Title: Prioritization of global software requirements' engineering barriers: An analytical hierarchy process
Abstract: IET SoftwareVolume 15, Issue 4 p. 277-291 ORIGINAL RESEARCH PAPEROpen Access Prioritization of global software requirements' engineering barriers: An analytical hierarchy process Muhammad Azeem Akbar, Corresponding Author Muhammad Azeem Akbar [email protected] orcid.org/0000-0002-6880-4991 Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China Correspondence Muhammad Azeem Akbar, Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China. Email: [email protected] Ahmed Alsanad, STC's Artificial Intelligence Chair, Department of Information Systems, College of Computer and Information Sciences, King Saud University, Riyadh 11451, Saudi Arabia. Email: [email protected] for more papers by this authorWishal Naveed, Wishal Naveed Punjab Information Technology Board, Lahore, PakistanSearch for more papers by this authorSajjad Mahmood, Sajjad Mahmood Information and Computer Science Department, King Fahd University of Petroleum and Minerals, Dhahran, Saudi ArabiaSearch for more papers by this authorSaima Rafi, Saima Rafi School of Computer Science and Technology, Chongqing University of Posts and Telecommunication, Chongqing, ChinaSearch for more papers by this authorAhmed Alsanad, Corresponding Author Ahmed Alsanad [email protected] Department of Information Systems, STC's Artificial Intelligence Chair, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi Arabia Correspondence Muhammad Azeem Akbar, Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China. Email: [email protected] Ahmed Alsanad, STC's Artificial Intelligence Chair, Department of Information Systems, College of Computer and Information Sciences, King Saud University, Riyadh 11451, Saudi Arabia. Email: [email protected] for more papers by this authorAbeer Abdul-Aziz Alsanad, Abeer Abdul-Aziz Alsanad Information Systems Department, College of Computer and Information Sciences, Imam Muhammad Ibn-Saud Islamic University, Saudi ArabiaSearch for more papers by this authorAbdu Gumaei, Abdu Gumaei orcid.org/0000-0001-8512-9687 Department of Information Systems, STC's Artificial Intelligence Chair, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi ArabiaSearch for more papers by this authorAbdulrahman Alothaim, Abdulrahman Alothaim Department of Information Systems, STC's Artificial Intelligence Chair, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi ArabiaSearch for more papers by this author Muhammad Azeem Akbar, Corresponding Author Muhammad Azeem Akbar [email protected] orcid.org/0000-0002-6880-4991 Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China Correspondence Muhammad Azeem Akbar, Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China. Email: [email protected] Ahmed Alsanad, STC's Artificial Intelligence Chair, Department of Information Systems, College of Computer and Information Sciences, King Saud University, Riyadh 11451, Saudi Arabia. Email: [email protected] for more papers by this authorWishal Naveed, Wishal Naveed Punjab Information Technology Board, Lahore, PakistanSearch for more papers by this authorSajjad Mahmood, Sajjad Mahmood Information and Computer Science Department, King Fahd University of Petroleum and Minerals, Dhahran, Saudi ArabiaSearch for more papers by this authorSaima Rafi, Saima Rafi School of Computer Science and Technology, Chongqing University of Posts and Telecommunication, Chongqing, ChinaSearch for more papers by this authorAhmed Alsanad, Corresponding Author Ahmed Alsanad [email protected] Department of Information Systems, STC's Artificial Intelligence Chair, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi Arabia Correspondence Muhammad Azeem Akbar, Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China. Email: [email protected] Ahmed Alsanad, STC's Artificial Intelligence Chair, Department of Information Systems, College of Computer and Information Sciences, King Saud University, Riyadh 11451, Saudi Arabia. Email: [email protected] for more papers by this authorAbeer Abdul-Aziz Alsanad, Abeer Abdul-Aziz Alsanad Information Systems Department, College of Computer and Information Sciences, Imam Muhammad Ibn-Saud Islamic University, Saudi ArabiaSearch for more papers by this authorAbdu Gumaei, Abdu Gumaei orcid.org/0000-0001-8512-9687 Department of Information Systems, STC's Artificial Intelligence Chair, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi ArabiaSearch for more papers by this authorAbdulrahman Alothaim, Abdulrahman Alothaim Department of Information Systems, STC's Artificial Intelligence Chair, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi ArabiaSearch for more papers by this author First published: 20 April 2021 https://doi.org/10.1049/sfw2.12022AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Abstract Software industry is adopting global software development (GSD) due to its potential to produce quality products at a lower cost. However, the GSD firms face many challenges that make development activities more complicated, especially related to the requirements engineering (RE) process. The objectives of this article are to investigate and prioritize the barriers faced by the GSD organizations during the RE process. First, we identified 17 barriers related to the RE process in the GSD projects. Next, the identified barriers were further validated with real-world GSD practitioners using a questionnaire survey. Finally, we applied the analytical hierarchy process to prioritize the investigated barriers with respect to their significance for the RE process in the GSD domain. The results show that coordination is the most significant barrier category for the RE process in GSD projects. Lack of standard and procedure for RE in GSD, lack of synchronized communication infrastructure, and lack of mutual understanding between the overseas RE teams are also high-ranked barriers for the RE process in GSD. The authors believe that the findings of this study will assist practitioners and researchers in developing effective strategies and plans for the successful implementation of the RE process in the GSD context. 1 INTRODUCTION Requirements engineering (RE) is the process of clarifying the goal by defining the needs of stakeholders and documenting these needs in such a way that can be analysed and implemented [1]. The RE process is an important phase of every software development process as it provides base for the subsequent software development activities. According to Jeremy [1], the RE process consists of five core phases, namely, requirements extraction or elicitation, requirements analysis and design, requirements specification, requirements validation, and requirements management. Over the last two decades, software firms have increasingly adopted the phenomena of global software development (GSD) [2, 3]. The GSD is a software development process in which development activities are carried out by experts who are based in different geographical locations around the globe [4, 5]. The GSD is an effective development paradigm as it facilitates to hire skilled human resources at a low cost [6]. By adopting the GSD, the development activities are carried out round the clock (24/7), which is significant for in-time project delivery. For example, Mighetti and Hadad [7] underlined that the GSD paradigm assists in developing quality software due to the availability of the latest technological tools in geographically distributed sites. They further indicated that across the globe, the development provides the opportunity to GSD firms to keep update with the international market. However, besides the advantages, GSD faces several barriers, especially in requirements elicitation and management activities [8, 9]. For example, Damian, et al. [10] Lloyd, et al. [9] and Khan, et al. [11] indicate that the RE faces a number of barriers in the GSD domain due to the teams distributed across the globe. Similarly, Hsieh [12] underlined the communication and collaboration barriers associated with the RE process due to geographical distance between the RE teams of different GSD sites. Furthermore, the RE teams face various barriers while deploying RE activities such as 'establishing relationships between geographically distributed team members', 'managing time differences', and 'cultural issues' [13, 14]. Significant work has been conducted to understand RE in the GSD [15-17]. However, no work has been carried out that identifies barriers that influence the RE process in GSD projects. Moreover, no studies have been conducted that empirically prioritize barriers associated with the RE process in GSD context. This research gap motivated us to conduct a study to better understand the barriers of the RE paradigm in the geographically distributed development environment. This study has two key goals: (i) to identify the RE barriers in the context of GSD by adopting the literature review approach and to validate them with real-world practitioners using a questionnaire survey; and (ii) to prioritize identified barriers by applying analytical hierarchy process (AHP). The prioritization of barriers will assist the practitioners to consider the high priority barriers that influence the RE process in GSD projects. The prioritization will also assist the practitioners to better understand, plan and manage RE activities in GSD context. To achieve the objectives of this study, we address the following research questions: [RQ1] What are the barriers faced by the GSD firms during the requirements engineering process, as reported in the literature survey? [RQ2] Do the real-world practitioners agree with the findings of the literature survey? [RQ3] How the investigated barriers are prioritized? 2 BACKGROUND AND MOTIVATION To develop software projects according to the expectations of the stakeholders, the RE process plays an important role. Various studies were carried out to address the challenges of the RE process. For example, Pandey, et al. [13] proposed a model for requirements elicitation. Pandey's model contained four core phases, namely: 'requirements elicitation and development', 'documentation of requirements', 'validation and verification of requirements', and 'requirements management and planning'. The model provides a brief description to collect and analyse the requirements according to the customers' expectations [3]. Beecham, et al. [18] introduced a requirements process improvement model, namely R-CMM. The R-CMM model contains a detailed description of the RE process. The maturity level of the R-CMM model enables software firms to assess and improve their RE activities. In the software industry prospective, the R-CMM also provides the best practices which are useful to address the faced barriers during the RE process [18]. Mellado, et al. [19] introduced a security RE process. This model is based on the common criteria (CC) (ISO/IEC 15408) [19]. This standard-based model has an agreement with the requirements security at the primary phases of software development is an iterative and systematic way by using a security resources repository; together with the combination of the CC (ISO/IEC 15408). This model is useful to address the barriers related to the RE process concerning the requirements security at the initial phase of the development life cycle [20]. Various other studies were conducted to address the challenges of the RE process [20-23, 25, 26]. A number of studies have also been conducted to better understand barriers associated with RE process in GSD projects. For example, Bhat et al. [27] indicated that RE teams working in client-vendor outsouricing projects face a different that are not addressed by traditional RE processes. The study findings indicate that shared goals, shared culture, shared process, shared responsibilyt and trust are best practices to successfully carry out RE process in client-vendor offshore outsourcing relationships. Damian and Zowghi [28] investigated RE challenges introduced by stakeholders who are geographically distributed in a multi-site organization. They found that remote communication, knowledge sharing, cultural diversity and time zone differences negatively impact the RE process in multi-site software development organizations. Simialarly, Alnuem et al. [29] applied structured and semi-strucutred interview techniques to understand RE challenges faced by the Saudi Arabian industry during GSD projects. They found documents management and competence management as key challenges faced by the Saudi Arabian organizations during GSD projects. Jain and Suman [30] carried out a systematic literature review on the GSD life cycle and reported that temporal, socio-cultural, linguisitic, communication and coordination are challenges associated with RE process of a GSD project. However, the findings of the above discussed studies indicate that no study has been conducted to prioritize the barriers associated with RE process in the GSD context. In this study, we apply the AHP to prioritize the barriers related to the RE process in GSD projects. 3 RESEARCH METHODOLOGY To meet the objectives of the study, we applied three different research approaches; (i) literature review, (ii) empirical study (questionnaire survey), and (iii) AHP. All the approaches are briefly discussed in subsequent sections and graphically presented in Figure 1. FIGURE 1Open in figure viewerPowerPoint Conducted research methodology. AHP, analytical hierarchy process; GSD, global software development; RE, requirements engineering 3.1 Literature review process To identify the barriers of the RE process from the existing literature, we used the literature review approach. 3.1.1 Data sources To extract the research articles related to study objectives, we applied search string on the most popular digital repositories. The databases were selcted using the suggestion of Khan, et al. [4], Lloyd, et al. [9] Niazi [31] that is, 'ACM Digital Library', 'IEEE Xplore', 'Wiley Inter-Science', 'Springer Link', 'Science Direct', 'IET Digital Library' and 'Google Scholar'. 3.1.2 Search string The search string was developed using keywords and their respeoctive synonyms. The Boolean operators "AND" and "OR" were used to develop search strings for different databases by concatenating keywrods and their respective synonyms. Since, different databases have unique format for writing and executing search strings, in this article, we report a generic structure of the search string. The format of the search string executed on each selected databases are given at: https://tinyurl.com/y944gj2a. An example of the search string is given below: ('barriers' OR 'obstacles' OR 'hurdles' OR 'difficulties' OR 'impediments' OR 'hindrance' OR 'challenges') AND ('requirements engineering' OR 'RE' OR 'requirements collection' OR 'requirements elicitation' OR 'requirements updating' OR 'requirements process assessment') AND ('global software development' OR 'GSD' OR 'distributed software development' OR 'offshore software development' OR 'outsourcing' OR 'multisite software development'). The search terms were collected based on our research experience and using the references, Shafiq et al. [2], Khan et al. [6], Khan et al. [4], Lloyd et al. [9], Niazi et al. [31]. The developed search string was carefully executed on the selected digital databases, and initially, a total of 823 research articles were collected. 3.1.3 Inclusion and exclusion critera Next, we used the inclusion and exculsion criteria to select primary publications for the study. The inclusion and execution criteria were developed by adopting the criteria published in the software engineering domain [4, 32]. The inclusion and exclusion criteria for the study are as follows: Inclusion Criteria: (1) Articles that are published as a journal paper, conference article or book chapter, (2) articles written in English language, (3) findings of the articles based on empirical investigation (4) include the most complete version of an article from duplicate studies. Exclusion Criteria: (1) The articles that do not focus on the requirement engineering in GSD context, (2) articles that do not address the research questions, (3) articles that do not explore the success factors of requirement engineering in GSD (4) articles written in other language than English. 3.1.4 Final selection of studies We have applied the steps of tollgate approach proposed by Afzal, et al. [33] to select the final primary studies. By executing the search string on the selected databases and after applying the inclusion and exclusion criteria, initially, 823 studies were extracted. By carefully applying the five steps of the tollgate approach [33], finally we have selected 29 studies (Figure 2). The list of the final selected studies is given in appendix-D. FIGURE 2Open in figure viewerPowerPoint The phase of the adopted tollgate approach The selected studies were carefully reviewed to extract the data to answer the RQs of the study. The first and second authors participated in the data extraction process. First, they extracted the barriers from the finally selected studies. Later, the remaining research team carefully reviewed the extracted data. We have further conducted the reliability test with the aim to check the inter-personal biasness. As a result, two independent external reviewers were requested for the reliability test. They have randomly selected seven articles and carefully performed all the phases of data extraction process. We calculated the non-parametric Kendalls coefficient of concordance (W) value to check the inter-rater agreement between the authors and external independent reviewers [34-36]. The value of W = 1 indicates a complete agreement and W = 0 shows a complete disagreement. The calculated results of reliability test for seven selected articles shows that W = 0.87 (p = 0.003), which indicted significant similarities between the findings of the authors and the external reviewers. The selected studies were further analysed concerning the publication years and the adopted research methodologies (Figure 3). According to the summarized results there is an increasing trend of publication in recent years. It is also observed that 19% 'questionnaire surveys' (QS), 21% 'case studies' (CS), 12% 'content analyses' (CA), 13%) 'grounded theory' (GT), 15 (12%) 'action research' (AR), and 23% of the studies used mixed methods (MM). Hence, it is observed that MM, CS and QS are the most widely adopted research approaches. FIGURE 3Open in figure viewerPowerPoint Temporal distribution and the research methodologies used in selected studies. AR, action research; CA, content analyses; CS, Case studies; GT, grounded theory; MM, mixed methods; QS, questionnaire survey 3.2 Empirical study To validate the barriers explored through the literature review, the questionnaire survey approach was applied. The phases adopted to conduct data collection form from the respondents are presented in the subsequent sections. 3.2.1 Survey questionnaire development In light of the literature review investigations, we developed an online survey instrument to explore the barriers of the RE process in GSD. The survey method can obtain information from a large targeted community [4]. Also, we can obtain data through a survey method that is hard to obtain using other observational methods [37, 38]. We created a close-ended questionnaire to collect information from the RE and GSD practitioners. The survey questions are based on the 17 barriers that are investigated in the literature review. We used a five-pointed Likert scale with the following feasible reactions: 'strongly agree, agree, neutral, disagree, and strongly disagree'. According to Finstad [39], there is no critical drawback in using the 'neutral' option on a five-pointed Likert scale. Moreover, most of the researchers verified that it is a justifiable reaction to respond as neutral to any statement or subject matter [4, 40]. If there is no 'neutral' option, it might force respondents to make a negative or positive decision which makes the result biased. 3.2.2 Questionnaire pilot assessment A pilot assessment of the survey instrument was conducted with five software engineering experts to get their feedback regarding the design, measurement scale, readability, and understandability of the survey questionnaire. Three experts were from the industry (i.e. belong to DCSL SOFTWARE, ITECHART GROUP, and RADIX WEB) and the other two were from academia (i.e. 'Chongqing University of Posts and Telecommunications, Chongqing, China' and 'Nanjing University, Nanjing, China'). These experts were selected as they have direct experience in the RE and GSD paradigm. They provided suggestions to modify the structure of the questionnaire by listing all the identified barriers in tabular form. Moreover, they recommended to rephrase some of the barriers and make them more understandable. We finally updated the survey questionnaire based on the feedback of the experts and the sample of the updated questionnaire survey is given in Appendix-A. We assured the survery participants that the survey data would only be used for research purposes, and the identity of the survey respondents will not be disclosed to any organization or any individual in any circumstances. 3.2.3 Data sources The purpose of this study is to identify the barriers of the RE process in GSD context. Hence, it was important to gather information from various professionals with relevancy to RE experiences in the GSD domain. The survey participants are invited using the snowball strategy [41]. All participants were invited via different sources, for example Email, Facebook, LinkedIn, and through their professional contacts. We developed a link to the questionnaire survey and sent it to the different in-contact practitioners to participate in the survey. In addition to the questionnaire survey link, an invitation letter was also attached to the request to invite their team members to participate in the survey study. The data was collected from March 2019 to June 2019. We collected a total of 56 filled questionnaires from the experts. Although feedback was manually checked to identify incomplete entries, none was identified. We found that most of the developing firms are located either in Europe or Asia. Survey respondent's positions are ranged from software developers to software project managers. All of them have vast experiences in the RE process as well as in GSD. The detail demographics of the respondents are provided in Appendix-B. 3.2.4 Survey data analysis The survey responses are summarized using the frequency analysis method. The frequency analysis is an effective way to investigate the variable groups and for ordinal and numeric information [42]. The frequency analysis is performed, and the summarized results are presented in the tabular form. The same analysis approach has been adopted by various existing studies, for example, [4, 37]. 3.3 Analytical hierarchy process AHP is a multi-criteria decision-making process used for prioritizing the influencing factor(s) of a specific project. Various studies have already been adopted in this method to inline the complex decision-making problems and prioritize the factors of different other software engineering domains, for example [43, 44]. The same approach is adopted in this study to prioritize the investigated barriers of the GSD RE process with respect to their significance. Table 1 shows the core phases of AHP. TABLE 1. List of the AHP phase Sr. no Phases of AHP Phase 1 Classifying the goal, categories (factors), and their corresponding barriers (sub-factors) (as shown in Figure 4). Phase 2 Develop the pair-wise comparison matrix of barriers based on the experts' opinions. Phase 3 Calculate the priority weight of each barrier category and sub-category by using pair-wise comparisons. Phase 4 Check the consistency of the judgements. Phase 5 Rank the barriers in their corresponding categories (local ranking of barriers). Phase 6 Determine the global weights of barriers (final rank of barriers). Phase 7 Prioritizing the barriers. Abbreviation: AHP, analytical hierarchy process. FIGURE 4Open in figure viewerPowerPoint Hierarchy of the analytical hierarchy process 3.3.1 Classifying the goal, categories (factors), and their corresponding barriers (sub-factors) In this phase, we decomposed a complicated problem in interlined decision elem -making problem is divided at three levels in the hierarchy structure (as shown in Figure 4). In this hierarchy structure, the basic goal is indicated at the first level; the subcategories are indicated at the second and third levels respectively. 3.3.2 Development of pair-wise comparison matrix of barriers based on the expert's opinions At this level, we used a pair-wise comparison technique [46, 47]. The purpose of adopting the pair-wise technique is to calculate the weight (W) of each reported barrier and category of barriers. In a pair-wise approach, all the barriers and their categories are compared with each other at every level based on their effect on the RE process in GSD. Based on the comparison, the pair-wise matrix was developed, and weight (W) was calculated. The weight (W) indicated the priority level of each barrier within the category and among the categories. However, experts' panel (discussed in section 3.2.1) was involved in checking the comparison score of barriers within the category and among the categories. The questionnaire used to report the responses of the experts' panel is available at the link: https://tinyurl.com/yachfe6h. For reporting the respondents' opinions, we used a 'standardized nine-pointed comparison scale' (Table 2). The comparison matrix [44, 47] was developed for each reported barrier and their categories, as shown in matrix-A (n × n). TABLE 2. Description of the intensity scale Description Significance intensity Equally important 1 Moderately important 3 Strongly more important 5 Very strongly more important 7 Extremely more important 9 Intermediate value 2, 4, 6, 8 Matrix - A ( n×n ) [ 1 a 12 a 1 n a 21 1 a 2 n an 1 an 2 1 ] We further explain the pairwise comparison Matrix-A as follows: We compare the two barriers, that is, BA1 and BA2 with respect to their degree of significance for RE in GSD. If BA1 has 3 greater importance than BA2, then BA2 is one-third with respect to BA1. Using the same practice we have developed the pairwise comparison matrixes for all the barrier factors and their categories in Section 4.3. 3.3.3 Calculation of priority weight for each barrier category and sub-category by using pair-wise comparisons To calculate the priority weight (W) of the investigated barriers and their categories, all the developed matrixes of the barriers were normalized. For the normalization process, all the barriers were divided by the sum of their column and calculate the average of each row. The calculated average score of each row is the priority weigh of each barrier (Section 4.3). 3.3.4 Checking the consistency of the judgements This phase is performed to check the consistency of the pairwise comparison. To do this, the following equations are used: Consistency Index ( CI ) = ( λ − n ) ( n − 1 ) where, λ = [sum of each column of pairwise comparison matrix] × Weight (W) vector and n = number of criteria. Consistency Ratio ( CR ) = CI RI where, RI = Random consistency index with respect to matrix size (Table 3). TABLE 3. RI with respect to matrix size Size of matrix 1 2 3 4 5 6 7 8 9 10 RI 0 0 0.58 0.9 1.12 1.24 1.32 1.41 1.45 1.49 Abbreviation: RI, random consistency index. 4 FINDINGS OF THE STUDY The results and analysis are presented in this section. 4.1 Literature review (RQ 1) Due to potential business gains, software firms are motivated to transform their business activities from colocated (single site) development environment to GSD [6]. However, adoption of GSD is not straightforward. There are various barriers are associated with GSD, particularly for the RE process [40]. A case study conducted by Regnell et al. [48] shows that the barriers of the RE process are one of key causes for project failure. Prikladnicki et al. [8] and Nicolás et al. [49] indicated that there is a huge research gap in the RE process in the GSD context. This encouraged us to explore the RE process barriers faced by GSD practitioners. We have reviewed the existing literature and observed that various researchers discussed the barriers f