Seminar Group 1

Academic Writing (AW)

Andy Weeger

Neu-Ulm University of Applied Sciences

November 23, 2025

Agenda

Structure
Literature integration
Research question clarity
Overall clarity
Language and style (Anja Zenk)
Summary

We’ll use real examples from your cohort to illustrate each point.

Structure

The 5-paragraph formula

Every introduction needs:

Hook - Why this topic matters now
Background - What we know from literature
Tension - What’s missing or unresolved
Resolution - Your approach to address it
Contribution - Expected value of your work

Key principle: Each element gets its own paragraph(s)

Example #1

When and Why Small Language Models Outperform Large Language Models in Resource-Constrained Environments and Domain-Specific Applications (Muhammad Khan)

Well-structured introduction with clear paragraphs, not lengthy, one clear purpose:

Paragraph 1 (Hook): “The rapid advancement of large language models (LLMs) such as GPT-4… However, these achievements come with substantial computational demands… accessibility and practical deployment of LLMs remain restricted…” but: no sources for the claims made

Paragraph 2 (Background): “Recent developments in small language models (SLMs) have emerged… Wang et al. (2024) provide a comprehensive survey showing that SLMs offer significant advantages…” but: superficial synthesis of research, only 2 references

Paragraph 3 (Tension): “Despite these advancements, these platforms often lack sufficient physical realism…” but: claims grounded in literature — no references at all

Paragraph 4 (Resolution): “This thesis aims to address this research gap by investigating when and why SLMs …” but: RQ missing, method is not getting clear (reference to similar studies?)

Paragraph 5 (Contribution): “The expected contributions of this research include…” but: contributions fuzzy/superficial, no reference to the BoK

Example #2

Declarative Decentralized Analytics Framework for FAIR Data Utilization (Sabina Adhikari)

Structure follows the 5-paragraph formula:

Paragraph 1 (Hook): “Over the last several years, data are being generated at an exponential rate… This data holds vast potential… As most information cannot be easily shared… privacy concerns… GDPR… significantly limits cross-institutional data use” but: first half lacks citations; what exactly is the problem (“the lack of methodology …”)

Paragraph 2 (Background): “In order to solve these challenges, researchers have tried to develop frameworks… DAMS… Personal Health Train… However, most of the existing architectures today are of procedural nature…” but: lists examples (methods/frameworks) rather than synthesizing the BoK; ends with a tension

Paragraph 3 (Tension): “Despite these huge technical innovations … they lack standardized abstractions… central orchestration points contradict the core idea of decentralization” but: how do poor reproducibility and central orchestration relate to the problem as well as to DAMs und PHTs?

Paragraph 4 (Resolution): “This paper proposes to overcome the above limitations by developing a conceptual framework… design science and conceptual analysis… based on existing metadata standards like DAMS” but: no explicit research question; link to the problem; method could be clearer (DSR?)

Paragraph 5 (Contribution): “This research shows how declarative design principles can simplify… It tends to create contribution to the academic discussion… bridges the connection between AI-driven analytics and distributed data management” but: contributions need to be more specific/concrete

Example #3

Simulation Scenario Generation based on OpenStreetMap Data (Niharika Patil)

Paragraphs blur together, problem statement scattered throughout:

Paragraph 1 (Hook): “Autonomous vehicles today continue to be an area… the methods used to test them… remain largely manual… Hence, the need for automatic generation…” establishes practical relevance, but already states RQ; does not explain concepts like SUMO, carla

Paragraph 2 (Background): “The recent development shows that OSM data has been utilized… several platforms—such as SUMO, CARLA, and CommonRoad—have successfully incorporated…” no review of the BoK given

Paragraph 3 (Tension): “Despite these advancements, these platforms often lack sufficient physical realism… BeamNG as a high-fidelity, physics-based simulation…” identifies some gap, but claims are not backed by literature, reads rather like the hook

Paragraph 4 (Resolution): “The phenomenon this thesis examines is the feasibility… The idea is to replace or remove the current manual process… creating a pipeline developed in python…” restates problem and background literatur; method is not clear

Paragraph 5 (Contribution): “The contributions made by this research will be both practical and theoretical…” attempts to state contribution, but vague, non-committal, no grounding in BoK

Example #4

Predictive Beamforming (Fatih Hazir)

No clear paragraph structure — duplicate content, overlapping purposes, unclear flow:

Paragraph 1: “Radar systems play a critical role… Air Traffic Control (ATC) radars… Eurocontrol regularly sets standards… Despite advances in radar signal processing, predictive beamforming remains a challenging task… Existing approaches often fail…” Hook elements, but also problem statement mixed in one long paragraph

Paragraph 2 & 3 (Background): “Several previous studies have proposed machine learning–based classification methods… Some approaches utilize handcrafted feature extraction…” no synthesis of literature, only two sources

Paragraph 4: “Despite advances in radar signal processing, predictive beamforming remains a challenging task…” Same text as paragraph 2; copy-paste error; lack of careful editing

Paragraph 4 (Resolution): “The proposed reinforcement learning–based predictive beamforming model will be evaluated using real ATC radar data…” No goal, no RQ, method not clear

Paragraph 5 (Contribution): “The expected contribution of this thesis is a robust AI-driven predictive beamforming framework…” Attempts contribution, but unspecific as not related to the BoK; still describing method/evaluation

Example #5

Automated production planning with the help of artificial intelligence (Maximilian Franz)

No visual paragraph structure — entire text runs together (use template), zero references:

Paragraph 1: “The increasing complexity of production processes… production planning is currently carried out mainly using traditional IT systems such as SAP ERP… Although this method has proven itself in practice, it does have structural weaknesses… To compensate for these shortcomings, experienced production planners rely on heuristics…” No clear paragraph break, combines hook, background, and problem into one dense block; zero references

Paragraph 2 (Tension): “Even advanced planning and scheduling (APS) systems are often still model-driven… There is a clear research gap between the theoretical performance of modern AI methods and their practical applicability…” Again, no references to support claims about clear research gap

Paragraph 3 and 4 (Resolution): “This work addresses precisely this gap… The central research question is: How can an AI-supported prototype for production planning take bottlenecks…” RQ stated, but no connection to the literature — what do we not know to built such a prototype?

Paragraph 5 (Contribution): “Significant scientific and practical benefits are expected if the thesis shows how data-driven decision-making mechanisms can extend classic planning logics…” Vague, conditional contribution — not grounded in BoK

Paragraph 6 (Outline): “Structured in eight chapters, the master’s thesis covers everything from…” Too extensive; partly contradicts prior statements (e.g., DSR vs. research model)

Takeaways

Always use an outline first.

Have one clear message per paragraph.

Connect the paragraph logically.

Literature integration

Example #1

Understanding the Success Dynamics of Digital Servitization: A Conceptual Synthesis through the RBV and DCV Lenses (Rabia Zehra Karataş)

Sources well weaved into the narrative:

Servitization generally describes a company’s shift from a product focused approach to a service-focused approach in its business model (Kowalkowski et al., 2017; Raddats et al., 2019). It offers a structure that can provide mutual benefits for both the manufacturer and the customer. It offers sustainable revenue streams and differentiation for the manufacturer, and it also creates the opportunity for comprehensive solutions and high satisfaction for the customer (Raddats et al., 2019). Information technologies play a critical role in this. Digital technologies such as advanced sensors, big data, and the Internet of Things (IoT) are impacting the relationships manufacturers establish with their customers and the way they do business (Kamp & Parry, 2017).

Example #2

Simulation Scenario Generation based on OpenStreetMap Data (Niharika Patil)

Citation dump — the studies seem to be randomly selected (plus acronyms are not explained):

For example, several platforms—such as SUMO, CARLA, and CommonRoad—have successfully incorporated OSM-based workflows… Zilske et al. (2011) showed how OSM data could be automatically transformed… Similarly, Maierhofer et al. (2021) expanded this method…

Further examples

Declarative Decentralized Analytics Framework for FAIR Data Utilization (Sabina Adhikari)

Factual claims without any evidence, e.g.:

Over the last several years, data are being generated at an exponential rate from the various institutions such as healthcare, research institutes, and business organizations.

Multi-Model, Hardware Aware Neural Network Optimization for Edge Devices (Upanishadh Iyer)

Bibliography given, but no source is cited in the text.

Other observations

To few references (the introduction already provides a synthesis of current research).

Bibliographical data often not complete.

Often pre-dominantly references from questionable sources (often pre-prints from Arxiv high quality outlets missing). Potential reasons:

AI-driven research (no access to papers beyond the paywalls)
Topic does not relate to (current) conversations in academia

Takeaways

Your sources are the product of thorough literary research.

Weave these into your narrative — build an argument supported by studies.

The citation supports the claim but doesn’t interrupt the flow.

Tell a story that happens to be supported by research.

Every factual claim needs support. “Common knowledge” is rarer than you think.

Use a citation manager, check if bibliographical data is complete.

Research question clarity

Good research questions

As discussed in class, a good research question should be:

Explicit - it must be clearly stated
Clear and focused - the question should clearly state what you need to do
Motivated - it is totally clear why you are asking it (aligned with the hook and gap)
Not too easy to answer - easy to locate in the text
Researchable - you must have access to data required to answer the question.

Example #1

Automated production planning with the help of artificial intelligence (Maximilian Franz)

How can an AI-supported prototype for production planning take bottlenecks (such as personnel, capacities, and setup times) into account and thus automatically and soundly replace the previous manual backward scheduling?

Explicit: Clearly stated in one sentence
Clear and focused: Specifies the artifact (AI-supported prototype; better: an AI system), the constraints (bottlenecks), and the outcome (replace manual scheduling)
Motivated: Built up through the hook (complexity of production), why we cannot answer the question (i.e., it is not motivated by instabilities in the BoK)
Not too easy: Requires design, implementation, and evaluation of a prototype
Researchable: Access to Diehl Aviation data and ERP systems mentioned, but the question still seems to be too broad

Example 2

Exploring Leadership Approaches to AI-Driven Cultural Transformation in Organizations (Oluwaseun Oke)

Which leadership approaches prove most effective in driving cultural transformation during AI adoption? How do leaders influence organizational culture in the face of AI-driven change? What critical success factors underpin leadership-driven AI cultural transformation? And how can organizations assess the effectiveness of their leadership approaches in this domain?

Explicit: Clearly stated
Clear and focused: Multiple questions risk scope creep; primary RQ not identified
Motivated: Well-motivated by AI transformation challenges, but lacks focus (too broad; too many relationships)
Not too easy: Requires multi-case study analysis
Researchable: Method (Stake’s framework) and data sources specified; theory missing

Example 3

Definition of Cloud Sovereignty (Mihael Hristov)

How is cloud sovereignty defined?”

Explicit: Very clear
Clear and focused: Extremely focused
Motivated: Motivation present (confusion in terminology)
Not too easy: Too narrow
Researchable: Highly researchable (maybe too researchable)

Recommendation: have a look at Hund et al. (2021) - they propose a definition and a theoretical framework

Example 4

Simulation Scenario Generation based on OpenStreetMap Data (Niharika Patil)

“How practical is it to use BeamNG.drive to automatically create driving simulation scenarios from OpenStreetMap (OSM) data while preserving precise GPS-based positioning data for simulated vehicles?”

Explicit: Clear
Clear and focused: Focused on specific tool and data source
Motivated: Manual scenario creation is time-consuming; instability in BoK not clear
Not too easy: “How practical is it” sounds like feasibility test, not research question
Researchable: Tools are available

Overall clarity

Common issues

We observed following weaknesses regarding clarity:

Paragraph blending: multiple ideas in one paragraph
Missing transitions: paragraphs often don’t connect
Vague language: generic terms without specifics (concepts not introduced)

Paragraph blending

Exploring Leadership Approaches to AI-Driven Cultural Transformation in Organizations (Oluwaseun Oke)

One large paragraph covering at least 4 distinct ideas:

What agentic AI is, where it’s being used, ethical questions that arise, academic debates, responsibility issues, transparency needs, and value alignment.

Rule: One paragraph, one main idea. This has at least 4 distinct ideas.

Potential fix:

Paragraph 1: What is agentic AI and where is it used (Hook)
Paragraph 2: Ethical challenges from autonomy (Hook)
Paragraph 3: Academic debates about accountability (Background)
Paragraph 4: Specific research gaps (Background)

Missing transitions

Predictive Beamforming (Fatih Hazir)

Paragraph ends:

… which makes it challenging to reproduce and extend such systems to new analytical tasks (Šimko et al., 2021).

Next paragraph starts:

Despite these huge technical innovations, the systems are not yet having the full features …

Problem: “Despite” suggests contrast, but the previous paragraph was also about limitations. These should be in the same paragraph OR need a clearer transition.

Vague language

Vague (Maximilian Franz):

The increasing complexity of production processes in industry poses major challenges for companies when it comes to planning and controlling their manufacturing operations.

Generic terms: challenges, complexity

Specific (Alekseeva-Aryuna):

Accurate credit risk assessment is fundamental to modern banking, playing a crucial role in regulatory compliance, financial stability, and informed decision-making. Central to this process is the calculation of expected credit losses (ECL)…

Concrete terms: ECL, capital planning, provisioning

Be specific. Avoid “challenges” or “issues” without saying what they are.

Summary

Key takeaways

Structure

Follow the 5-paragraph formula: Hook — Background — Tension — Resolution — Contribution
One paragraph = one main idea
Use transitions between paragraphs

Literature

Take literature work seriously
Weave citations into your narrative naturally
Don’t just list studies—synthesize them
Every factual claim needs support

Research question

State it explicitly as a question
Set it up—explain why we need to ask it

Clarity

Use specific, concrete language
Avoid vague terms like “challenges” or “issues” without specifying
Use the “reverse outline” test: Can you summarize each paragraph in one sentence?

Final advices

Literature is key: gain a solid understanding of the literature. Read, read, read.
Structure is scaffolding: use it until it becomes natural.
Make your RQ unmissable: bold it, frame it, highlight it.
Claims need to be backed: every factual claim has a citation.
Gaps need specificity: more research needed says nothing.
Literature tells a story: show evolution and debate.
Clarity beats complexity: clear writing reflects clear thinking.

:::

Great introductions aren’t written, they’re rewritten!

Q&A

Hund, A., Wagner, H.-T., Beimborn, D., & Weitzel, T. (2021). Digital innovation: Review and novel perspective. The Journal of Strategic Information Systems, 30(4), 101695.