by determining the problem to the final step. The researchers map the challenges in DQM with data quality improvement activities according to DMBOK. The research stages are in Fig. 1.

B. Problems Identification and Define Framework
Accurate and reliable data quality expectation has been stated in the 2016  evidence is the misclassification of GOP that is not in line with the codification of GOP. Real life GOPs locations which are not presented in the application also contribute to the inaccuracy of data. These realities cause the BMKG financial statements and the state property statement to be inaccurate. Therefore, it is necessary to measure the maturity of DQM and recommend activities to improve the maturity.
To determine the framework, the researchers conduct a study of literature on previous research. Based on several researches in Table I, the researchers define the framework proposed by David Loshin as the model. It emphasizes on the impact of data quality to analyze the data in BMKG.

C. Collecting Data
The characteristic in every component of Loshin's Data Quality Maturity model is used as a checklist to assess the compliance with the state of DQM. Checklists are formed to resemble a matrix to simplify the assessment process. The presentation of the assessment data includes a characteristic of ID code adjusting to the level of maturity. The codes are expectation component (H), dimension (D), policy (K), procedure (P), governance (G), standardization (S), technology (T), and performance management (M). For maturity code, initial (I), repeatable (R), defined (D), managed (M), and optimized (O). The characteristic ID is a combination of characteristic code, the code of maturity level, and the serial number of characteristics in each component per level of maturity in the theory of David Loshin [11].
Due to a large number of checklists, the first step in data collection and compatibility assessment is done by observing the documents. The documents are regulations related to the management of GOP such as Central Government Regulation, Presidential Regulation, Minister of Finance Regulation, and BMKG Regulation, financial statements, GOP statement, and other reports related GOP managed by BMKG. Several documents cannot evaluate some of the checklists. Some of them need an evaluation from people who ever do the activities on the checklist or see the activities.
Moreover, the interview section is conducted with open questions. It aims to explore more information from interviewees. Then, the interviewees focus on elaborating on the situation. It is not just "yes" or "no" answers for every point in the checklist. The interviewees are two GOP Officers in BMKG head office. They have experienced the management of GOP for nine years. There are 197 BMKG offices in Indonesia, and every office has at least one GOP Officer. However, the head office, where all transactions are collected and all BMKG GOP regulations are made, has ten officers. The result of the interviews completes the checklist.

D. Calculating Data Quality Maturity
Calculation of the maturity level of each component is done by summing the value of each level of maturity. Each component has a maximum maturity level (1). It is derived from the average value of the overall characteristics of erach component. If the characteristics are by BMKG practice, it is 1. Otherwise, it is 0. For example, in expectation component (H) with initial level (I), there are three characteristics (HI1, HI2, HI3). HI1 and HI3 are appropriate with circumstances in BMKG, then each value is 1. Meanwhile, HI2 is not fit, so the value is 0. Then, the expected component value for the initial level is as follows: The number of characteristics (3)

E. Recommendation Analysis Improved Data Quality Management
Based on the characteristic, the matrix shows the characteristics of points that have not been met by BMKG. These characteristics are mapped into the activities that need to be done according to the DQM in DMBOK rules. It expects the DQM performed by BMKG can get better in the future. For a better presentation of DQM in DMBOK mapping activity, the code is given for each activity starting from the code to develop and promote data quality awareness (DQM1) to monitor operational SOP and DQM performance (DQM12).

IV. RESULTS AND DISCUSSION
The results of the assessment are presented in Tables A3-A5 (see Appendix). The justification for condition is based on documents observation and interview. Reference number for every evidence for the document is presented in Table A6 (see Appendix). Meanwhile, the interview-based evidence is given a code W1 for the first interviewee and W2 for the second interviewee.
The maturity position of the DQM and the target of maturity are based on the position above the assessment of the current condition. The results of the assessment and target of maturity are presented in  expectation of data quality regarding data problems anticipation and reporting. The lacks of this component lies in the absence of methods for measuring business impact when data errors occur and the absence of benchmarks in the measurement of data quality. Measuring data error impact on business process is a necessary action to do especially in data interchange matter [26].

2) Dimension. Dimensional component assessment
shows that the management of GOP data quality already uses the dimension of data quality in general. However, there is no determination of data quality dimension in the form of regulation. The absence of rules by governing data quality dimensions leads to no reports of data quality. The absence is seen in the various ways of data quality assessment in each GOP Officer in head office. 3) Policy. The condition of existing policy components in BMKG and coordination do the policymaking process. The regulated policies include restrictions on access rights to data and historical data changes. Things that have not applied in BMKG are the SLA regarding the data quality, and there is still unsuitable standard notification for data transaction. Other policies that need to be considered are the handling of data problems and the certification process regarding the sources of data quality. 4) Procedure. The condition of GOP DQM for procedural components is done with coordination at the technical service unit level and head office level. It also includes coordination of data correction as and coordination related to data source in searching data. The search does not include syntax and data structures since it is the authority of the Ministry of Finance. Moreover, the validation and auditing have been done by involving several other applications. 5) Governance. Governance implemented in BMKG still has not applied the data steward, and there is no organizational structure to supervise data governance. However, to overcome the problem, it has been communicated to GOP Officers in technical service unit to head office. The GOP Officers have realized that the data problem is not only the problem of IT. Regarding appreciation of the GOP data management business, the Ministry of Finance has also held awards as a form of appreciation of the ministry/institution that has proper GOP management. 6) Standardization. The condition of standardization component can be seen in the existing of standard and definition of managed data and business terminology. The existence of reference data also supports it. GOP transaction data can be identified by referring to any information. The guidelines for data exchange are well organized and executed.
On the other hand, metadata management does not exist, and the master data are still managed with transaction data. 7) Technology. Technology components in DQM are illustrated by the availability of applications to find, match, and connect data. GOP Officers have also realized that the problem of data will impact the other parts. It also provides dashboard and reporting applications to support impact analysis caused by data errors. 8) Performance Management. Performance management conditions in BMKG GOP DQM has the absence of regional characterization as the impact of poor data quality, and the absence of profiling that can be used to identify data errors. Moreover, there is no framework to analyze the impact or detect data errors. Continous profiling is needed since the size of GOP data is big. Profiling process is performed while data are created or updated. Profiling also determines the common properties or heterogeneity of data, so that inconsistent data can be found easily [27]. Moreover, there is no framework to analyze the impact or detect data errors. BMKG and the Ministry of Finance must continue to update and give more rules as a foundation for improving the management of functional data quality.

V. RECOMMENDATION
The result of the assessment based on the compatibility with the condition of DQM also yields characteristics that have not fulfilled. The fulfillment of DQM characteristics is anticipated by the application of DQM activities on DMBOK. Table III shows the mapping of characteristic that has not met the DQM activity in DMBOK.
The most critical issue is DQM3, BMKG must promote the awareness of data quality to every employee  There is no documented data of quality dimension DQM2 HD4 The methods for assessing business impact is not available DQM5 Managed HM4 There is no scheduling of data quality assessment DQM8 Optimized HO1 There is no data quality benchmark DQM4 HO2 It is not associated with individual performance targets DQM11 HO3 The level of industry proficiency has not been used DQM5 There is no categorization of data quality problems DQM3 Defined DD3 The report on data quality measurement is not available DQM3 Managed DM1 There is no grouping of data quality dimensions to business impact DQM5 DM2 There is no report of data quality DQM4 DM3 There is no data steward DQM1 Optimized DO1 There is no SLA related to data quality DQM7 DO2 There is no SLA related to data quality DQM7 DO3 There is no definition of data quality dimensions DQM2 There is no certification process regarding data quality DQM6 KD4 There is no SLA about data quality DQM7 Managed KM3 DQM is missing DQM5 KM4 The policy has not driven performance management DQM11 KM5 There is no SLA about data quality DQM7 Optimized KO1 There is no automatic notification if there is any inappropriate data DQM10 KO2 It has not implemented a system with independent data governance DQM5 There is no working principle of data quality DQM2 Defined GD1 There is no organizational structure oversees data governance DQM1 GD2 There is no documentation of working principles and data governance DQM5 GD3 There is no standard data stewardship view DQM8 GD4 There are no SOPs in governing data governance DQM11 Managed GM1 There are no committees in dealing with data governance yet DQM1 GM2 It has not handled data governance DQM1 GM3 There is no SLA DQM7 GM4 There is no data governance framework DQM11 GM5 There is no report of data governance DQM11 There is no regional characterization of the impact of poor data quality DQM3 MR2 There is no profiling data DQM3 Defined MD1 There is no framework to analyze the impact DQM3 MD2 There is no data quality service component DQM7 Managed MM1 There is no data quality metrics DQM4 MM2 There are no determined data quality dimensions DQM2 using GOP data directly or indirectly. Informing them of the impact on data issues and giving socialization about the data quality issue are not only a technology matter. Next critical issue is DQM1. BMKG must identify the business usage of GOP data set to list potential anomalies. These anomalies must be analyzed with subject matter expert to determine if it is categorized as data flaw or not. They can evaluate the potential impact on business caused by that anomaly. DQM5 is another concern for BMKG. After the expectation of data quality is determined, the next stage is to set business rules related transactions. It is inputted into the system including giving notification to data steward if there is a transaction that has the potential to reduce the quality of data.  2). Thus, the average maturity level of 2.8. In other words, maturity is still at the level of repeatable to defined. Repeatable level shows that BMKG has essential organizational management and information sharing. BMKG also can recognize good practice and try to implement it in their process. However, it has limited documentation of processes, plans, standards, and practices.
The characteristic assessment also leaves 54 characteristics that still need to be a concern for BMKG to achieve the highest level of maturity. These characteristics are mapped into DQM in DMBOK activities as a recommendation for improving the maturity of GOP DQM. The most critical issue is DQM3, DQM1, and DQM5. There are many concerns in how BMKG delivers awareness according to data quality, and how it must identify which transaction that may cause a flaw in data and how to avoid it. BMKG may need to consider to start determining SLA for data quality to specify the organizational expectation for response and remediation. With SLA of data quality, BMKG can monitor the compliance of data to the organizational expectation, and how well the employee performs the procedure associated with data errors.
Research shows that Loshin's Data Quality Maturity model can be used as a measurement of maturity in DQM. Therefore, it is expected that further research can be done in the ministry or other government institutions, especially in the Ministry of Finance as the builder in GOP management and as an agency that develops applications of SIMAK-BMN and Inventory. Moreover, further research can raise the subject of information system aspects in assessing the maturity of DQM.

APPENDIX
The Appendices can be seen in the next page.      Data quality control also does SAIBA (7) 1 GM2 There is no handle on data governance (W2) 0 SM2 Reference data already exist (7) 1 PM3 The weakness of data is unknown since the beginning (W1) 0 GM3 There is no SLA (W2) 0 SM3 Data exchange standards are maintained (7) 1 PM4 There is a process of normalization (7) 1 GM4 There is no governance framework (W2)