Use of Search Tools in Software Development:

A Study of Microservice-Based Team Projects

Yi Meng Lau

, Christian Michael Koh

and Lingxiao Jiang

School of Computing & Information Systems, Singapore Management University, Singapore

Keywords: Software Development, Microservices Architecture, Search Engines, Large Language Models,

Collaborative Learning, Peer Learning, Self-Directed Learning.

Abstract: Universities are increasingly integrating real-world projects into software engineering curricula to prepare

students for careers involving complex concepts like Microservices Architecture (MSA). Students frequently

struggle with such concepts within limited class time and turn to various search tools and online resources for

additional help. Search tools are also widely used in the software development industry. While search engines,

like Google and Yahoo!, can provide quick solutions, they pose the risk of information overload. Large

Language Models (LLMs) such as ChatGPT, offer the advantage of delivering more precise answers. Studies

have shown that LLMs can comprehend codes, assist in system architectural design, and suggest solutions,

potentially enhancing the learning experience for students. This study aims to determine how students make

use of search tools for their team projects in a software development course that teaches MSA. It will also

explore if search tools can enhance learning in team projects by facilitating collaborative, peer, and self-

directed learning, and propose methods to address any limitations.

1 INTRODUCTION

Learning in software engineering has extended beyond

attending instructor-delivered lessons focusing on

knowledge acquisition and computational thinking.

Universities are incorporating real-world practical

projects into their software engineering and STEM

courses to align theoretical knowledge with workplace

demands, integrating technical and soft skills (Ceh-

Varela, Canto-Bonilla, & Duni, 2023; Guo, Saab, Post,

& Admiraal, 2020). These projects provide students

with the opportunity to explore technical concepts,

working individually and in teams to prepare them for

their future careers (Podeschi, 2016; Ceh-Varela,

Canto-Bonilla, & Duni, 2023). For example, many

universities have included Microservices Architecture

(MSA) (Bogner, Fritzsch, Wagner, & Zimmermann,

2021) in their software engineering curricula (Lau,

Koh, & Jiang, 2024). MSA involves breaking down the

functions of an application into small, self-contained

code modules called microservices that are highly

reusable for agile software development (Bucchiarone,

https://orcid.org/0009-0000-7470-2556

https://orcid.org/0009-0004-1149-7478

https://orcid.org/0000-0002-4336-8548

Dragoni, Dustdar, Larsen, & Mazzara, 2018). These

concepts can be abstract and hard for students to

understand within limited class time.

Studies have shown that software developers and

students extensively leverage online resources to learn

new programming languages and technical concepts,

clear doubts, and augment their existing knowledge

(Brandt, Guo, Lewenstein, Dontcheva, & Klemmer,

2009; Liu, et al., 2021). Search engines such as Google

and Yahoo! are widely used by industry practitioners

to source and locate information (Rangaswamy, Giles,

& Seres, 2009). Likewise, online resources are

valuable information sources for students working on

software projects, especially when they are tackling

new or unfamiliar topics (Griffiths & Brophy, 2005;

Walraven, Brand-Gruwel, & Boshuizen, 2009). The

abundant online resources available in software

development cater to students seeking to learn

programming at both novice and advanced levels (Lu

& Hsiao, 2017).

While search engines have been the primary tools

for students to quickly and conveniently start their

232

Lau, Y. M., Koh, C. M. and Jiang, L.

Use of Search Tools in Software Development: A Study of Microservice-Based Team Projects.

DOI: 10.5220/0013206000003932

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 1, pages 232-242

ISBN: 978-989-758-746-7; ISSN: 2184-5026

information searches (Liu, Zamir, Li, & Hastings,

2018), search results can be overwhelming or

unfavorable. Students face challenges of information

overload, uncertainty about which keywords to use,

and the need to spend more time reading and

interpreting the results (Griffiths & Brophy, 2005;

Hahnel, Goldhammer, Kröhne, & Naumann, 2018;

Mahdi, Ahmad, Ismail, Natiq, & Mohammed, 2020).

This has evolved with the introduction of Large

Language Models (LLMs), which enable more

conversational queries with more precise answers.

LLMs can enhance productivity in software

development through features like auto-completion,

code suggestions, and interactive chat dialogues

(Valový & Buchalcevova, 2023). Software

engineering graduates will increasingly find

themselves in LLM-driven environments, suggesting a

need for software engineering education to adapt and

innovate to accept the use of LLMs (Kirova, Ku,

Laracy, & Marlowe, 2024).

Given the complex nature of software

development, students are likely to depend on search

engines or LLMs (collectively called “search tools” in

this paper) for their courses. Rather than viewing

search tools as merely a tool, the search process can

also be part of learning (Brandt, Guo, Lewenstein,

Dontcheva, & Klemmer, 2009; Lu & Hsiao, 2017).

Online searches seem like an individual activity

(Ghosh, Rath, & Shah, 2018).

This study aims to determine how students use

search tools for their team projects in a software

development course that teaches MSA, where students

need to learn and practice breaking down a monolithic

application into small, self-contained, reusable

microservices, developing the solution as an assembly

of microservices, and deploying the tested solution in

containerized environments. We will also evaluate if

search tools can be integrated as learning resources to

enhance the student’s learning experience and facilitate

collaborative, peer, and self-directed learning through

a proposed framework.

Since teamwork is essential for software

developers to collaboratively analyze problems from

multiple perspectives (Shuto, et al., 2016), we will also

examine the limitations of search engines and LLMs in

facilitating collaborative efforts (Raibulet & Fontana,

2018) and suggest methods to address these challenges.

We gathered feedback through an online survey

from students enrolled in a software development

course. With the insights from the survey, we

developed a web-based prototype to address these

gaps. The prototype is piloted with a smaller group of

students.

The study seeks to address the following research

questions:

 RQ1: Are students using search engines or

LLMs in their team projects?

 RQ2: Can search engines enhance learning in

team projects by facilitating collaborative,

peer, and self-directed learning?

 RQ3: Can LLMs enhance learning in team

projects by facilitating collaborative, peer, and

self-directed learning?

 RQ4: If there is an application that can

integrate the benefits of various search tools,

what additional features would students like to

have?

The rest of this paper is organized as follows:

Section 2 shares related works. Section 3 describes the

proposed framework. Section 4 presents the approach

and findings of this study. Section 5 discusses the

limitations of this study that present potential areas for

future work. Section 6 concludes.

2 RELATED WORKS

2.1 Software Development Projects

The teaching of software engineering courses is

complex as they involve a wide range of concepts,

algorithms, software design principles, and tools (Ceh-

Varela, Canto-Bonilla, & Duni, 2023). Software

development courses tend to emphasize technical skills

at the expense of soft skills (Tubino, Morgan, Wood-

Bradley, & Cain, 2023).

Industries have cited soft skills as essential work-

ready skills and found that novice software developers

are reportedly disadvantaged by their lack of soft skills

(Christensen & Paasivaara, 2022). Software

developers are expected to collaborate effectively in

teams to solve real-world problems (Kuusinen &

Albertsen, 2019) prompting universities to introduce

team-based projects to address this need (Hamer,

Quesada-López, Martínez, & Jenkins, 2021;

Christensen & Paasivaara, 2022). Team projects have

proven effective in improving technical proficiency

and soft skills and encouraging students to take

responsibility for their learning (Ceh-Varela, Canto-

Bonilla, & Duni, 2023). To encourage collaboration

and teamwork, universities have explored the use of

industry tools like Microsoft Project, GitHub, and

Trello in software development courses team projects

(Raibulet & Fontana, 2018; Macak, Kruzelova, Chren,

& Buhnova, 2021; Christensen & Paasivaara, 2022).

Use of Search Tools in Software Development: A Study of Microservice-Based Team Projects

233

A challenge instructors often face in team projects

is to assess individual student contributions versus

team contributions and to establish effective feedback

mechanisms to track students’ project development

and facilitate timely interventions (Hamer, Quesada-

López, Martínez, & Jenkins, 2021). Git logs have been

used and analyzed to track team progress and

understand the team dynamics (Hamer, Quesada-

López, Martínez, & Jenkins, 2021; Macak, Kruzelova,

Chren, & Buhnova, 2021).

2.2 Searching as Part of Learning

Ghosh et al. share that learning is a significant outcome

of searching (Ghosh, Rath, & Shah, 2018). As we seek

information to address specific problems, we often

explore related areas (Rieh, Collins-Thompson,

Hansen, & Lee, 2016). The consideration of keyword

choices and relevance assessments are cognitive

processes akin to thinking and learning (Vakkari,

2016).

Online searches are a three-stage search process of

formulating the query, selecting sources, and

interacting with them (Vakkari, 2016) exhibits

parallels between search behavior and learning

behavior suggesting that it is critical learning rather

than passive reception (Rieh, Collins-Thompson,

Hansen, & Lee, 2016). These search activities allow

individuals to integrate pieces of information, create

new insights, evaluate, and apply knowledge, and

reconstruct their understanding of the search topic

(Vakkari, 2016; Ghosh, Rath, & Shah, 2018). At the

same time, students learn to recognize and appreciate

diverse perspectives and viewpoints (Rieh, Collins-

Thompson, Hansen, & Lee, 2016).

2.3 Use of Search Engines in Software

Development

Search engines have become a key source of

information for industry practitioners, including those

in software development (Rangaswamy, Giles, &

Seres, 2009; Sadowski, Stolee, & Elbaum, 2015).

Developers do code searches to increase productivity

and quality. It is observed that 74% of the time these

searches are queried via natural language, although

recent advancements in machine learning aim to

improve search accuracy (Liu, et al., 2021). Brandt et

al. share that developers often concurrently engage in

web-based research, learning, and code writing,

adopting a just-in-time approach to learn new skills or

refresh their memory on existing context or syntax

(Brandt, Guo, Lewenstein, Dontcheva, & Klemmer,

2009).

With the sheer amount of information available

online, code searching is not limited to finding reusable

codes but also understanding what a particular piece of

code is doing, or fixing a bug (Sadowski, Stolee, &

Elbaum, 2015; Grazia & Pradel, 2023). Search engines

have the challenge of having a vast amount of available

information, complicating the task of locating relevant

data (Mahdi, Ahmad, Ismail, Natiq, & Mohammed,

2020). Researchers have investigated various methods

to mitigate information overload with algorithms,

systems, filtering tools, and custom querying

languages (Mahdi, Ahmad, Ismail, Natiq, &

Mohammed, 2020; Grazia & Pradel, 2023).

2.4 Use of Large Language Models in

Software Development

Large Language Models (LLMs) have transformed

software development by assuming traditional

functions previously completed on search engines. The

study by Nam et al suggests that LLMs aid in code

comprehension and suggest codes and API endpoints

during application development (Nam, Macvean,

Hellendoorn, Vasilescu, & Myers, 2024). Researchers

have attempted to use LLMs to generate code

explanations and compile them into an e-book for web

software development to determine whether these

explanations generated by LLMs are useful for

learning (MacNeil, et al., 2023). For more advanced

topics such as software architecture designs, bots are

developed using ChatGPT to help novice software

architects gain experience through artificial

intelligence decision support to provide rapid

architecting software systems (Ahmad, et al., 2023).

LLMs like ChatGPT have significantly impact

STEM education. Beyond serving as an information

repository, ChatGPT functions as a practice question

bank, revision tool, code debugger for students, or even

as an educator lesson plan assistant (Banerjee,

Srivastava, Adjeroh, Reddy, & Karimian, 2023). By

adapting to changes in question structure, it is a

valuable tool for both students and instructors rather

than a threat to the education system (Banerjee,

Srivastava, Adjeroh, Reddy, & Karimian, 2023).

While students find that these tools contribute to

their learning and have become an integral part of

coding, they agree that fundamental concepts in

software engineering should be learned without relying

on AI technologies (Valový & Buchalcevova, 2023;

MacNeil, et al., 2023).

CSEDU 2025 - 17th International Conference on Computer Supported Education

234

3 PROPOSED FRAMEWORK TO

EVALUATE SEARCH TOOLS

Students use search engines or LLMs to support their

learning (Rieh, Collins-Thompson, Hansen, & Lee,

2016). However, the effectiveness of these search tools

in supporting teamwork, particularly in collaborative,

peer, and self-directed learning among students

working on microservice-based team projects, has not

been studied. Figure 1 presents the proposed

framework, along with the team project schedule.

Search tools can be used throughout the team project

phases.

Collaborative learning happens when students

work together to solve a problem with shared

responsibilities (Laal & Laal, 2012; Loes, 2022).

Through this interaction, students teach and learn from

each other, leading to peer learning (Topping, 2005;

Lane, 2016). This encourages self-directed learning as

students take initiative, identify resources, and develop

strategies to contribute effectively to the group

(Topping, 2005). Collaborative and peer learning

enhance teamwork and soft skills that students can

develop while working together (Lane, 2016), essential

for the software development industry (Kuusinen &

Albertsen, 2019; Christensen & Paasivaara, 2022;

Ceh-Varela, Canto-Bonilla, & Duni, 2023).

In collaborative learning, we examine if search

tools can facilitate the following in students: (1)

knowledge and ideas exchange in “Group Activity”,

which helps to strengthen their comprehension of the

subject matter and hone their (2) “Communication”

skills by sharing perspectives and articulating thoughts

as they work toward a (3) “Common Goal”. They are

encouraged to (4) “Provide Feedback” as part of (5)

“Active Participation” in the learning process,

contributing their viewpoints and insights, rather than

passively receiving information (Laal & Laal, 2012;

Loes, 2022).

In peer learning, we examine if students can (1)

"Gain New Perspectives" by reviewing problems and

exchanging information based on their peers’ findings.

This mirrors the process of learning through searches,

where students learn to appreciate diverse perspectives

and develop flexible thinking skills and the ability to

construct reasoned conclusions (Rieh, Collins-

Thompson, Hansen, & Lee, 2016). This can potentially

(2) “Accelerate their Learning” process instead of

searching for data individually. Students will benefit

from (3) “Giving and Receiving Feedback” of search

results with their peers, enabling them to be (4)

“Actively Engaged” in the learning process (Topping,

2005).

In self-directed learning, students take ownership

of their learning (Loyens, Magda, & Rikers, 2008)

which is a necessary skill for a software developer. The

concept of information searching, querying, and

evaluating search results aligns with self-directed

learning (Loyens, Magda, & Rikers, 2008; Vakkari,

2016). Students need to be (1) “Ready to Learn” and

aware of their “Learning Needs” to know what and

how to query. To reach their (3) “Individual Goals”,

students display (4) “Active Involvement” in

understanding, analyzing, and evaluating search

results (Rieh, Collins-Thompson, Hansen, & Lee,

2016), creating new knowledge that can contribute to

their team.

Figure 1: Proposed Framework to Evaluate Search Tools for Collaborative, Peer, and Self-Directed Learning.

Use of Search Tools in Software Development: A Study of Microservice-Based Team Projects

235

4 STUDY APPROACH AND

FINDINGS

The pilot study was conducted with Information

Systems students enrolled in the Year 2 undergraduate

Enterprise Solutions Development course. At the end

of their team project, students complete a voluntary

survey to provide their feedback. The response rate is

59.1% (71 out of 120 students).

With insights from this survey, we identified the

gaps in search tools and developed a prototype to

address these gaps. Eight students separated into two

groups were involved to test out the prototype in a 4-

hour face-to-face session. Students completed a short

survey after that which helped us verify our

understanding of the identified challenges.

Both surveys included Likert scale questions where

1 represents the least favorable response and 10

represents the most favorable, open-ended, and multi-

select questions. Both surveys are endorsed by the

university’s Institutional Review Board. Numbers in

brackets in subsequent sections represent the average

Likert scores for the questions.

4.1 Enterprise Solutions Development

Course Team Project

The Enterprise Solutions Development (ESD) course

explores how enterprises adapt applications to meet

evolving organizational and customer needs. While

monolithic applications were traditionally favored for

their simplicity, the need for faster releases has led

organizations to adopt microservices architecture.

Figure 2: ESD Course Team Project Schedule.

In this course, students design and implement MSA

using tools like Flask and Flask-SQLAlchemy for

Python-based microservices and databases. Key topics

include communication patterns (e.g., HTTP, AMQP),

data transformation (e.g., JSON, XML), web interfaces

(e.g., HTML, CSS, JavaScript), Docker deployment,

and API Gateway management (e.g., Kong). The

course is assessed by class participation, quizzes, a

team project and an examination The course schedule

is provided in Figure 2.

The team project requires students to work in

groups of 5-6 to identify a real-world problem and

build a solution using concepts and technologies taught

in the course. Their solution must include at least three

user scenarios, integrating both complex and simple

microservices with independent data stores. It should

involve third-party service calls, establish HTTP or

AMQP communication between microservices, and

feature a web-based user interface. The microservices

must be containerized and may be managed via

Kubernetes. Project teams must incorporate at least one

technology not covered in the course, such as using a

new programming language or a database type.

Students are graded as a team and individually based

on their contributions.

Due to limited class time, although all the technical

concepts needed for the project solutions were

introduced in class, they were not covered in detail.

Students were encouraged to read additional self-study

materials beyond class time and use various search

tools as they like.

4.2 Use of Search Engines or LLMs

Survey results showed that 67 (94.4%) students use

LLMs or search engines for their team projects. Four

students (5.6%) indicated they do not use either, as

they prefer to get help from their peers and reason that

they do not trust LLMs’ coding abilities.

In their use of search engines, students indicated

that the results from search engines often provide a

broad perspective of information (7.328) (see SE1 in

Figure 3) and require them to discern applicable

information (7.657) (see SE2 in Figure 3). They need

precise keywords and multiple queries (8.149) (see

SE3 in Figure 3) to find specific information. Some

students have difficulty remembering what was

previously queried (6.881) (see SE4 in Figure 3).

Students prefer LLMs’ search results over those

of search engines (7.925) (see L1 in Figure 4). On

some occasions, LLMs provide a generalized

perspective (8.164) (see L2 in Figure 4) and require

well-crafted prompts to get the desired results (8.746)

(see L3 in Figure 4), but their natural language

capability simplifies prompt creation (8.284) (see L4

in Figure 4). LLMs facilitate easier revisiting of prior

search results (8.358) (see L5 in Figure 4).

For their project requirements, students prefer the

results from LLMs, as they can identify different

personas and how these personas will interact with

the application they intend to build (7.821). Students

CSEDU 2025 - 17th International Conference on Computer Supported Education

236

can leverage the results of LLMs to identify required

microservices for their project (7.776). The code

structures offered by LLMs provide a good starting

point for their solutions (7.970) and accelerate their

development process (7.806).

Answer to RQ1: The survey data shows that

students use search tools in their projects with a

higher preference for LLMs.

4.3 Search Engines for Collaborative,

Peer and Self-Directed Learning

Next, we evaluate the effectiveness of search engines

for students in facilitating learning in their team

projects with the proposed framework in Figure 1.

The team project itself has created a “Common

Goal”, which is to be completed collaboratively, with

each student contributing their share before

integrating these contributions into a final solution.

This is their “Group Activity”.

While the search engine is not a tool to create a

“Common Goal”, students use search engines to find

relevant information and resources to help them solve

problems. From the response to the open-ended

question in the survey, students mentioned that the

first “Group Activity” is for all team members to

search for ideas for their team project.

The broad perspective of information (7.328) (see

SE1 in Figure 3) from search results serves as a

valuable starting point to facilitate knowledge and

idea exchange within the team. However, most

students will conduct search processes individually

(7.394) (see SE6 in Figure 3), making sense of the

data individually before using external tools such as

Google Docs to share with teammates (7.239) (see

SE5 in Figure 3) which are usually copying and

pasting of search results without much context, with

little “Communication” among the team. “Active

Participation” primarily occurs through their

contributions in Google Docs and verifying the search

results of their teammates (7.881) (see SE7 in Figure

3) to “Provide Feedback”.

In peer learning, sharing search results allows

team members to “Gain New Perspectives” from the

findings and can potentially “Accelerate Learning”. If

one teammate is unable to find information through

their search, another may have the answer. “Giving

and Receiving Feedback” can be achieved by

verifying each other’s search results via external tools

(7.881) (see SE7 in Figure 3). Evidence of being

actively engaged can be seen through the history of

shared document tools.

Students must contribute to their team. Their

“Individual Goals” are to complete their part of the

project before integrating it with others. Their

“Learning Needs” are met when they identify the

resources necessary to complete their tasks. By

independently assessing the validity of search results

(7.657) (see SE2 in Figure 3), creating precise

keywords (8.149) (see SE3 in Figure 3), and

overcoming challenges such as retracing previous

searches (6.881) (see SE4 in Figure 3), they

demonstrate “Active Involvement”. Their openness

to feedback indicates their “Readiness to Learn”.

Figure 3: Results from the Survey on the Use of Search Engines.

Use of Search Tools in Software Development: A Study of Microservice-Based Team Projects

237

Students noted that search engines often fail to

resolve bugs or explain issues, requiring them to

adapt code to fit project specifics, which highlights

another aspect of self-directed learning.

Answer to RQ2: Search engines often lack an

effective means for team members to communicate

easily, provide feedback to one another, and share

information. Tracing and sharing of search results can

be challenging. However, search engines do enhance

learning in other areas.

4.4 Large Language Models for

Collaborative, Peer, and

Self-Directed Learning

The effectiveness of LLMs in facilitating students'

learning in their team projects is also evaluated using

the proposed framework in Figure 1.

Like search engines, the team project serves as a

“Common Goal” and creates opportunities for

“Group Activity”. The scenarios generated by LLMs

provide a good starting point for project teams to

define their user scenarios and the microservices

needed to develop solutions (7.493) (see L9 in Figure

4), as students work toward their “Common Goal”.

Although search results can be shared through the

URLs provided by LLMs, students are using external

tools to “Provide Feedback” to their team members,

which they value (8.106) (see L8 in Figure 4). One

concern that students have is that the URLs will not

contain information from future searches that

continue later (6.896) (see L6 in Figure 4). While

LLMs do facilitate some forms of “Communication”

through sharing, “Active Participation” still relies

heavily on contributions to shared documents,

although team members can be actively searching for

information individually via the LLMs.

Peer learning supported by LLMs resembles

search engines but enhances the sharing of search

results by providing documented steps on how each

result is derived, making the process more accessible.

LLMs can facilitate self-directed learning. With a

clear “Individual Goal” it defines the “Learning

Needs”. Students use LLMs to find information to

solve their problems (7.925) (see L1 in Figure 4). By

crafting well-structured prompts (8.746) (see L3 in

Figure 4) in natural language (8.248) (see L4 in Figure

4) and assessing the validity of LLMs’ responses

(8.164) (see L2 in Figure 4), students become

“Actively Involved” in their learning. LLMs also

facilitate easy retrieval of past searches (8.358) (see L5

in Figure 4), helping students plan their project’s next

steps efficiently. They appreciate teammate validation

of their search results (8.106) (see L8 in Figure 4)

indicating their “Readiness to Learn”.

Answer to RQ3: The survey results show that

LLMs contribute to self-directed learning with their

natural language prompts with better results.

Although LLMs provide an easier method for sharing

and tracking back search results, they lack the ability

for team members to communicate and provide

feedback to one another.

Figure 4: Results from the Survey on the Use of Large Language Models.

CSEDU 2025 - 17th International Conference on Computer Supported Education

238

4.5 Additional Features that Can

Complement Search Tools

The survey asked students if an application that

integrates the benefits of search tools would assist

them in their team project and what additional

features they would like to have.

Students expressed the need for a way to integrate

and share search results. Search results like the initial

definition of business scenarios related to the

problem, along with the structure of API endpoints

and code architecture, were highlighted as

particularly useful (7.985). Additionally, students

noted that providing a way to offer timely feedback

in a consolidated manner (7.803) was particularly

useful. Given the complexity of MSA solutions,

which require extensive teamwork, such features

would be highly beneficial.

The ability to keep track of search queries

throughout the development process (7.923) would

allow teams to easily refer to previous search results,

providing valuable resources for fixing bugs or

finding solutions. A means of creating documentation

alongside the search results would also be useful

(8.721).

Answer to RQ4: Students have requested an

easier way to integrate their search results and share

them with the team. They also want the application to

provide timely feedback, allow them to track all

queries as they build the application, and assist with

documentation throughout the development process.

4.6 Addressing the Limitations of

Search Tools with Prototype

The survey results revealed that search tools lacked

features to effectively facilitate “Communication”,

“Providing Feedback” and “Active Participation”

within the team. Together with the suggested

additional features, we developed a prototype to

address these limitations.

The prototype is developed as a web application

hosted on Amazon Web Services (AWS) which is

query ChatGPT-3.5 via available APIs. ChatGPT was

chosen as the search tool because it has available APIs

and shares similarities with the results from search

engines and LLMs.

The prototype allows the team leader to define the

team’s problem statement (“Common Goal”) at the

outset. At any stage, students can seek feedback or

assistance from their team members (“Timely

Feedback on”) and contribute to the documentation of

the application they are developing. Search results are

tracked and made available to the team, promoting

transparency (“Shared Search Results”). This approach

fosters “Active Participation” within the project team.

Figure 5 shares the screenshots of the prototype.

Figure 5: Sample Screenshot of the Prototype.

Use of Search Tools in Software Development: A Study of Microservice-Based Team Projects

239

4.6.1 Evaluating the Prototype

A small group of 8 students from the same course was

engaged to test the prototype by building a scaled-

down version of their team project. They provided

feedback on the prototype through a short survey. The

survey results are presented in Table 1.

Table 1: Survey Results for the Prototype.

Survey Questions (n=8)

Average

(Std Dev)

The application allows my team to share

search results (e.g., business scenarios,

API endpoints, code structure, etc.) and is

useful for us.

8.875

(0.927)

The application that keeps track of our

search queries as we build our solutions

would be useful for us.

8.125

(1.763)

The application allows me or members

of my team to revisit the results of

previous searches which is useful for us.

8.375

(1.798)

The application that enables my

teammates or peers to provide timely

feedback to each other is useful for us.

8.125

(1.452)

The application can assist us in building

our documentation (e.g., project

specifications, API endpoints,

microservices design) as we develop our

solutions would be useful for us.

7.625

(1.495)

Students valued the collaborative features in the

prototype, particularly the ability to share search results

(8.875), track search queries (

8.125), and revisit

previous search results (8.375). Students also found the

tool beneficial for facilitating timely feedback among

peers when they needed clarification or assistance

(8.125). However, the scores were slightly lower for

the prototype's ability to help them with

documentation, likely due to the primitive

functionality of accepting only text data in the

application.

Overall, the feedback highlights the application's

effectiveness in enhancing collaboration and

documentation during the project development

process.

5 LIMITATIONS AND FUTURE

WORKS

The study focused on a single software development

course, potentially limiting its scope. Similar studies

can be conducted with other software development

courses, with more participants, to verify the

effectiveness of the proposed framework and that

search tools are indeed helpful for students learning

software development (Sadowski, Stolee, & Elbaum,

2015; Brandt, Guo, Lewenstein, Dontcheva, &

Klemmer, 2009; Nam, Macvean, Hellendoorn,

Vasilescu, & Myers, 2024).

Likewise, the study can be extended to include

other types of search tools and the development of

effective prompts for LLM queries.

Feedback received was collected via a voluntary

online survey, reflecting student perceptions. Factors

like search patterns, accuracy, and prompt usage were

not evaluated. These could provide valuable insights

into students’ understanding and interests in particular

technical topics and if the use of search tools spurs

interest in software development.

The prototype that is developed has a very basic

user interface and features. Enhancements could

involve integrating with search engines like Google or

other LLMs, not limiting to ChatGPT. Team project

often use multiple forms of communication such as

Telegram or Zoom that can be integrated.

Another area worth exploring could involve

analyzing search trends by the students. These trends

could indicate when students are searching aimlessly

or are uncertain about their next steps, potentially

signaling a need for assistance (Hamer, Quesada-

López, Martínez, & Jenkins, 2021) .

The issue of individual and team contributions is

consistently a concern for instructors (Macak,

Kruzelova, Chren, & Buhnova, 2021). The

individuals’ contributions to the project can be used to

evaluate the effectiveness of collaborative, peer, and

self-directed learning in the project. The search results

and project team documentation can serve as

references for future similar projects, allowing them to

build upon existing work rather than starting from

scratch.

6 CONCLUSION

Software development courses cover a wide range of

topics and content to prepare students for industrial

workplace demands. The complexity of MSA and

various technical solutions add challenges to teaching

and learning the topics and can make it difficult for

students to apply these concepts.

Software developers and students are using search

tools to assist them with productivity, solve bugs, and

learn new languages and techniques. Rather than

being viewed as solely a tool used to find answers,

CSEDU 2025 - 17th International Conference on Computer Supported Education

240

search tools can be integrated as part of the learning

process.

This study examines the use of search tools in team

projects within a software application development

course involving MSA, focusing on how these tools

can facilitate collaborative, peer, and self-directed

learning. The study first investigates how students are

using search engines and Large Language Models

(LLMs). With a proposed framework, it evaluates the

potential gaps in the effectiveness of these tools.

It was found that both search engines and LLMs

have limitations in supporting teamwork effectively,

primarily due to the lack of features for sharing

information, communicating among team members,

and providing feedback. Since teamwork is an essential

part of software development, alongside the required

technical skills, addressing these challenges is

essential.

Based on student feedback, a prototype was built

and tested with a smaller group of students to verify its

effectiveness and understanding. The prototype's

features—facilitating information sharing, problem-

solving, and team feedback—could further enhance the

effectiveness of search tools to better facilitate and

support collaborative, peer, and self-directed learning

in software development courses.

ACKNOWLEDGMENTS

The authors would like to express gratitude to the

students and instructors involved in this course for

making this study possible.

REFERENCES

Ahmad, A., Waseem, M., Liang, P., Fahmideh, M., Aktar,

M. S., & Mikkonen, T. (2023). Towards Human-Bot

Collaborative Software Architecting with ChatGPT.

27th International Conference on Evaluation and

Assessment in Software Engineering, (pp. 279 - 285).

Oulu, Finland.

Banerjee, P., Srivastava, A., Adjeroh, D., Reddy, Y. R., &

Karimian, N. (2023). Understanding ChatGPT: Impact

Analysis and Path Forward for Teaching Computer

Science and Engineering. TechRxiv.

Bogner, J., Fritzsch, J., Wagner, S., & Zimmermann, A.

(2021). Industry practices and challenges for the

evolvability assurance of microservices: An interview

study and systematic grey literature review. Empirical

Software Engineering, 26(104).

Brandt, J., Guo, P. J., Lewenstein, J., Dontcheva, M., &

Klemmer, S. R. (2009). Two Studies of Opportunistic

Programming: Interleaving Web Foraging, Learning,

and Writing Code. SIGCHI Conference on Human

Factors in Computing Systems, (pp. 1589-1598).

Boston, USA.

Bucchiarone, A., Dragoni, N., Dustdar, S., Larsen, S. T., &

Mazzara, M. (2018). From monolithic to microservices:

An experience report from the banking domain. IEEE

Software, 35(3), 50-55.

Ceh-Varela, E., Canto-Bonilla, C., & Duni, D. (2023).

Application of Project-Based Learning to a Software

Engineering course in a hybrid class environment.

Information and Software Technology, 158, 107189.

Christensen, E. L., & Paasivaara, M. (2022). Learning Soft

Skills through Distributed Software Development.

International Conference on Software and System

Processes and International Conference on Global

Software Engineering, (pp. 93-103). Pittsburgh, PA,

USA.

Ghosh, S., Rath, M., & Shah, C. (2018). Searching as

Learning: Exploring Search Behavior and Learning

Outcomes in Learning-Related Tasks. 2018 Conference

on Human Information Interaction & Retrieval, (pp.

22-31). New Brunswick, NJ.

Grazia, L. D., & Pradel, M. (2023). Code Search: A Survey

of Techniques for Finding Code. ACM Computing

Surveys, 55(11), 1-31.

Griffiths, J. R., & Brophy, P. (2005). Student searching

behavior and the web: use of academic resources and

Google. In library trends.

Guo, P., Saab, N., Post, L. S., & Admiraal, W. (2020). A

review of project-based learning in higher education:

Student outcomes and measures. International Journal

of Educational Research, 102, 101586.

Hahnel, C., Goldhammer, F., Kröhne, U., & Naumann, J.

(2018). The role of reading skills in the evaluation of

online information gathered from search engine

environments. Computers in Human Behavior, 78, 223-

234.

Hamer, S., Quesada-López, C., Martínez, A., & Jenkins, M.

(2021). Using git metrics to measure students’ and

teams’ code contributions in software development

projects. CLEI Electronic Journal , 24(2).

Kirova, V. D., Ku, C. S., Laracy, J. R., & Marlowe, T. J.

(2024). Software Engineering Education Must Adapt

and Evolve for an LLM Environment. 55th ACM

Technical Symposium on Computer Science Education

V. 1, 1, pp. 666 - 672. Portland, USA.

Kuusinen, K., & Albertsen, S. (2019). Industry-Academy

Collaboration in Teaching DevOps and Continuous

Delivery to Software Engineering Students: Towards

Improved Industrial Relevance in Higher Education.

2019 IEEE/ACM 41st International Conference on

Software Engineering: Software Engineering

Education and Training (ICSE-SEET). Montreal, QC,

Canada.

Laal, M., & Laal, M. (2012). Collaborative learning: what

is it? Procedia - Social and Behavioral Sciences, 31,

491-495.

Lane, S. (2016). Promoting collaborative learning among

students. American Journal of Educational Research,

4(8), 602-607.

Use of Search Tools in Software Development: A Study of Microservice-Based Team Projects

241

Lau, Y. M., Koh, C. M., & Jiang, L. (2024). Teaching

Software Development for Real-World Problems using

a Microservice-Based Collaborative Problem-Solving

Approach. IEEE/ACM 46th International Conference

on Software Engineering Education and Training

(ICSE-SEET’24),, (p. 12). Lisbon, Portugal.

Liu, C., Xia, X., Lo, D., Gao, C., Yang, X., & Grundy, J.

(2021). Opportunities and Challenges in Code Search

Tools. ACM Computing Surveys (CSUR), 54(9), 1-40.

Liu, J., Zamir, H., Li, Y., & Hastings, S. K. (2018). Search

systems and their features: What college students use to

find and save information. Library and Information

Science Research, 40, 118-124.

Loes, C. N. (2022). The Effect of Collaborative Learning

on Academic Motivation. Teaching & Learning

Inquiry, 10.

Loyens, S. M., Magda, J., & Rikers, R. M. (2008). Self-

Directed Learning in Problem-Based Learning and its

Relationships with Self-Regulated Learning.

Educational Psychology Review, 20, 411-427.

Lu, Y., & Hsiao, I.-H. (2017). Personalized Information

Seeking Assistant (PiSA): from programming

information seeking to learning. Information Retrieval

Journal, 20, 433-455.

Macak, M., Kruzelova, D., Chren, S., & Buhnova, B.

(2021). Using process mining for Git log analysis of

projects in a software development course. Educational

and Information Technologies, 26, 5939–5969.

MacNeil, S., Tran, A., Hellas, A., Kim, J., Sarsa, S., Denny,

P., . . . Leinonen, J. (2023). Experiences from Using

Code Explanations Generated by Large Language

Models in a Web Software Development E-Book. 54th

ACM Technical Symposium on Computer Science

Education V. 1, (pp. 931 - 937). Toronto, Canada.

Mahdi, M. N., Ahmad, A. R., Ismail, R., Natiq, H., &

Mohammed, M. A. (2020). Solution for Information

Overload Using Faceted Search–A Review. IEEE

Access, 8, 119554-119585.

Nam, D., Macvean, A., Hellendoorn, V., Vasilescu, B., &

Myers, B. (2024). Using an LLM to Help With Code

Understanding. IEEE/ACM 46th International

Conference on Software Engineering, (pp. 1-13).

Lisbon, Portugal.

Podeschi, R. J. (2016). Building IS professionals through a

real-world client project in a database application

development course. Information Systems Education

Journal, 14(6), 34.

Raibulet, C., & Fontana, F. A. (2018). Collaborative and

teamwork software development in an undergraduate

software engineering course. Journal of Systems and

Software, 144, 409-422.

Rangaswamy, A., Giles, C. L., & Seres, S. (2009). A

Strategic Perspective on Search Engines: Thought

Candies for Practitioners and Researchers. Journal of

Interactive Marketing, 23, 40-60.

Rieh, S. Y., Collins-Thompson, K., Hansen, P., & Lee, H.-

J. (2016). Towards searching as a learning process: A

review of current perspectives and future directions.

Journal of Information Science, 42(1), 19-34.

Sadowski, C., Stolee, K. T., & Elbaum, S. (2015). How

Developers Search for Code: A Case Study. 2015 10th

Joint Meeting on Foundations of Software Engineering,

(pp. 191 - 201). Bergamo, Italy.

Shuto, M., Washizaki, H., Kakehi, K., Fukazawa, Y.,

Yamato, S., & Okubo, M. (2016). Learning

effectiveness of team discussions in various software

engineering education courses. 2016 IEEE 29th

International Conference on Software Engineering

Education and Training (CSEET), (pp. 227-231).

Dallas, Texas.

Topping, K. J. (2005). Trends in Peer Learning.

Educational Psychology, 25(6), 631-645.

Tubino, L., Morgan, K., Wood-Bradley, G., & Cain, A.

(2023). Shaping a Tool for Developing Computing

Students’ Professional Identity - Industry Perspectives.

2023 IEEE/ACM 45th International Conference on

Software Engineering: Software Engineering

Education and Training (ICSE-SEET). Melbourne,

Australia.

Vakkari, P. (2016). Searching as learning: A

systematization based on literature. Journal of

Information Science, 42(1), 7-18.

Valový, M., & Buchalcevova, A. (2023). The

Psychological Effects of AI-Assisted Programming on

Students and Professionals. IEEE International

Conference on Software Maintenance and Evolution

(ICSME) (pp. 385-390). IEEE.

Walraven, A., Brand-Gruwel, S., & Boshuizen, H. P.

(2009). How students evaluate information and sources

when searching the World Wide Web for information.

Computers & Education, 52(1), 234-246.

CSEDU 2025 - 17th International Conference on Computer Supported Education

242