Material Phase 3: Literature Review, SLR & Reference Management

Complete study guide for Course Unit 3. Master the Systematic Literature Review (SLR) protocol, the 4-Stage PRISMA framework, digital search strategies (ACM, IEEE), reference tools (Zotero/Mendeley), and Harvard citation rules.

1. Purpose of the Literature Review

A Literature Review is a critical synthesis of existing academic publications relevant to a research problem. It is NOT a simple summary or chronological listing of papers ("Author A said X, Author B said Y").

1. Identify the Knowledge Gap

Pinpoints contradictions, unaddressed edge cases, outdated benchmark assumptions, or unsolved technical limitations in published work.

2. Avoid Reinventing the Wheel

Prevents redundant research by evaluating algorithms, datasets, and architectures that have already been tested or proven inefficient.

3. Establish Theoretical Context

Situate your study within established frameworks, paradigms, and mathematical foundations.

4. Benchmark Selection

Identifies state-of-the-art baseline models against which your new CS artifact or algorithm will be empirically evaluated.

2. Systematic Literature Review (SLR) & The 4-Stage PRISMA Diagram

A Systematic Literature Review (SLR) follows a repeatable, audit-ready protocol to collect and analyze all relevant studies answering specific Research Questions.

The 4-Stage PRISMA Statement (Preferred Reporting Items for Systematic Reviews)

Stage 1: Identification

Execute structured search queries across digital databases (ACM Digital Library, IEEE Xplore, Scopus, Google Scholar). Record initial total hits (e.g., N = 1,420 records).

Stage 2: Screening

Remove duplicate papers across databases. Screen titles, abstracts, and keywords against explicit inclusion/exclusion criteria (e.g., retain N = 380 records).

Stage 3: Eligibility

Retrieve full-text PDFs. Assess full articles against methodological rigor, publication quality, and direct relevance (e.g., exclude non-peer-reviewed preprints, retain N = 45 records).

Stage 4: Inclusion

Final corpus of peer-reviewed articles included in qualitative synthesis and quantitative meta-analysis (e.g., Final N = 28 primary studies).

💡 Distinction Rule: IMRaD Format Limitations

Standard IMRaD (Introduction, Methods, Results, and Discussion) structures standalone journal articles well, but final-year CS research theses require a dedicated Chapter 2: Literature Review and Chapter 3: System Architecture & Artifact Design to properly dissect complex technical dependencies.

3. Digital Libraries & Boolean Search Queries

Database	Primary Domain Focus	Special Features
ACM Digital Library	Core Computer Science theory, software engineering, systems.	High peer-review rigor, authoritative ACM Computing Classification System (CCS).
IEEE Xplore	Electrical engineering, computer networking, hardware, signal processing.	Standards documentation (IEEE 802.11, etc.), hardware benchmarks.
Google Scholar	Broad interdisciplinary coverage, gray literature, citations.	Massive index, but includes unvetted preprints; requires manual filtering.

Boolean Query Formulation Example

("microservice architecture" OR "serverless") AND ("latency optimization" OR "cold start") AND NOT ("monolith")

4. Reference Management & Citation Rules

Reference managers like Zotero and Mendeley store metadata, generate automatically formatted bibliographies, and export BibTeX files for LaTeX integration.

⚠️ NUST Examination Standards (<5 Year Recency Rule):
- At least 70% to 80% of cited literature in Chapter 2 must be published within the last 5 years to ensure the work reflects state-of-the-art developments.
- Historical/seminal papers (e.g., Turing 1950, Dijkstra 1968) are allowed only to introduce foundational concepts.
- Direct quotes must be minimized; information should be paraphrased and cited in Harvard format e.g., (Moyo et al., 2024).